Article pubs.acs.org/crt
Estrogenic Activity Data Extraction and in Silico Prediction Show the Endocrine Disruption Potential of Bisphenol A Replacement Compounds Hui Wen Ng, Mao Shu, Heng Luo, Hao Ye, Weigong Ge, Roger Perkins, Weida Tong, and Huixiao Hong* Division of Bioinformatics and Biostatistics, National Center for Toxicological Research, U.S. Food and Drug Administration, 3900 NCTR Road, Jefferson, Arkansas 72079, United States S Supporting Information *
ABSTRACT: Bisphenol A (BPA) replacement compounds are released to the environment and cause widespread human exposure. However, a lack of thorough safety evaluations on the BPA replacement compounds has raised public concerns. We assessed the endocrine disruption potential of BPA replacement compounds in the market to assist their safety evaluations. A literature search was conducted to ascertain the BPA replacement compounds in use. Available experimental estrogenic activity data of these compounds were extracted from the Estrogenic Activity Database (EADB) to assess their estrogenic potential. An in silico model was developed to predict the estrogenic activity of compounds lacking experimental data. Molecular dynamics (MD) simulations were performed to understand the mechanisms by which the estrogenic compounds bind to and activate the estrogen receptor (ER). Forty-five BPA replacement compounds were identified in the literature. Seven were more estrogenic and five less estrogenic than BPA, while six were nonestrogenic in EADB. A two-tier in silico model was developed based on molecular docking to predict the estrogenic activity of the 27 compounds lacking data. Eleven were predicted as ER binders and 16 as nonbinders. MD simulations revealed hydrophobic contacts and hydrogen bonds as the main interactions between ER and the estrogenic compounds.
■
INTRODUCTION Bisphenol A (BPA) is arguably one of the most controversial compounds in the recent decades. First synthesized by the Russian chemist, A. P. Dianin, in 1891, BPA has been used in almost every industry since ca. 1950,1 most notably in the production of polycarbonate plastics and epoxy resins. Examples of products containing or formerly containing BPA include infant’s milk bottles, food and beverage containers, thermal paper, compact discs, adhesives, water piping, automobile parts, dental sealants, and metal can coatings. BPA production exceeds six billion pounds per annum, growing at some 6 to 10%.2 Moreover, environmental release is estimated at about one million pounds per annum (http:// www.epa.gov/oppt/existingchemicals/pubs/actionplans/bpa. html). The ubiquitous presence of BPA leads to widespread exposure to humans through indoor air, dust, packaged food, and water.3−5 Estimates for 2003−2004 were that 92.6% of the United States general population aged six and older had detectable BPA levels in their urine.6 The past couple of decades saw BPA concerns enter the public spotlight, leading to substantial scrutiny by regulators, the scientific community, and media alike, due to its putative This article not subject to U.S. Copyright. Published XXXX by the American Chemical Society
endocrine-interfering nature and potential to cause a myriad adverse health effects. Although the mechanism of BPA endocrine activity remained unclear, it was reported to bind to a variety of receptors such as the estrogen receptors (ER), androgen receptor, and thyroid hormone receptor.7 The ability of BPA to mimic the endogenous estrogens and act as a weak xenoestrogen8 through binding to the notoriously promiscuous ERs9 led to an extensive study of BPA. Animal studies showed that at high doses, BPA produced estrogen-like effects, e.g., increased uterine and prostate weight.6 At lower doses typical of human exposure or below the lowest adverse effect level (LOAEL) of 50 mg/kg body weight/day,4 effects on the mammary glands10 and urogenital system,11 including morphological changes in testes and sperm,12 were observed. The latter effect was proposed to be a product of BPA binding to membrane-bound ERs.13 Although the safety of BPA currently remains contentious, it is generally considered safe at the current levels occurring in foods and is continuing to be approved for uses in food container/packaging by the U.S. Received: June 9, 2015
A
DOI: 10.1021/acs.chemrestox.5b00243 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Chemical Research in Toxicology
Article
Figure 1. Chemical structures of EST, BPA, and BPA replacement compounds collected from the literature. For chemicals with more than one stereochemistry, the possible stereoisomers were also included. Apart from the EST and BPA, the compounds were arranged according their names in alphabetical order. Marked A to D is the conventional nomenclature of the EST steroidal rings.
Not surprisingly, as the replacement “BPA-free” products were reported to also manifest detectable estrogenic activities,20−24 it was deja vu as the controversy emerged anew25,26 with calls for action by the public and regulators. Given the increasing production and use of BPA replacement compounds, increasing the public’s concerns over product safety, the public has demanded appropriate scientific investigations regarding risks to the public and environmental health. Models or inferences made from data cannot be any better than the data itself. Thus, care must be taken to curate a sufficient amount of quality data. Literature data is, of course, unstructured and must be vetted by a qualified person. Qualification of data needs to include ascertaining whether the experimental protocols used for the data were appropriate and comparable. In this study, only data from the recently
Food and Drug Administraion (FDA).14 However, some countries, e.g., European Commission (http://faolex.fao.org/ docs/pdf/eur100741.pdf) and the government of Canada (http://www.gazette.gc.ca/rp-pr/p2/2010/2010-03-31/html/ sor-dors53-eng.html), have enacted laws to ban BPA from certain products such as infant’s milk bottles. To date, the safety of BPA remains controversial due to conflicting findings as well as disputes about experimental designs used for BPA safety evaluations.15 Understandably, a considerable number of compounds were substituted in the market to avoid or circumvent the BPA controversy. These BPA replacement compounds range from BPA-like analogues to other more structurally diverse compounds. BPA analogues such as bisphenol B (BPB),16 bisphenol E (BPE), bisphenol F (BPF), and bisphenol S (BPS)17−20 were commonly used to replace BPA in food/beverage can coatings, for example. B
DOI: 10.1021/acs.chemrestox.5b00243 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Chemical Research in Toxicology
Article
Figure 2. Overall study design. In panel a, the estrogenic activity data for the BPA replacement compounds were extracted through EADB search. In panel b, an in silico model was developed based on molecular docking using the compounds with experimental estrogenic data to predict the estrogenic activity of the compounds without data. Finally, MD simulations were conducted to elucidate the ER-ligand binding interactions.
updated Estrogenic Activity Database (EADB)27 were used, data that is stored in the NCTR’s Endocrine Disruptor Knowledge Base.28 These data were carefully curated, vetted by hand, and stored in the EADB’s Oracle relational database. Both ligand-based models and protein structure-based predictive models have long been commonplace in drug candidate discovery and predictive toxicology. When the target protein structure is not available, the ligand-based methods such as pharmacophore searching,29−32 comparative molecular field analysis (CoMFA),33 and quantitative structure−activity relationship (QSAR)34−38 can be effective for screening or predicting potentially active compounds. If three-dimensional (3D) structure of the target protein is determined, structurebased methods such as molecular docking39−41 can be used to construct predictive models. Fortunately, many crystal structures of ER bound with a variety of ligands have been determined and made available in Protein Data Bank (PDB) (http://www.rcsb.org/pdb/home/home.do). For the BPA replacement compounds for which we were not able to obtain experimental data from EADB, we constructed a two-tier in silico docking model for predicting estrogenic activity. Molecular dynamics (MD) simulations are techniques that use classical molecular mechanics force fields to predict particle motions within a system as a function of time. Therefore, the MD simulations of biologicals system are useful in providing detailed information on the conformational changes and fluctuations of the molecules in the system. In this vein, MD simulations are now routinely used to refine molecular structures, investigate the dynamics and thermodynamics of a given molecular system, and elucidate atomic-level interactions between molecules.41−47 Such an investigation can provide
significant insights into the molecular-level dynamics involved in the mechanisms of action, in this case possibly leading to a disruption of normal homeostatic ER-mediated responses. To gain additional insights from our in silico predictions, the docking results of the experimentally determined and predicted active chemicals in ER were energetically investigated using MD simulations. In brief, from the literature, we found 45 compounds noted as BPA replacements (see Figure 1). We next extracted available experimental data on their estrogenic activities from the EADB, finding that 18 had been tested for estrogenicity, and that 12 of these were found to be active. For the 27 chemicals without experimental data, 11 were predicted to be ER binders using our docking model trained by the 18 chemicals with experimental data and the two reference compounds, 17β-estradiol (EST) and BPA. Our results support the need for thorough safety evaluation of chemicals used to replace BPA.
■
MATERIALS AND METHODS
Overall Study Design. Figure 2 depicts the work flow in two main parts in the study. Part one was a literature search to determine the apparent replacement chemicals of relevance and to gather the associated bioactivity data. BPA and EST were added to serve as the reference active compounds. Figure 1 provides names and shows chemical structures for all of the compounds selected for the study. A search of the EADB then provided curated estrogenic activity data for the compounds and enabled labeling them as active or inactive, or having no data (Figure 2a). The second part was developing a two-tiered in silico modeling approach based on molecular docking for predicting estrogenic activity for the compounds lacking data. All 47 compounds were first docked to the 3D structure of the ERα binding pocket, with the best docking C
DOI: 10.1021/acs.chemrestox.5b00243 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Chemical Research in Toxicology
Article
score used for the assessment of the fitness of chemicals in ER. Binding free energy values were additionally calculated using Prime Molecular Mechanics with Generalized Born Surface Area (MMGBSA) for the estrogenic compounds docked in the ERα structure. The two-tiered predictive model was calibrated based on both the “best” docking scores and binding free energies using just the compounds experimentally determined to be active or inactive. Thereafter, the model was used to predict estrogenic activity of the 27 compounds lacking experiment data. Finally, MD simulations were performed on all compounds either experimentally determined to be active, or predicted to be, in order to elucidate the possible binding interactions between the receptor and the chemicals in a dynamic environment. Data Extraction. EADB was downloaded from the FDA Web site (http://www.fda.gov/ScienceResearch/BioinformaticsTools/ EstrogenicActivityDatabaseEADB/default.htm) and accessed using the Instant J Chem v6.2.1 software (http://www.chemaxon.com). The compounds gathered from the literature were searched via the structure search function in EADB. All of the estrogenic activity data obtained were output to a spreadsheet for subsequent evaluation of endocrine activity potential through ER-mediated responses. For a compound with multiple data from different experiments, the median value of its estrogenic activity data were determined and used in subsequent model development. ER and Small Molecule Preparation. The ERα 3D structure from the PDB, PDB ID 1GWR,48 that we used in a previous docking study40 was prepared using the Protein Preparation Wizard tool within Maestro (http://www.schrodinger.com/Maestro/). Hydrogen atoms were added to the protein structure, and bond orders were assigned. All crystallographic waters except that which formed the “triumvirate hydrogen bond network” with Glu353 and Arg3949 were deleted. The hydrogen bonds were then optimized at pH 7 using the PROPKA program49 within Maestro, after which a restrained minimization was performed using the OPLS_2005 force field50 with the convergence for the heavy atoms set at root-mean-squared-deviations (RMSD) 0.3 Å. The two-dimensional (2D) structures of the BPA replacement compounds were downloaded from PubChem in SDF format. For the compounds not contained in PubChem, 2D structures were drawn using ChemBioDraw Ultra 12.0 (https://www.cambridgesoft.com/ Ensemble_for_Chemistry/ChemBioDraw/) and then output in SDF format. The 3D structures of all compounds, including all combinations of stereoisomers for the nonsteroid structures, were prepared using Ligprep in Maestro. Possible ionization states were generated at pH 7.0 (± 2) using Epik.39,51 Docking. A docking grid was generated using Maestro with the grid box centered at the cognate ligand (EST) and the maximum length of the ligands set to 20 Å. Glide Extra Precision52 was used to perform docking with the following parameters: flexible ligand sampling, 2.5 kcal/mol energy window for ring sampling, 5000 poses per ligand at the initial phase of docking, 400 poses per ligand kept for energy minimization, and maximum of 100 minimization steps. Postdocking minimization was performed with five poses included per ligand. Finally, one pose was written out per ligand in the output file. Binding Free Energy Calculations. The binding free energies of the estrogenic BPA replacement compounds, as well as EST and BPA, in ER were calculated postdocking using Prime MM-GBSA v3.6 within Maestro (http://www.schrodinger.com/Prime/). The residues within 3.0 Å of the docked compound were allowed to be flexible during these calculations, while the rest of the parameters remained as default. MD Simulations. MD simulations were performed using Desmond53 on the docked complexes of compounds either measured or computationally predicted to be active. The simulation systems were built using the System Builder tool within Maestro with simple point-charge (SPC) used as the water model. An orthorhombic box shape with 10 Å solvent buffer distance to each dimension of the box was applied. The total charges of the systems were neutralized by adding 30 Na+ and 26 Cl−, achieving a salt concentration of 0.15 M. The force field used was OPLS 2005.54 Six system-relaxation steps
were then done in stages as follows using the default settings in Desmond: (1−2) steps one and two were minimization steps of the protein−ligand complex with and without restraint on the solute, respectively; (3) step three was a 12 ps (ps) constant number, volume, and temperature (NVT) ensemble simulation carried out at 10K with a Berendsen barostat as well as a fast temperature relaxation constant. Velocity resampling was set at every 1 ps, and restraints were applied on all heavy atoms of the protein−ligand complex; (4) step four was a 12 ps constant number, pressure, and temperature (NPT) ensemble simulation carried out at 10K and 1 atmospheric pressure (atm) using a Berendsen thermostat and barostat with a fast temperature relaxation constant and a slow pressure relaxation constant; velocity resampling was set at every 1 ps, and restraints were applied on all heavy atoms of the protein−ligand complex; (5) step five was a 12 ps NPT simulation carried out at 300 K and 1 atm using the same thermostat, barostat, temperature relaxation constants, pressure relaxation constants, velocity resampling steps, and restraints; and (6) the last step six was a 24 ps NPT simulation, again carried out at 300 K and 1 atm, and using a fast temperature and normal pressure relaxation constants. Finally, the production simulations were conducted under 300 K and 1 atm under NPT conditions for 20 ns (ns) with an energy recording interval of 5 ps and a trajectory recording interval of 20 ps. After the simulations, calculations of RMSDs for the whole duration of the production simulation, and the root-mean-square fluctuations (RMSFs), as well as an analysis of the binding interactions that took place in the last nanosecond were carried out using the Simulation Interactions Diagram tool in Maestro. Statistical Analysis. The linear regression model was constructed using the LinearModel.fit function in the Statistics toolbox in MATLAB (http://www.mathworks.com/products/matlab/). The pvalue of the linear model was calculated using F-statistic to assess the statistical significance of the model from a constant model. The linear correlation coefficient was calculated from Pearson’s correlation using the corr function in the Statistics toolbox in MATLAB.
■
RESULTS Experimental Data Indicated Potential Endocrine Activity. To evaluate the endocrine activity potential through ER-mediated responses of the 45 BPA replacement compounds, the experimental estrogenic activity data were extracted from EADB using structure search. No data existed for 27 of the compounds of interest. The experimental estrogenic data for the remaining 18 compounds indicated that 12 were estrogenic in ER binding and reporter gene assays using various species such as human, rat, mouse, etc. (see Table S1 for the detailed estrogenic data). The median relative binding affinity values (logRBA) of the 12 estrogenic BPA replacement compounds as well as the reference compounds EST (EST logRBA set to 2.0) and BPA were given in Figure 3. Among the 12 estrogenic BPA replacement compounds, five (TMBPA, HPP, PHBB, BPF, and BPS) showed weaker estrogenic activity than BPA. However, seven compounds (BPC, BPAF, BPZ, NDCP, BPP, BPB, and BPAP) exhibited moderate estrogenic activity exceeding that of BPA. Predictive Model. To construct a predictive model based on molecular docking, the 18 BPA replacement compounds having experimental estrogenic data together with the reference compounds EST and BPA were docked in the ligand binding domain of ER. The best extra precision docking score was obtained for each (with the exception of BPP which was not successfully docked; see results in Table 1). Figure 4a plots the resultant docking scores for both experimentally determined positives (the left notched box, including the reference compounds EST and BPA) and negatives (the right notched box), indicating a straightforward segregation of active and inactive compounds based on the scores. The resultant binary D
DOI: 10.1021/acs.chemrestox.5b00243 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Chemical Research in Toxicology
Article
energy optimized to calculate the binding free energies using the Prime MM-GBSA method in Maestro. The binding free energies of these 11 compounds were converted to logRBA values using the linear model as listed in Table 1. Binding Interactions. The convergences of the MD simulations of complexes of ER bound by the 24 active compounds (13 measured, BPP not included due to unsuccessful docking, and 11 predicted) were assessed at the end of the 20 ns. The RMSD values are plotted in Figure S3. Comparing the plots, while having different RMSD patterns, the systems exhibit stable progression to convergence after initial larger fluctuations, with the backbone RMSD values at around 2 Å (Å) and side chain RMSD values around 3 Å. The key interactions between ER and its ligands during the final 1 ns simulations were analyzed. The RMSF values were calculated and plotted in Figure S4. Examining the RMSF plots, it was shown that the residues of ER generally fluctuated mildly, i.e., at 1−2 Å with the exception of the Ile326-Glu339 and Tyr459-Lys472 regions, which showed more vigorous residue movements due to the loops and breaks (missing residues between Tyr331and Arg335 as well as Phe461 and Thr465) of the ER crystal structure in these regions. The rather nonspecific hydrophobic contacts between the nonpolar residues lining the ER ligand binding pocket and the compounds studied were found to be a major contributor to the overall protein−ligand interactions (see Figure 5 and Figure S5). Leu346, Ala350, Leu387, Met388, Leu391, Phe404, His524, and Leu525 were the most important for nonpolar contacts. Phe404 played a significant role in forming π-π interactions with the bound molecules. Ala350 and Phe404, with the latter forming distinctive face-to-edge π-π interactions with one of the aromatic rings of the binding chemicals, were noted to sandwich many of these chemicals in a region which corresponds to the A-ring of EST (see Figure 5 and Figure 1 for EST ring nomenclature) when bound to ERα. Another major interaction that secures these chemicals in positions is hydrogen bonding. Chemicals such as BPA, BPE, BPF, TMBPA, and TGSA were found to adopt positions that favor hydrogen bonding of their respective hydroxyl groups with His524 and Glu353 throughout the simulations, while BPSMAE, MBHA, TBBPA, and TCBPA, although adopting similar poses, only formed a hydrogen bond with His524. However, BPC, BPAF, BPAP, and NDCP assumed poses that more favored hydrogen bonding with Thr347 and Glu353. Similarly, while BPS, BPB, and BPZ adopted similar poses, the former only formed a hydrogen bond with Glu353, while BPB and BPZ did with Thr347. The “signature” hydrogen bonding interactions as displayed by EST in the ER crystal structures deposited in the PDB, i.e., (i) the “triumvirate hydrogen bond network” between a water molecule, Glu353 and ARG394, and (ii) the D-ring hydroxyl group with His524, were only partially preserved during the simulations. In this case, the involvement of Arg394 was lost, probably due to the departure of the water molecule from the crucial position during the course of the simulations. BPS-MPE, D-8, DD-70, PHBB, and 2,4-BPS, although found to form hydrogen bonds with the receptor, adopt diverse poses. D-90 and HPP did not form any hydrogen bonds.
Figure 3. Experimental estrogenic activity data. The logRBA values of the BPA replacement and reference compounds are denoted by the empty and black bars, respectively. Using the experimental logRBA value of BPA as the yardstick, higher logRBA values are represented by up-bars, while lower values are represented by down-bars.
predictor classifies a compound to be estrogenic if the score is less than −6.614, and otherwise, it is predicted to be inactive, with just one false positive (2,2-BPF is nonestrogenic but was classified as estrogenic). A linear, least-squares fit of experimental logRBA versus docking score yielded logRBA = −10.480−1.039·docking score. The fit had a p-value of 0.007, indicative of statistical significance. When the linear model was used to convert docking scores to logRBA (Table 1), the logRBA values correlated with the actual experimental data with a moderate correlation coefficient of 0.705 (Figure S1). The binding free energy for the ligands from Prime MMGBSA calculations are listed in Table 1. Similarly, the binding free energy data could be used as a qualitative model to separate the active compounds from the inactive compounds as demonstrated in Figure S2, with a slightly narrower margin compared to the model based on the best docking score. The linear fit of logRBA = −7.719−0.0860·binding free energy can be used to convert binding free energy to logRBA. The logRBA values calculated based on binding free energies correlate with the experimental binding affinity data better than the logRBA predicted from docking scores, with a correlation coefficient of 0.833 (see Figure 4b). On the basis of the performance of the models, a two-tier in silico prediction model was proposed for predicting the estrogenic activity of the chemicals without experimental data. Tier-one of the estrogenic activity predictor for compounds lacking experimental data uses the best ER docking score for assigning a compound as active or inactive. For compounds predicted as active in tier-one, a second confirmatory tier-two prediction based on ER binding free energy is conducted to determine its binding affinity. Prediction of Estrogenic Activity. For the 27 BPA replacement compounds without experimental estrogenic data, their potential estrogenic activities were predicted using the two-tier in silico model. Each of the 27 chemicals was docked to ER and their best docking score determined (Table 1). Four compounds failed in docking and thus were predicted as ER nonbinders. Eleven of the 23 successfully docked compounds had the best docking scores, better than the model activity cutoff of (−6.614), and thus predicted as estrogenic. The remaining 12 compounds had docking scores larger than the cutoff, and they were predicted to be nonestrogenic. The ERligand complexes obtained from docking were then further
■
DISCUSSIONS We identified 45 potential BPA replacement compounds from the literature and extracted data for them from the FDA’s EADB. Eighteen of these compounds had experimental data E
DOI: 10.1021/acs.chemrestox.5b00243 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Chemical Research in Toxicology
Article
Table 1. Median Estrogenic Activity Data, Docking Scores, Calculated Binding Free Energies, and Predicted Estrogenic Activity Data compda EST BPA BPAF BPAP BPB BPC BPF BPP BPS BPZ HPP NDCPb PHBB TMBPA DMT methacrylic acid styrene TPA 1,4-dihydroxybenzene 2,2-BPF acrylic acid acrylonitrile BisOPP-A BPA bis(diphenylphosphate) BPE BPS-MAE BPS-MPE BTUM cis-CHDM trans-CHDM D-8 D-90 DD-70 D-lactic acid L-lactic acid MBHA methyl methacrylate norbornene Pergafast201 TBBPA TCBPA TGSA cis-TMCD trans-TMCD UU 1,4-bis(4-chlorophenyl) sulfone 2,4-BPS
median data (logRBA)c
XP lowest docking score
calculated binding free energy (kcal/mol)
predicted logRBA from docking score
predicted logRBA from binding free energy
2.000 −1.680 −0.050 −1.100 −0.950 0.360 −3.050 −0.930 −3.050 −0.670 −2.280 −0.746 −2.532 −1.710 inactive inactive inactive inactive inactive inactive PI PI PI PI
−10.304 −8.263 −9.646 −9.453 −8.611 −8.750 −7.047
−106.484 −67.359 −69.765 −77.949 −73.992 −84.794 −55.138
0.226 −1.895 −0.458 −0.658 −1.533 −1.389 −3.158
1.439 −1.926 −1.719 −1.015 −1.356 −0.427 −2.977
−8.175 −10.206 −8.552 −9.125 −8.017 −10.129 −5.265 −3.392 −4.752 −6.180 −4.446 −7.692 −2.323 −1.126
−61.863 −85.326 −69.438 −84.313 −61.953 −89.150 −52.158 −32.049 −38.940 −40.637 −32.674 −58.938
−1.986 0.1240 −1.594 −0.999 −2.150 0.044
−2.399 −0.381 −1.747 −0.468 −2.391 −0.052
−8.407 −7.629 −9.586
−61.688 −80.799 −83.544
−2.414 −0.770 −0.534
−79.388 −80.827 −105.371
−0.892 −0.768 1.343
−65.103
−2.120
−111.296 −109.504 −95.765
1.852 1.698 0.517
PA PA PA PI PI PI PA PA PA PI PI PA PI PI PI PA PA PA PI PI PI PI PA
−4.966 −5.166 −7.505 −8.722 −8.968 −3.311 −3.427 −9.125 −2.702 −3.569 −0.511 −8.323 −9.498 −9.664 −4.131 −3.911 −6.544 −7.818
−60.749
a
−2.495 b
The full name of the compounds can be found in Abbreviation List and Table S1. Note that two of the compounds identified were referred to as “BPC” in the literature. We refer to this compound as NDCP (Nonox DCP), i.e., one of the synonyms listed in PubChem, to avoid confusion. cThe experimentally determined or predicted classes of estrogenic activity for each compound: the values provided are the median logRBA of the experimentally determined active compounds; “inactive” indicates experimentally determined inactive compounds; “PA” indicates compounds which are predicted active, while “PI” indicates compounds which are predicted inactive.
showing 12 to be estrogenic and six to be inactive at ER. There was no experimental data for the remaining 27 compounds. Among the 12 active compounds, BPC, BPAF, BPZ, NDCP, BPP, BPB, and BPAP had been measured to be more
estrogenic than BPA. For the 27 compounds having no estrogenic activity data, 11 compounds were predicted to be estrogenic using our two-tiered in silico model. Our results demonstrated 23 BPA replacement compounds have endocrine F
DOI: 10.1021/acs.chemrestox.5b00243 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Chemical Research in Toxicology
Article
∼0.36 could be considered a strong ER binder, more than 100fold stronger than BPA itself. BPAF, BPZ, NDCP, BPP, BPB, and BPAP were moderate ER binders and also showed stronger estrogenic activity than BPA. The other five BPA replacement compounds were weaker ER binders than BPA. The data used for the comparison with BPA were the median of different data. It is worth noting that BPS, one of the weaker ER binders commonly used as a BPA replacement compound, was detected in 81% of 315 urine samples collected in the USA and seven other Asian countries55 and has been measured to activate the estrogen and androgen receptors;56,57 accordingly, questions about its safety have been previously raised.58−60 This reiterates that a compound with weaker (or no) ER binding does not necessarily mean it is a “safe” compound. A paper on the safety of five BPA alternatives was recently published21 during the preparation of this manuscript. For the four out of their five BPA alternatives cited by Rosenmai, we had extracted similar experimental estrogenic activity data from EADB in the course of our work. The remaining one, BPE, is not contained in the EADB, and our in silico model predicted it to be estrogenic. Furthermore, our quantitative prediction on BPE’s estrogenic activity, i.e., logRBA = −2.318, about 4.35-fold weaker than BPA (logRBA = −1.68), was consistent with the recently published experimental data (using maximum effect data and 50% effect concentration data, BPE was 1.35-fold and 10.25-fold weaker than BPA, respectively). In short, the recent safety study of the five BPA alternatives confirmed our results. It is worth pointing out that this study only investigated the estrogenic activity of the 45 BPA replacement compounds as opposed to their safety, hence potentially indicating the need for further safety evaluation on endocrine disruption through ER-mediated responses, the major safety concern on BPA. Indeed, the endocrine disruption potential of these compounds would be better evaluated if their capability to bind serum proteins such as α-fetoprotein61 and sex hormone binding globulin62 were available. Other aspects such as the heat resistance of the finished products and the ease of leaching of these chemicals from them are also important points for consideration in ascertaining the individual product safety. The compounds which have not been found to possess estrogenic activity consist of chemicals more structurally divergent from BPA: methacrylic acid, styrene, 1,4-dihydroxybenzene, DMT, and TPA (2,2-BPF being the only exception). In general, these structures are smaller compared to the BPA analogues and may or may not contain a ring structure. From a structural point of view, these chemicals are too small in size to allow good anchoring to the ER binding pocket and do not fulfill commensurate criteria of a good ER binder, i.e., one that consists of two hydroxyl groups suitably spaced by a hydrophobic core.63,64 2,2-BPF is the only compound that is successfully docked in ER, predicted as positive but experimentally determined as nonestrogenic. In other words, it could be a false positive of the docking model. However, only one experimental result (see Table S1) was found for this compound. Therefore, further experimental validation would help to ascertain its estrogenic activity. No information could be found for 27 potential BPA replacement compounds surveyed in this study. Using the in silico model developed, 11 were predicted to be estrogenic, while the remaining 16 compounds were predicted to be nonestrogenic. Structurally, BPE, 2,4-BPS, BPS-MAE, BPSMPE, TGSA, D-8, D-90, MBHA, TBBPA, and TCBPA showed only minor differences from other BPA analogues (mainly
Figure 4. Performance of the two-tier in silico predictive model. The box plot of the lowest XP docking gscores of the positive and negative compounds is shown in a. The one-way ANOVA returned a p-value of 6.29 × 10−6. The predicted estrogenic activity data from the binding free energies by the tier-2 quantitative model for the estrogenic BPA replacement compounds were plotted against their actual estrogenic activity data as circles in b, indicating a good correlation between the predicted and the actual estrogenic activity data.
disruption potential through ER-mediated responses. Replacement of BPA, especially by these compounds should be performed with extra caution as the concern on BPA’s low-dose toxicological effects such as endocrine disruption might remain. That said, the compounds which have not been shown to demonstrate estrogenic activity also warrant further study for activity at other endocrine receptors also found to be involved in the interference of endocrine signaling, e.g., the androgen receptor and thyroid hormone receptors. Solid safety evaluations should be conducted before the products containing these compounds are introduced to the market in order to protect public health. The experimentally determined estrogenic activity data of the 12 potential BPA replacement compounds were ranked as BPC > BPAF > BPZ > NDCP > BPP > BPB > BPAP BPA > TMBPA > HPP > PHBB > BPF > BPS. BPC with logRBA of G
DOI: 10.1021/acs.chemrestox.5b00243 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Chemical Research in Toxicology
Article
Figure 5. Binding interactions between ER and the estrogenic BPA replacement compounds (experimentally determined and in silico model predicted), EST and BPA. The binding poses of the 24 estrogenic compounds in the ERα ligand binding pocket were shown to illustrate the key residues involved in binding. Hydrophobic contacts (protein nonpolar surface, gray; π-π interactions, blue dotted line) appear to play a major role in conferring binding affinity to these molecules. Hydrogen bonding interactions (protein polar surface, red for electronegative; blue for electropositive; hydrogen bond, yellow dotted line) are crucial to anchor these molecules in the binding pocket. The corresponding interaction fractions chart can be found in Figure S5.
having flexible extensions to the common bisphenol skeleton), thus supporting their calculated estrogenicity. DD-70 is highly flexible to fit in and bind with the ER binding pocket. However, 1,4-bis(chlorophenyl)sulfone, BisOPP-A, Pergafast201, BTUM, UU, and BPA bis(diphenylphosphate), which were also structurally related to BPA, were predicted to be nonestrogenic.
It is worth noting that 1,4-bis(chlorophenyl)sulfone, albeit resembling BPS, lacks any hydroxyl moieties to anchor the molecule to the polar residues flanking the ER ligand binding pocket, a major structural reason that may account for its nonestrogenic nature. BisOPP-A, Pergafast201, BTUM, UU, and BPA bis(diphenylphosphate), despite being structurally H
DOI: 10.1021/acs.chemrestox.5b00243 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Chemical Research in Toxicology
Article
The case with BPA and its replacement compounds is typical of a scenario where unfavorable scientific findings of a commonly, often extensively, used chemical had led to a frenzy of replacement efforts by the industry with other assumed safer compounds.71 With the BPA replacement chemicals, a significant number of compounds included in this study had already been demonstrated to potentially interfere with the estrogenic pathways in the body; in some cases, the mechanism(s) of actions and ability of these chemicals to cause effects to the body and environment are still largely, poorly understood. It cannot be ruled out that these compounds may in time become the next big public health concern, not unlike the situation that arose in the past with polybrominated diphenyl ethers (PBDEs) and perfluorooctanesulfonate (PFOS).71 In this sense, extensive studies, safety evaluations, and prudent, preventive steps are of paramount importance in preventing history from repeating itself. Given the history of BPA being a scientific and public concern, a cautionary thorough evaluation of replacement compounds for estrogenic activity seems warranted. It is worth noting that a lack of predicted estrogenic response does not necessarily mean that it is endocrine-disruptor free as endocrine interference may occur at mechanisms other than the estrogenic signaling pathway.
related to BPA, were predicted as nonestrogenic due, unsurprisingly, to their sheer size, i.e., simply too large for the ER binding pocket; Pergafast201, albeit successfully docked, was not favorably scored probably due to its rather rigid structure, which led to difficulty in adopting poses complementary to the ER pocket. For the exact opposite reason (i.e., too small in size), the other 10 compounds were also predicted to be nonestrogenic: CHDM (both cis and trans forms) and TMCD (both cis and trans forms), acrylic acid, acrylonitrile, lactic acids (both D and L forms), methyl methacrylate, and norbornene. Note that despite being predicted to be nonestrogenic, acrylonitrile is a toxic compound reported to cause cancers.65 The key protein−ligand interactions between ER and the 22 estrogenic BPA replacement compounds (BPP was not included due to unsuccessful docking because of its large size) as well as the two reference compounds were elucidated by MD simulations and compared to results in the literature,66 where the key interacting residues of ER for BPA, BPAF, and BPC were identified using the 3D structures from crystals of the complexes of ER bound with these ligands. Our findings from the MD simulations showed that BPA, BPE, BPF, TMBPA, and TGSA formed and mostly maintained hydrogen bonds with Glu353 and His524. These interactions were found to be consistent with those that stabilized the “BPA conformation” as described in the article. However, the hydrogen bonding patterns formed by BPC, BPAF, BPAP, and NDCP with Glu353 and Thr347 were found to be consistent with the interactions described to stabilize the “BPC conformation.” An interaction with Leu525, which was suggested to be crucial in holding helix 3 and helix 11 in place and maintaining the AF2 surface to facilitate coactivator binding and activation,66 was not observed with EST but was observed with other chemicals studied. Finally, the mobility of helix 12, i.e., the “molecular switch” for the ER (Tyr537His547), during the simulations was not evaluated among these complexes, as a meaningful observation could not be made due to the fact that a coactivator protein that helped to stabilize helix 12 was not included in the simulation. Plastic products generally comprise a wide range of building blocks. Ideally, any whole product testing that yields estrogenic activity could be further assessed through docking of the individual monomer components to help with rationalization of the results and provide a clearer picture as to which of these are/is the true “culprit(s).” However, in vitro experimental data and in silico predictions may not be consistent with in vivo outcomes. In vitro studies mostly exclude metabolic effects that can either yield a compound less estrogenic or not active, or yield a more active compound.67−69 For instance, while the testing of poly(ether sulfone) yielded estrogenic activity,22 the in vitro data extracted in this study indicated that one compound in the product, 1,4-dihydroxybenzene, did not show estrogenic activity in the rat ER binding assay and the yeast two-hybrid assay;70 the other compound, 1,4-bis(chlorophenyl)sulfone, was also predicted as nonestrogenic using our in silico model. One speculation could be that the metabolites of 1,4-dihydroxybenzene and 1,4-bis(chlorophenyl)sulfone may be estrogenic. Another possibility is the very weak, equivocal nature of estrogenic activity for 1,4bis(chlorophenyl)sulfone because its ER binding score (−6.544) is very close to the model cutoff for a binder (−6.612).
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.chemrestox.5b00243. Experimental estrogenic data extracted for the 45 BPA replacement compounds and the reference compounds, EST and BPA; performance of the quantitative model based on docking scores; box plot of the postdock MMGBSA values for the experimentally determined active and inactive compounds; RMSDs of the ER backbone and side chains during the 20 ns MD simulations; RMSFs of the ER backbone and side chains during the final nanosecond of the MD simulations; key interactions between the ER and the bound ligands during the final nanosecond of the MD simulations (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Funding
This research was supported in part by an appointment to the Research Participation Program at the National Center for Toxicological Research (H.W.N., M.S., H.L., and H.Y.) administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the U.S. Department of Energy and the U.S. Food and Drug Administration. Notes
The content is solely the responsibility of the authors and does not necessarily represent the official views of the Food and Drug Administration. The authors declare no competing financial interest. I
DOI: 10.1021/acs.chemrestox.5b00243 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Chemical Research in Toxicology
■
Article
(7) Wetherill, Y. B., Akingbemi, B. T., Kanno, J., McLachlan, J. A., Nadal, A., Sonnenschein, C., Watson, C. S., Zoeller, R. T., and Belcher, S. M. (2007) In vitro molecular mechanisms of bisphenol A action. Reprod. Toxicol. 24, 178−198. (8) Kuiper, G. G., Lemmen, J. G., Carlsson, B., Corton, J. C., Safe, S. H., van der Saag, P. T., van der Burg, B., and Gustafsson, J. A. (1998) Interaction of estrogenic chemicals and phytoestrogens with estrogen receptor beta. Endocrinology 139, 4252−4263. (9) Ng, H. W., Perkins, R., Tong, W., and Hong, H. (2014) Versatility or Promiscuity: The Estrogen Receptors, Control of Ligand Selectivity and an Update on Subtype Selective Ligands. Int. J. Environ. Res. Public Health 11, 8709−8742. (10) Durando, M., Kass, L., Piva, J., Sonnenschein, C., Soto, A. M., Luque, E. H., and Munoz-de-Toro, M. (2007) Prenatal bisphenol A exposure induces preneoplastic lesions in the mammary gland in Wistar rats. Environ. Health Perspect. 115, 80−86. (11) Timms, B. G., Howdeshell, K. L., Barton, L., Bradley, S., Richter, C. A., and vom Saal, F. S. (2005) Estrogenic chemicals in plastic and oral contraceptives disrupt development of the fetal mouse prostate and urethra. Proc. Natl. Acad. Sci. U. S. A. 102, 7014−7019. (12) Goodman, J. E., McConnell, E. E., Sipes, I. G., Witorsch, R. J., Slayton, T. M., Yu, C. J., Lewis, A. S., and Rhomberg, L. R. (2006) An Updated Weight of the Evidence Evaluation of Reproductive and Developmental Effects of Low Doses of Bisphenol A. Crit. Rev. Toxicol. 36, 387−457. (13) Alonso-Magdalena, P., Laribi, O., Ropero, A. B., Fuentes, E., Ripoll, C., Soria, B., and Nadal, A. (2005) Low doses of bisphenol A and diethylstilbestrol impair Ca2+ signals in pancreatic alpha-cells through a nonclassical membrane estrogen receptor within intact islets of Langerhans. Environ. Health Perspect. 113, 969−977. (14) U.S. Food and Drug Administration (2015) Bisphenol A (BPA): Use in Food Contact Application. (15) Borrell, B. (2010) Toxicology: The big test for bisphenol A. Nature 464, 1122−1124. (16) Grumetto, L., Montesano, D., Seccia, S., Albrizio, S., and Barbato, F. (2008) Determination of bisphenol A and bisphenol B residues in canned peeled tomatoes by reversed-phase liquid chromatography. J. Agric. Food Chem. 56, 10633−10637. (17) Satoh, K., Ohyama, K., Aoki, N., Iida, M., and Nagai, F. (2004) Study on anti-androgenic effects of bisphenol a diglycidyl ether (BADGE), bisphenol F diglycidyl ether (BFDGE) and their derivatives using cells stably transfected with human androgen receptor, AR-EcoScreen. Food Chem. Toxicol. 42, 983−993. (18) Sueiro, R. A., Suarez, S., Araujo, M., and Garrido, M. J. (2003) Mutagenic and genotoxic evaluation of bisphenol F diglycidyl ether (BFDGE) in prokaryotic and eukaryotic systems. Mutat. Res., Genet. Toxicol. Environ. Mutagen. 536, 39−48. (19) Gallart-Ayala, H., Moyano, E., and Galceran, M. (2011) Analysis of bisphenols in soft drinks by on-line solid phase extraction fast liquid chromatography−tandem mass spectrometry. Anal. Chim. Acta 683, 227−233. (20) Chen, M. Y., Ike, M., and Fujita, M. (2002) Acute toxicity, mutagenicity, and estrogenicity of bisphenol-A and other bisphenols. Environ. Toxicol. 17, 80−86. (21) Rosenmai, A. K., Dybdahl, M., Pedersen, M., van VugtLussenburg, B. M. A., Wedebye, E. B., Taxvig, C., and Vinggaard, A. M. (2014) Are structural analogues to bisphenol A safe alternatives? Toxicol. Sci. 139, 35−47. (22) Yang, C. Z., Yaniger, S. I., Jordan, V. C., Klein, D. J., and Bittner, G. D. (2011) Most plastic products release estrogenic chemicals: a potential health problem that can be solved. Environ. Health Perspect. 119, 989−996. (23) Bittner, G. D., Yang, C. Z., and Stoner, M. A. (2014) Estrogenic chemicals often leach from BPA-free plastic products that are replacements for BPA-containing polycarbonate products. Environ. Health 13, 41. (24) Glausiusz, J. (2014) Toxicology: The plastics puzzle. Nature 508, 306−308.
ABBREVIATIONS FDA, U.S. Food and Drug Administration; BPA, bisphenol A; EADB, Estrogenic Activity Database; MD, molecular dynamics; ER, estrogen receptor; LOAEL, lowest adverse effect level; CoMFA, comparative molecular field analysis; QSAR, quantitative structure−activity relationship; 3D, three-dimensional; PDB, Protein Data Bank; EST, 17β-estradiol; MMGBSA, Molecular Mechanics with Generalized Born Surface Area; 2D, two-dimensional; SPC, simple point-charge; NVT, constant number, volume, and temperature; NPT, constant number, pressure, and temperature; atm, atmospheric pressure; RMSDs, root-mean-square deviations; RMSFs, root-mean-square fluctuations; logRBA, log relative binding affinity; BPAF, bisphenol AF; BPAP, bisphenol AP; BPB, bisphenol B; BPC, bisphenol C; BPF, bisphenol F; BPP, bisphenol P; BPS, bisphenol S; BPZ, bisphenol Z; HPP, 4-cumylphenol; NDCP, nonox DCP; PHBB, benzyl 4-hydroxybenzoate; TMBPA, tetramethylbisphenol A; DMT, dimethyl terephthalate; TPA, terephthalic acid; 2,2-BPF, 2,2-bisphenol F; BisOPP-A, 4,4″-isopropylidenebis2-phenylphenol; BPE, bisphenol E; BPS-MAE, bis(4hydroxyphenyl)sulfonemonoallylether; BPS-MPE, 4-hydroxy4″-benzyloxydiphenylsulfone; BTUM, 4,4′-bis-N-carbamoyl-4methylbenzenesulfonamidediphenylmethane; cis-CHDM, cis1,4-cyclohexanedimethanol; trans-CHDM, trans-1,4-cyclohexanedimethanol; D-8, 4-hydroxyphenyl 4-isoprooxyphenylsulfone); D-90, 4-[4′-[(1′-methylethyloxy)phenyl]sulfonyl]phenol; DD-70, 1,7-bis-4-hydroxyphenylthio)-3,5-dioxaheptane; MBHA, methyl(bis-4-hydroxyphenyl)acetate; TBBPA, tetrabromobisphenol A; TCBPA, tetrachlorobisphenol A; TGSA, bis-3-allyl-4-hydroxyphenyl sulfone; cis-TMCD, cis2,2,4,4-tetramethyl-1,3-cyclobutanediol; trans-TMCD, trans2,2,4,4-tetramethyl-1,3-cyclobutanediol; UU, urea urethane compound; 2,4-BPS, 2,4-bisphenol S; PBDEs, polybrominateddiphenylethers; PFOS, perfluorooctanesulfonate
■
REFERENCES
(1) Vogel, S. A. (2009) The politics of plastics: the making and unmaking of bisphenol a ″safety″. Am. J. Public Health 99 (Suppl 3), S559−S566. (2) Burridge, E. (2003) Bisphenol A: product profile. Eur. Chem. News, 14−20. (3) Kubwabo, C., Kosarac, I., Stewart, B., Gauthier, B. R., Lalonde, K., and Lalonde, P. J. (2009) Migration of bisphenol A from plastic baby bottles, baby bottle liners and reusable polycarbonate drinking bottles. Food Addit. Contam., Part A 26, 928−937. (4) vom Saal, F. S., Akingbemi, B. T., Belcher, S. M., Birnbaum, L. S., Crain, D. A., Eriksen, M., Farabollini, F., Guillette, L. J., Jr, Hauser, R., Heindel, J. J., Ho, S.-M., Hunt, P. A., Iguchi, T., Jobling, S., Kanno, J., Keri, R. A., Knudsen, K. E., Laufer, H., LeBlanc, G. A., Marcus, M., McLachlan, J. A., Myers, J. P., Nadal, A., Newbold, R. R., Olea, N., Prins, G. S., Richter, C. A., Rubin, B. S., Sonnenschein, C., Soto, A. M., Talsness, C. E., Vandenbergh, J. G., Vandenberg, L. N., Walser-Kuntz, D. R., Watson, C. S., Welshons, W. V., Wetherill, Y., and Zoeller, R. T. (2007) Chapel Hill bisphenol A expert panel consensus statement: Integration of mechanisms, effects in animals and potential to impact human health at current levels of exposure. Reprod. Toxicol. 24, 131− 138. (5) Loganathan, S. N., and Kannan, K. (2011) Occurrence of bisphenol A in indoor dust from two locations in the eastern United States and implications for human exposures. Arch. Environ. Contam. Toxicol. 61, 68−73. (6) Calafat, A. M., Ye, X., Wong, L. Y., Reidy, J. A., and Needham, L. L. (2008) Exposure of the U.S. population to bisphenol A and 4tertiary-octylphenol: 2003−2004. Environ. Health Perspect. 116, 39−44. J
DOI: 10.1021/acs.chemrestox.5b00243 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Chemical Research in Toxicology
Article
(25) Osimitz, T. G., Eldridge, M. L., Sloter, E., Welsh, W., Ai, N., Sayler, G. S., Menn, F., and Toole, C. (2012) Lack of androgenicity and estrogenicity of the three monomers used in Eastman’s Tritan copolyesters. Food Chem. Toxicol. 50, 2196−2205. (26) Bittner, G., and Yaniger, S. (2012) Comment on “Lack of androgenicity and estrogenicity of the three monomers used in Eastman’s Tritan copolyesters” by Osimitz et al.(2012). Food Chem. Toxicol. 50, 4236−4237. (27) Shen, J., Xu, L., Fang, H., Richard, A. M., Bray, J. D., Judson, R. S., Zhou, G., Colatsky, T. J., Aungst, J. L., Teng, C., Harris, S. C., Ge, W., Dai, S. Y., Su, Z., Jacobs, A. C., Harrouk, W., Perkins, R., Tong, W., and Hong, H. (2013) EADB: an estrogenic activity database for assessing potential endocrine activity. Toxicol. Sci. 135, 277−291. (28) Ding, D., Xu, L., Fang, H., Hong, H., Perkins, R., Harris, S., Bearden, E. D., Shi, L., and Tong, W. (2010) The EDKB: an established knowledge base for endocrine disrupting chemicals. BMC Bioinf. 11 (Suppl 6), S5. (29) Nicklaus, M. C., Neamati, N., Hong, H., Mazumder, A., Sunder, S., Chen, J., Milne, G. W., and Pommier, Y. (1997) HIV-1 integrase pharmacophore: discovery of inhibitors through three-dimensional database searching. J. Med. Chem. 40, 920−929. (30) Hong, H., Neamati, N., Wang, S., Nicklaus, M. C., Mazumder, A., Zhao, H., Burke, T. R., Pommier, Y., and Milne, G. W. A. (1997) Discovery of HIV-1 Integrase Inhibitors by Pharmacophore Searching. J. Med. Chem. 40, 930−936. (31) Neamati, N., Hong, H., Sunder, S., Milne, G. W., and Pommier, Y. (1997) Potent inhibitors of human immunodeficiency virus type 1 integrase: identification of a novel four-point pharmacophore and tetracyclines as novel inhibitors. Mol. Pharmacol. 52, 1041−1055. (32) Hong, H., Neamati, N., Winslow, H. E., Christensen, J. L., Orr, A., Pommier, Y., and Milne, G. W. (1998) Identification of HIV-1 integrase inhibitors based on a four-point pharmacophore. Antiviral Chem. Chemother. 9, 461−472. (33) Hong, H., Fang, H., Xie, Q., Perkins, R., Sheehan, D. M., and Tong, W. (2003) Comparative molecular field analysis (CoMFA) model using a large diverse set of natural, synthetic and environmental chemicals for binding to the androgen receptor. SAR QSAR Environ. Res. 14, 373−388. (34) Hong, H., Tong, W., Fang, H., Shi, L., Xie, Q., Wu, J., Perkins, R., Walker, J. D., Branham, W., and Sheehan, D. M. (2002) Prediction of estrogen receptor binding for 58,000 chemicals using an integrated system of a tree-based model with structural alerts. Environ. Health Perspect. 110, 29−36. (35) Shi, L., Tong, W., Fang, H., Xie, Q., Hong, H., Perkins, R., Wu, J., Tu, M., Blair, R. M., Branham, W. S., Waller, C., Walker, J., and Sheehan, D. M. (2002) An integrated ″4-phase″ approach for setting endocrine disruption screening priorities–phase I and II predictions of estrogen receptor binding affinity. SAR QSAR Environ. Res. 13, 69−88. (36) Tong, W., Hong, H., Xie, Q., Shi, L., Fang, H., and Perkins, R. (2005) Assessing QSAR limitations-A regulatory perspective. Curr. Comput.-Aided Drug Des. 1, 195−205. (37) Hong, H., Tong, W., Xie, Q., Fang, H., and Perkins, R. (2005) An in silico ensemble method for lead discovery: decision forest. SAR QSAR Environ. Res. 16, 339−347. (38) McPhail, B., Tie, Y., Hong, H., Pearce, B. A., Schnackenberg, L. K., Ge, W., Valerio, L. G., Fuscoe, J. C., Tong, W., Buzatu, D. A., Wilkes, J. G., Fowler, B. A., Demchuk, E., and Beger, R. D. (2012) Modeling chemical interaction profiles: I. Spectral data-activity relationship and structure-activity relationship models for inhibitors and non-inhibitors of cytochrome P450 CYP3A4 and CYP2D6 isozymes. Molecules 17, 3383−3406. (39) Shen, J., Zhang, W., Fang, H., Perkins, R., Tong, W., and Hong, H. (2013) Homology modeling, molecular docking, and molecular dynamics simulations elucidated alpha-fetoprotein binding modes. BMC Bioinf. 14 (Suppl14), S6. (40) Ng, H. W., Zhang, W., Shu, M., Luo, H., Ge, W., Perkins, R., Tong, W., and Hong, H. (2014) Competitive molecular docking approach for predicting estrogen receptor subtype alpha agonists and antagonists. BMC Bioinf. 15, S4.
(41) Vernall, A. J., Stoddart, L. A., Briddon, S. J., Ng, H. W., Laughton, C. A., Doughty, S. W., Hill, S. J., and Kellam, B. (2013) Conversion of a non-selective adenosine receptor antagonist into A 3selective high affinity fluorescent probes using peptide-based linkers. Org. Biomol. Chem. 11, 5673−5682. (42) Gibbons, D. L., Pricl, S., Posocco, P., Laurini, E., Fermeglia, M., Sun, H., Talpaz, M., Donato, N., and Quintás-Cardama, A. (2014) Molecular dynamics reveal BCR-ABL1 polymutants as a unique mechanism of resistance to PAN-BCR-ABL1 kinase inhibitor therapy. Proc. Natl. Acad. Sci. U. S. A. 111, 3550−3555. (43) Ng, H. W., Laughton, C. A., and Doughty, S. W. (2013) Molecular dynamics simulations of the adenosine A2a receptor: structural stability, sampling, and convergence. J. Chem. Inf. Model. 53, 1168−1178. (44) Ng, H. W., Laughton, C. A., and Doughty, S. W. (2014) Molecular dynamics simulations of the adenosine A2a receptor in POPC and POPE lipid bilayers: effects of membrane on protein behavior. J. Chem. Inf. Model. 54, 573−581. (45) Shaw, D. E., Maragakis, P., Lindorff-Larsen, K., Piana, S., Dror, R. O., Eastwood, M. P., Bank, J. A., Jumper, J. M., Salmon, J. K., Shan, Y., and Wriggers, W. (2010) Atomic-Level Characterization of the Structural Dynamics of Proteins. Science 330, 341−346. (46) Compadre, C. M., Singh, A., Thakkar, S., Zheng, G., Breen, P. J., Ghosh, S., Kiaei, M., Boerma, M., Varughese, K. I., and Hauer-Jensen, M. (2014) Molecular Dynamics Guided Design of Tocoflexol: A New Radioprotectant Tocotrienol with Enhanced Bioavailability. Drug Dev. Res. 75, 10−22. (47) Compadre, C. M., Singh, A., Thakkar, S., Zheng, G., Breen, P. J., Ghosh, S., Kiaei, M., Boerma, M., Varughese, K. I., and Hauer-Jensen, M. (2014) Applications of Molecular Dynamics to the Design of Radioprotectant Tocotrienols with Enhanced Bioavailability. Drug Dev. Res. 75, 10−22. (48) Warnmark, A., Treuter, E., Gustafsson, J. A., Hubbard, R. E., Brzozowski, A. M., and Pike, A. C. (2002) Interaction of transcriptional intermediary factor 2 nuclear receptor box peptides with the coactivator binding site of estrogen receptor alpha. J. Biol. Chem. 277, 21862−21868. (49) Olsson, M. H., Søndergaard, C. R., Rostkowski, M., and Jensen, J. H. (2011) PROPKA3: consistent treatment of internal and surface residues in empirical p K a predictions. J. Chem. Theory Comput. 7, 525−537. (50) Banks, J. L., Beard, H. S., Cao, Y., Cho, A. E., Damm, W., Farid, R., Felts, A. K., Halgren, T. A., Mainz, D. T., Maple, J. R., Murphy, R., Philipp, D. M., Repasky, M. P., Zhang, L. Y., Berne, B. J., Friesner, R. A., Gallicchio, E., and Levy, R. M. (2005) Integrated Modeling Program, Applied Chemical Theory (IMPACT). J. Comput. Chem. 26, 1752−1780. (51) Greenwood, J. R., Calkins, D., Sullivan, A. P., and Shelley, J. C. (2010) Towards the comprehensive, rapid, and accurate prediction of the favorable tautomeric states of drug-like molecules in aqueous solution. J. Comput.-Aided Mol. Des. 24, 591−604. (52) Friesner, R. A., Murphy, R. B., Repasky, M. P., Frye, L. L., Greenwood, J. R., Halgren, T. A., Sanschagrin, P. C., and Mainz, D. T. (2006) Extra precision glide: docking and scoring incorporating a model of hydrophobic enclosure for protein-ligand complexes. J. Med. Chem. 49, 6177−6196. (53) Bowers, K. J., Chow, E., Xu, H., Dror, R. O., Eastwood, M. P., Gregersen, B. A., Klepeis, J. L., Kolossvary, I., Moraes, M. A., and Sacerdoti, F. D. (2006) Scalable algorithms for molecular dynamics simulations on commodity clusters, in SC 2006 Conference, Proceedings of the ACM/IEEE, pp 43−43, IEEE, New York. (54) Shivakumar, D., Williams, J., Wu, Y., Damm, W., Shelley, J., and Sherman, W. (2010) Prediction of absolute solvation free energies using molecular dynamics free energy perturbation and the OPLS force field. J. Chem. Theory Comput. 6, 1509−1519. (55) Liao, C., Liu, F., Alomirah, H., Loi, V. D., Mohd, M. A., Moon, H.-B., Nakata, H., and Kannan, K. (2012) Bisphenol S in urine from the United States and seven Asian countries: occurrence and human exposures. Environ. Sci. Technol. 46, 6860−6866. K
DOI: 10.1021/acs.chemrestox.5b00243 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Chemical Research in Toxicology
Article
(56) Kuruto-Niwa, R., Nozawa, R., Miyakoshi, T., Shiozawa, T., and Terao, Y. (2005) Estrogenic activity of alkylphenols, bisphenol S, and their chlorinated derivatives using a GFP expression system. Environ. Toxicol. Pharmacol. 19, 121−130. (57) Kinch, C. D., Ibhazehiebo, K., Jeong, J.-H., Habibi, H. R., and Kurrasch, D. M. (2015) Low-dose exposure to bisphenol A and replacement bisphenol S induces precocious hypothalamic neurogenesis in embryonic zebrafish. Proc. Natl. Acad. Sci. U. S. A. 112, 1475−1480. (58) Konkel, L. (2013) thermal reaction: the Spread of Bisphenol S via paper products. Environ. Health Perspect. 121, a76. (59) Viñas, R., and Watson, C. S. (2013) Bisphenol S disrupts estradiol-induced nongenomic signaling in a rat pituitary cell line: effects on cell functions. Environ. Health Perspect. 121, 352−358. (60) Ji, K., Hong, S., Kho, Y., and Choi, K. (2013) Effects of Bisphenol S Exposure on Endocrine Functions and Reproduction of Zebrafish. Environ. Sci. Technol. 47, 8793−8800. (61) Hong, H., Branham, W. S., Dial, S. L., Moland, C. L., Fang, H., Shen, J., Perkins, R., Sheehan, D., and Tong, W. (2012) Rat alphaFetoprotein binding affinities of a large set of structurally diverse chemicals elucidated the relationships between structures and binding affinities. Chem. Res. Toxicol. 25, 2553−2566. (62) Hong, H., Branham, W. S., Ng, H. W., Moland, C. L., Dial, S. L., Fang, H., Perkins, R., Sheehan, D., and Tong, W. (2015) Human Sex Hormone-Binding Globulin Binding Affinities of 125 Structurally Diverse Chemicals and Comparison with Their Binding to Androgen Receptor, Estrogen Receptor, and alpha-Fetoprotein. Toxicol. Sci. 143, 333−348. (63) Blair, R. M., Fang, H., Branham, W. S., Hass, B. S., Dial, S. L., Moland, C. L., Tong, W., Shi, L., Perkins, R., and Sheehan, D. M. (2000) The estrogen receptor relative binding affinities of 188 natural and xenochemicals: structural diversity of ligands. Toxicol. Sci. 54, 138−153. (64) Roy, U., and Luck, L. A. (2007) Molecular modeling of estrogen receptor using molecular operating environment. Biochem. Mol. Biol. Educ. 35, 238−243. (65) Haber, L., and Patterson, J. (2005) Report of an independent peer review of an acrylonitrile risk assessment. Hum. Exp. Toxicol. 24, 487−527. (66) Delfosse, V., Grimaldi, M., Pons, J. L., Boulahtouf, A., le Maire, A., Cavailles, V., Labesse, G., Bourguet, W., and Balaguer, P. (2012) Structural and mechanistic insights into bisphenols action provide guidelines for risk assessment and discovery of bisphenol A substitutes. Proc. Natl. Acad. Sci. U. S. A. 109, 14930−14935. (67) Okuda, K., Fukuuchi, T., Takiguchi, M., and Yoshihara, S. i. (2011) Novel pathway of metabolic activation of bisphenol A-related compounds for estrogenic activity. Drug Metab. Dispos. 39, 1696− 1703. (68) Quesnot, N., Bucher, S., Fromenty, B., and Robin, M.-A. (2014) Modulation of Metabolizing Enzymes by Bisphenol A in Human and Animal Models. Chem. Res. Toxicol. 27, 1463−1473. (69) Baker, M. E., and Chandsawangbhuwana, C. (2012) 3D models of MBP, a biologically active metabolite of bisphenol A, in human estrogen receptor α and estrogen receptor β. PLoS One 7, e46078. (70) Nishihara, T., Takatori, S., Kitagawa, Y., and Hori, S. (2000) Estrogenic Activities of 517 Chemicals by Yeast Two-Hybrid Assay. J. Health Sci. 46, 282−298. (71) Lakind, J. S., and Birnbaum, L. S. (2010) Out of the frying pan and out of the fire: the indispensable role of exposure science in avoiding risks from replacement chemicals. J. Exposure Sci. Environ. Epidemiol. 20, 115−116.
L
DOI: 10.1021/acs.chemrestox.5b00243 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX