In silico predication of nuclear hormone receptors for organic pollutants by homology modeling and molecular docking

Toxicology Letters 191 (2009) 69–73

Contents lists available at ScienceDirect

Toxicology Letters journal homepage: www.elsevier.com/locate/toxlet

In silico predication of nuclear hormone receptors for organic pollutants by homology modeling and molecular docking Bing Wu, Yan Zhang, Jie Kong, Xuxiang Zhang, Shupei Cheng ∗ State Key Laboratory of Pollution Control & Resource Reuse, School of the Environment, Nanjing University, Nanjing, 210093, China

a r t i c l e

i n f o

Article history: Received 10 July 2009 Received in revised form 4 August 2009 Accepted 6 August 2009 Available online 14 August 2009 Keywords: Nuclear hormone receptors Homology modeling Molecular docking Organic pollutants

a b s t r a c t Homology modeling and molecular docking were used to in silico predict the rat nuclear hormone receptors of different organic pollutants. Rat aryl hydrocarbon receptor (rAhR), constitutive androstane receptor (rCAR) and pregnane X receptor (rPXR) were chosen as the target nuclear receptors. 3D models of ligand binding domains of rAhR, rCAR and rPXR were constructed by MODELLER 9V6 and assessed by the Procheck and Prosa 2003. Surflex-Dock program was applied to bind the different organic pollutants into the three receptors to predict their affinities. The results of docking experiments demonstrated that three polybrominated dibenzofurans (PBDFs, including TretaBDF, PentaBDF and HexaBDF) and 3,3 ,4,4 ,5 pentachlorobiphenyl (PCB126) would be better categorized by rAhR-dependent mechanism, but four polybrominated diphenyl ethers (PBDEs, including BDE47, BDE80, BDE99 and BDE153) and 2,2 ,4,4 ,5,5 hexachlorobiphenyl (PCB153) by rCAR and rPXR-dependent mechanism. For benzo(a)pyrene and pyrene, they have high affinities with the three target receptors, which suggests that “crosstalk” among the receptors might occur during the receptor induction. The results of this study are consistent with those of animal experiments reported by previous literatures, which suggest that homology modeling and molecular docking would have the potential to predict the nuclear hormone receptors of environmental pollutants. © 2009 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Aryl hydrocarbon receptor (AhR), constitutive androstane receptor (CAR) and pregnane X receptor (PXR) belong to the nuclear hormone receptors which could bind and be activated by a large number of endogenous and xenobiotic ligands (Baroukiabi et al., 2007; Janosek et al., 2006; Kakizaki et al., 2008). The ligand activated nuclear receptors AhR, CAR and PXR could bind to their cognate DNA elements and then activate the transcription of cytochrome P450 1A (CYP1A), CYP2B and CYP3A, respectively (Jacobs et al., 2003). The expression of cytochrome P450 enzymes often acts to detoxify poisonous xenobiotics. However, in some cases, the intermediates in xenobiotic metabolism could themselves be the cause of toxic effects. Thus, it is important to study the interactions of xenobiotics and different nuclear receptors in order to analyze the metabolic process and toxicity of xenobiotics by cytochrome P450. Many xenobiotics, especially environmental contaminants, have been proved that they can reversibly bind into the special nuclear hormone receptors and activate the function of these receptors in the animal experiments. For example, AhR has high affinities

∗ Corresponding author. Tel.: +86 25 83595995; fax: +86 25 83595995. E-mail address: [email protected] (S. Cheng). 0378-4274/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.toxlet.2009.08.005

towards the halogenated aromatic hydrocarbons (HAHs) (Hahn, 1998) and polycyclic aromatic hydrocarbons (PAHs) (Savouret et al., 2003). Sanders et al. (2005) found that polybrominated diphenyl ethers (PBDEs) mediated toxicity would be better categorized by CAR and PXR-dependent mechanisms. However, for the increasing number of environmental contaminants, animal experiments are time-consuming and lack hypothesis-driven aim. The approach of homology modeling and molecular docking might be a potential tool to provide the hypothesis-driven aim of animal experiments. Both techniques have been applied to study the interactions between ligands and nuclear receptors (such as AhR, CAR and PXR). However, these studies always focused on the structural and functional characterization of the special receptors (Pandini et al., 2007; Tirona et al., 2004; Windshugel et al., 2007). Few studies were performed to compare the binding affinities of one ligand to the different receptors so as to predict the special receptor of a certain target ligand. In this study, homology modeling was used to construct the 3D model of ligand binding domains (LBDs) of rat AhR, CAR and PXR. Then, the predicted models were applied to bind the different organic pollutants by molecular docking. Free energy of binding was considered as the criteria to identify the binding affinities to analyze the specialization of interaction between receptor and ligand. The data from animal experiments was used to validate the results of docking experiments. This study is a useful attempt to

70

B. Wu et al. / Toxicology Letters 191 (2009) 69–73

in silico predict the special nuclear hormone receptors for different organic pollutants and might provide a potential tool to search hypothesis-driven aim of animal experiments. 2. Computational methods 2.1. Sequence alignment and homology modeling The primary sequences of rAhR, rCAR and rPXR were obtained from the Swiss-Prot database (http://www.expasy.ch/sprot/). The LBDs of three proteins were chosen as the target sequences (Table 1). BLAST algorithm against Protein Data Bank (PDB) (http://blast.ncbi.nlm.nih.gov/Blast.cgi) was used to carry out the sequence homology searches. The sequence and crystal structure of each template protein were extracted from Swiss-Prot and PDB databases (Table 1). Multiple sequence alignments among the target and template sequences were performed by ClustalW 2.0.10 program with default parameters (http://www.ebi.ac.uk/Tools/clustalw2/index.html). MODELLER 9v6 program (Sali and Blundell, 1993) was used to construct initial 3D structural model of rAhR, rCAR and rPXR LBDs. MODELLER can implement comparative protein structure modeling by satisfying spatial restraints in terms of probability density functions. In this study, 50 runs of modeler were carried out using standard parameters and the outcomes were ranked on the basis of the internal scoring function of the program. The model with the highest score was chosen as the target model. Then, energy minimizations of chosen models were performed using GROMACS 3.3 according to the software protocol (Pandini et al., 2007; Van der Spoel et al., 2005). 2.2. Model evaluation

2.4. Flexible molecular docking The Surflex-Dock program of Sybyl 7.3 was employed to dock the target pollutants into the rAhR, rCAR and rPXR LBDs, respectively. Surflex-Dock could automatically dock ligands into a receptor’s ligand binding site using a protomol based approach and assess the affinity by an empirically derived scoring function. The method has been proved to be one of the most effective docking techniques (Kellenberger et al., 2004). In this study, prior to docking, the hydrogen atoms were added in predicted models using the Biopolymer modulators of Sybyl 7.3. The Kollman-all atom charges were assigned to protein atoms. Protomol for Surflex-Dock was generated according to the software protocol. Two important factors, “proto bloat” and “proto thresh”, can significantly affect the size and extent of the protomol. “Proto thresh” determines how far the protomol extents into the concavity of the target site, while “proto bloat” impacts how far the protomol extents outside of the concavity (Holt et al., 2008). Considering the purposes of this study, “proto thresh” was set to 0.5 and “proto bloat” was set to 1 for all protomols generated. Other parameters were employed with default setting in all runs. Protomols were visualized with Sybyl 7.3 to ensure proper coverage of the desired target area. Surflex-Dock’s scoring function, which contains hydrophobic, polar, repulsive, entropic, and salvation terms, was trained to estimate the dissociation constant (Kd ) expressed in −log(Kd ) unit (Jain, 2007). After running Surflex-Dock, the scores of docked conformers could be ranked in a molecular spread sheet. The best score conformer would be selected as the docking results. In this study, the scores of binding were converted to the free energy of binding (kcal/mol) in order to better compare the binding affinities between ligand and three target receptors, and to predict the preference receptor. The free energy of binding was calculated as following equation, where RT = 0.59 kcal/mol (Holt et al., 2008): free energy of binding = RT ln(10−pKd ).

Model evaluation involved analysis of geometry, stereochemistry, and energy distribution of the predicted models. Firstly, 3D visualization programs SwissPdbViewer 4.01 (Guex and Peitsch, 1997) and Rasmol 2.7.4 (Goodsell, 2005) were carried out to peruse the reliability of the alignment and modeling of variable surface loops of predicated models (Lutfullah et al., 2008). Then, stereochemical quality of the homology models was checked by Procheck program (Laskowski et al., 1993). The energetic architecture of model folds was determined by Prosa 2003 program (Van Brussel et al., 1998). 2.3. Target pollutants According to the previous literatures, 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), 16,17-androstene-3-ol (ATE) and 16a-carbonitrile (PCN) have been proved to be the known ligands of rAhR, rCAR and rPXR, respectively (Hahn, 1998; Shan et al., 2004; Tirona et al., 2004). Thus, they were chosen as the reference ligands of docking experiment. A total of 11 environmental pollutants, whose data of animal experiment were available, were selected as target ligands to in silico predict their special nuclear hormone receptors, including 3,4-benzo(a)pyrene (BaP), pyrene, 3,3 ,4,4 ,5 -pentachlorobiphenyl (PCB126), (PCB153), 2,3,7,8-tetrabromodibenzofuran 2,2 ,4,4 ,5,5 -hexachlorobiphenyl (TetraBDF), 1,2,3,7,8-pentabromodibenzofuran (PentaBDF), 1,2,3,4,7,8hexabromodibenzofuran (HexaBDF), 2,2 ,4,4 -tetrabromodiphenyl ether (BDE47), 3,3,5,5-tetrabromodiphenyl ether (BDE80), 2,2 ,4,4 ,5 -pentabromodiphenyl ether (BDE99) and 2,2 ,4,4 ,5,5 -hexabromodiphenyl ether (BDE153). Initial conformations of compounds were obtained from the Chemical Book Database (http://www.chemicalbook.com/ProductIndex.aspx). Compounds not included in the Database were constructed from the structures of similar compounds (Yang et al., 2009). The geometries of these compounds were subsequently optimized in Sybyl 7.3 (Tripos Inc., St. Louis, MO). Relevant energy minimization of target compounds was conducted using Tripos Force Field (distance-dependent dielectric) with atom charge calculated by Gasteiger–Hückel method to reach a final energy convergence gradient value of 0.001 kcal/mol. The optimized structures offered reasonable starting conformations for further molecular docking.

3. Results and discussion 3.1. Construction of receptor model Crystal structures, as potential templates of target proteins, were obtained from the BLAST search for rAhR, rCAR and rPXR, respectively. Template selection was performed on the basis of sequence similarity, residues completeness, crystal resolution and functional similarity. Table 1 shows the basic information on the selected templates used in this study. For rAhR, among the available candidate templates, the sequence identities between target and templates were low (≤30%). Pandini et al. (2007) found that, despite low level of sequence similarity, mouse AhR could be well constructed using the crystal structures of HIf-2a and ARNT by homology modeling. HIf-2a and ARNT, like the AhR, belong to the basic helix-loop-helix (bHLH)/Per-Arnt-Sim- (PAS) family of transcriptional factors that are key regulators of gene expression networks underlying many essential biological processes. They have similar functions. Thus, three crystal structures of HIf-2a and ARNT LBD (PDB ID: 1P97, 2A24 and 3F1N) were selected as multi-templates to construct the 3D model of rAhR LBD. The sequence identities between rAhR LBD and its templates were 30% (Table 1). In general, sequence identities of 30% are enough to construct the 3D model of target proteins through the homology modeling. Fig. 1 shows the ribbon schematic representation of the final modeled structure of the rAhR LBD.

Table 1 Basic information on the target and template proteins. Target

Swiss-Prot ID

Residues of LBDa

PDB ID

Protein

Sequence identity

rAhR

P41738

273-384

1P97 2A24 3F1N

Human hypoxia-inducible factor, HIF-2␣ HIF-2a/ARNT PAS-B Heterodimer Heterodimer of HIF2 alpha and ARNT C-terminal PAS domains

30% 30% 30%

rCAR

Q9QUS1

114-345

1XNX

Mouse constitutive androstane receptor, mCAR

89%

rPXR

Q9R1A7

202-431

3CTB

Human tethered PXR-LBD/SRC-1p apoprotein

76%

a

the residues of chosen ligand binding domains.

Template

B. Wu et al. / Toxicology Letters 191 (2009) 69–73

Fig. 1. Ribbon schematic representations of the modeled structures of rAhR, rCAR and rPXR, which are viewed by Pytho Molecule Viewer (http://autodock.scripps.edu/).

Crystal structures of mouse CAR (mCAR, PDB ID: 1XNX) and human PXR apoprotein (hPXR, PDB ID: 3CTB) were chosen as the templates for the modeling of rCAR and rPXR LBDs, respectively (Table 1). The amino acid sequences identity between rCAR and mCAR is 89%, and that between rPXR and hPXR is 76%. Almost the entire structure was considered as conserved region for both alignments. Such high sequence identity and conservations should be enough to construct reliable 3D models in the light of homology modeling method. The final modeled structures of the rCAR and rPXR LBDs are displayed in Fig. 1. 3.2. Structural analysis of predicted models Final homology models of rAhR, rCAR and rPXR were assessed from energetic and geometric criteria with Prosa 2003 and Procheck. Fig. 2 shows the graphs of Prosa energy versus pro-

71

Fig. 2. z-Score versus protein residue position for the modeled structures of rAhR, rCAR and rPXR from Prosa 2003.

tein residue position of three predicted models, respectively. Prosa energy was calculated as z-score, which indicated the quality of protein structures. In general, the z-score should be below the zero point, which suggests no significant stressed or stained folds with high energies (Lutfullah et al., 2008). As shown in Fig. 2, the z-score of predicted models are all negative at all residue positions, which indicate reasonable side-chain interactions. Assessment of stereochemical properties of main-chain and side-chain residues was performed using the Procheck program. All the three constructed models satisfied stereochemical restraints and passed all criteria implemented in Procheck. Overall scores of the Procheck geometric assessment for rAhR, rCAR and rPXR are −0.16, 0.11 and 0.10, respectively. These values are higher than −0.5, which is considered as the indicator for a high quality structure (Li et al., 2008). Table 2 shows the percentage of ˚– angles in

72

B. Wu et al. / Toxicology Letters 191 (2009) 69–73

Table 2 The percentage of ˚– angles in different Ramchandran regions for the predicted models. Model

Residues in most favored regions

Residues in additional allowed regions

Residues in generously allowed regions

Residues in disallowed regions

rAhR rCAR rPXR

87.9% 97.2% 95.1%

12.1% 2.3% 4.9%

0% 0.5% 0%

0% 0% 0%

different Ramchandran regions for the predicted models. The ˚– angles in most favored Ramchandran regions was 87.9%, 97.2% and 95.1% for rAhR, rCAR and rPXR, respectively. Residues in disallowed regions for three predicted models were not found. Based on the results of energetic and geometric quality assessment, the predicted models of rAhR, rCAR and rPXR are reasonable and reliable for further docking studies. 3.3. Comparative analysis of affinity between pollutants and receptors

The third type of tested pollutants included BaP and pyrene (Fig. 3c). Both pollutants could effectively bind into the three receptors. Thus, the “crosstalk” among the receptors might occur during the nuclear receptor induction (Waxman, 1999). For example, Patel et al. (2007) found that CAR could be up-regulated in primary human hepatocytes in response to AhR activation by BaP. Naspinski et al. (2008) found PXR played an important role in protection against DNA damage by BaP. The data suggests that BaP and pyrene could be characterized by multi-receptors dependent mechanism.

Surflex-Dock program was used to identify the docking affinity between the different pollutants and three predicted models. Free energy of binding calculated from the scores of conformers was used to indicate affinitive ability between ligand and receptor. The lower free energy of binding indicates higher affinity. In this study, the free energy of binding of TCDD into rAhR, ATE into rCAR and PCN into rPXR was −4.76, −5.27 and −4.31 kcal/mol, respectively. The free energy of biding of reference ligands was used as the reference to identify the potential receptor of target pollutants. Fig. 3 shows the free energy of binding between tested pollutants and three nuclear receptors. Eleven tested pollutants could be divided into three types according to the number of free energy of binding from the docking experiment. The first type of pollutants included the PCB126, TretaBDF, PentaBDF and HexaBDF (Fig. 3a). These pollutants had higher affinity with rAhR than rCAR and rPXR. In animal experiments, these pollutants have been proved to be characterized high affinity AhR ligands (Andrysik et al., 2007; Bemis et al., 2005; McAlister et al., 2008; Staskal et al., 2005). PCB126 is coplanar and capable of producing TCDD-like toxicity, including carcinogenicity in experimental animal (Giesy and Kannan, 1998). For three polybrominated dibenzofurans (PBDFs, including TretaBDF, PentaBDF and HexaBDF), they could bind to the rAhR and also exhibited TCDD-like toxicity in experimental animals. However, the results of docking experiment display that HexaBDF has the highest affinity towards rAhR among three PBDFs, which is different from the results of Sanders et al. (2005) who found that TretaBDF was the most potent of the three PBDFs in the binding to rAhR in the rat experiment. The differences might be due to that transport process of PBDFs in the animal body is an important factor for the interactions between PBDFs and AhR. The second type of tested pollutants included PCB153 and four PBDEs (BDE47, BDE80, BDE99 and BDE153), which had higher affinity with rCAR and rPXR than rAhR (Fig. 3b). Both rCAR and rPXR belong to the member of the same nuclear receptor subfamily, sharing around 40% amino acid identity in their LBDs (Jacobs et al., 2003). This might be the reason that both receptors show the similar binding affinity with different ligands. The results of in vitro experiments have proved that non-coplanar PCBs (including PCB153) and PBDE congeners are poor AhR agonists (non-dioxinlike), however, they might induce CYP2Bs and CYP3As through respective activation of the CAR and PXR to reduce toxic and carcinogenic effects of these pollutants (Sanders et al., 2005). The results of animal experiments is quite line with this docking experiments, which suggest that PCB153 and PBDEs would be better categorized by CAR and PXR-dependent mechanisms, rather than the well-characterized AhR-dependent mechanism.

Fig. 3. The free energy of binding between pollutants and predicted models (rAhR, rCAR and rPXR) from Surflex-Dock program.

B. Wu et al. / Toxicology Letters 191 (2009) 69–73

4. Conclusion Homology modeling and molecular docking were used to in silico identify the potential nuclear hormone receptors (rAhR, rCAR and rPXR) of different organic pollutants. The results suggest that PBDFs and PCB126 have higher affinities towards rAhR than that towards rCAR and rPXR. However, PBDEs and PCB153 could effectively activate rCAR and rPXR. For BaP and pyrene, they have high affinities towards three receptors. These results are consistent with the reported conclusion of animal experiments, which suggest the docking experiments might be a potential tool for the prediction of nuclear hormone receptors of environmental pollutants. 5. Conflict of interest None. Acknowledgements This research work was financially supported by Jiangsu Department of Science and Technology (BE200970536) and Nanjing University Innovative Foundation (2006071009). References Andrysik, Z., Vondracek, J., Machala, M., Krcmar, P., Svihalkova-Sindlerova, L., Kranz, A., Weiss, C., Faust, D., Kozubik, A., Dietrich, C., 2007. The aryl hydrocarbon receptor-dependent deregulation of cell cycle control induced by polycyclic aromatic hydrocarbons in rat liver epithelial cells. Mutat. Res., Fundam. Mol. Mech. Mutagen. 615, 87–97. Baroukiabi, R., Coumoul, X., Fernandez-Salgueroc, P.M., 2007. The aryl hydrocarbon receptor, more than a xenobiotic-interacting protein. FEBS Lett. 581, 3608–3615. Bemis, J.C., Nazarenko, D.A., Gasiewicz, T.A., 2005. Coplanar polychlorinated biphenyls activate the aryl hydrocarbon receptor in developing tissues of two TCDD-responsive lacZ mouse lines. Toxicol. Sci. 87, 529–536. Giesy, J.P., Kannan, K., 1998. Dioxin-like and non-dioxin-like toxic effects of polychlorinated biphenyls (PCBs): implications for risk assessment. Crit. Rev. Toxicol. 28, 511–569. Goodsell, D.S., 2005. Representing structural information with RasMol. Curr Protoc Bioinformatics Chapter 5, Unit 5.4. Guex, N., Peitsch, M.C., 1997. SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling. Electrophoresis 18, 2714–2723. Hahn, M.E., 1998. The aryl hydrocarbon receptor: a comparative perspective. Comp. Biochem. Physiol., Part C: Toxicol. Pharmacol. 121, 23–53. Holt, P.A., Chaires, J.B., Trent, J.O., 2008. Molecular docking of intercalators and groove-binders to nucleic acids using Autodock and Surflex. J. Chem. Inf. Model. 48, 1602–1615. Jacobs, M.N., Dickins, M., Lewis, D.F.V., 2003. Homology modelling of the nuclear receptors: human oestrogen receptor beta (hER beta), the human pregnane X receptor (PXR), the Ah receptor (AhR) and the constitutive androstane receptor (CAR) ligand binding domains from the human oestrogen receptor alpha (hER alpha) crystal structure, and the human peroxisome proliferator activated receptor alpha (PPAR alpha) ligand binding domain from the human PPAR gamma crystal structure. J. Steroid Biochem. Mol. Biol. 84, 117– 132. Jain, A.N., 2007. Surflex-Dock 2.1: Robust performance from ligand energetic modeling, ring flexibility, and knowledge-based search. J. Comput. Aided Mol. Des. 21, 281–306.

73

Janosek, J., Hilscherova, K., Blaha, L., Holoubek, I., 2006. Environmental xenobiotics and nuclear receptors—interactions, effects and in vitro assessment. Toxicol. In Vitro 20, 18–37. Kakizaki, S., Yamazaki, Y., Takizawa, D., Negishi, M., 2008. New insights on the xenobiotic-sensing nuclear receptors in liver diseases—CAR and PXR. Curr. Drug Metab. 9, 614–621. Kellenberger, E., Rodrigo, J., Muller, P., Rognan, D., 2004. Comparative evaluation of eight docking tools for docking and virtual screening accuracy. Proteins: Struct., Funct., Bioinf. 57, 225–242. Laskowski, R.A., Macarthur, M.W., Moss, D.S., Thornton, J.M., 1993. Procheck—a program to check the stereochemical quality of protein structures. J. Appl. Crystallogr. 26, 283–291. Li, W.H., Tang, Y., Liu, H., Cheng, J.G., Zhu, W.L., Jiang, H.L., 2008. Probing ligand binding modes of human cytochrome P450 2J2 by homology modeling, molecular dynamics simulation, and flexible molecular docking. Proteins: Struct., Funct., Bioinf. 71, 938–949. Lutfullah, G., Amin, F., Khan, Z., Azhar, N., Azim, M.K., Noor, S., Shoukat, K., 2008. Homology modeling of hemagglutinin/protease [HA/P (vibriolysin)] from Vibrio cholerae: sequence comparision, residue interactions and molecular mechanism. Protein J. 27, 105–114. McAlister, D.R., Fern, M.J., Allen, R.L., 2008. Rapid aryl hydrocarbon receptor based polymerase chain reaction screening assay for polychlorinated dibenzo-pdioxins and furans in soil and sediment. Talanta 74, 992–997. Naspinski, C., Gu, X.S., Zhou, G.D., Mertens-Talcott, S.U., Donnelly, K.C., Tian, Y.A., 2008. Pregnane X receptor protects HepG2 cells from BaP-induced DNA damage. Toxicol. Sci. 104, 67–73. Pandini, A., Denison, M.S., Song, Y.J., Soshilov, A.A., Bonati, L., 2007. Structural and functional characterization of the aryl hydrocarbon receptor ligand binding domain by homology modeling and mutational analysis. Biochemistry (Mosc.) 46, 696–708. Patel, R.D., Hollingshead, B.D., Omiecinski, C.J., Perdew, G.H., 2007. Aryl-hydrocarbon receptor activation regulates constitutive androstane receptor levels in murine and human liver. Hepatology 46, 209–218. Sali, A., Blundell, T.L., 1993. Comparative protein modeling by satisfaction of spatial restraints. J. Mol. Biol. 234, 779–815. Sanders, J.M., Burka, L.T., Smith, C.S., Black, W., James, R., Cunningham, M.L., 2005. Differential expression of CYP1A, 2B, and 3A genes in the F344 rat following exposure to a polybrominated diphenyl ether mixture or individual components. Toxicol. Sci. 88, 127–133. Savouret, J.F., Berdeaux, A., Casper, R.F., 2003. The aryl hydrocarbon receptor and its xenobiotic ligands: a fundamental trigger for cardiovascular diseases. Nutr. Metab. Cardiovasc. Dis. 13, 104–113. Shan, L., Vincent, J., Brunzelle, J.S., Dussault, I., Lin, M., Ianculescu, I., Sherman, M.A., Forman, B.M., Fernandez, E.J., 2004. Structure of the murine constitutive androstane receptor complexed to androstenol: a molecular basis for inverse agonism. Mol. Cell 16, 907–917. Staskal, D.F., Diliberto, J.J., DeVito, M.J., Birnbaum, L.S., 2005. Inhibition of human and rat CYP1A2 by TCDD and dioxin-like chemicals. Toxicol. Sci. 84, 225–231. Tirona, R.G., Leake, B.F., Podust, L.M., Kim, R.B., 2004. Identification of amino acids in rat pregnane X receptor that determine species-specific activation. Mol. Pharmacol. 65, 36–44. Van Brussel, H., Wyns, J., Valckenaers, P., Bongaerts, L., Peeters, P., 1998. Reference architecture for holonic manufacturing systems: PROSA. Comput. Ind. 37, 255–274. Van der Spoel, D., Lindahl, E., Hess, B., Groenhof, G., Mark, A.E., Berendsen, H.J.C., 2005. GROMACS: fast, flexible, and free. J. Comput. Chem. 26, 1701–1718. Waxman, D.J., 1999. P450 gene induction by structurally diverse xenochemicals: central role of nuclear receptors CAR, PXR, and PPAR. Arch. Biochem. Biophys. 369, 11–23. Windshugel, B., Jyrkkarinne, J., Vanamo, J., Poso, A., Honkakoski, P., Sippl, W., 2007. Comparison of homology models and X-ray structures of the nuclear receptor CAR: assessing the structural basis of constitutive activity. J. Mol. Graphics Modell. 25, 644–657. Yang, W., Mu, Y., Giesy, J.P., Zhang, A., Yu, H., 2009. Anti-androgen activity of polybrominated diphenyl ethers determined by comparative molecular similarity indices and molecular docking. Chemosphere 75, 1159–1164.