Mechanistic insights into the synergistic activation of the RXR-PXR heterodimer by endocrine disruptor mixtures

Abstract

Humans are chronically exposed to mixtures of xenobiotics referred to as endocrine-disrupting chemicals (EDCs). A vast body of literature links exposure to these chemicals with increased incidences of reproductive, metabolic, or neurological disorders. Moreover, recent data demonstrate that, when used in combination, chemicals have outcomes that cannot be predicted from their individual behavior. In its heterodimeric form with the retinoid X receptor (RXR), the pregnane X receptor (PXR) plays an essential role in controlling the mammalian xenobiotic response and mediates both beneficial and detrimental effects. Our previous work shed light on a mechanism by which a binary mixture of xenobiotics activates PXR in a synergistic fashion. Structural analysis revealed that mutual stabilization of the compounds within the ligand-binding pocket of PXR accounts for the enhancement of their binding affinity. In order to identify and characterize additional active mixtures, we combined a set of cell-based, biophysical, structural, and in vivo approaches. Our study reveals features that confirm the binding promiscuity of this receptor and its ability to accommodate bipartite ligands. We reveal previously unidentified binding mechanisms involving dynamic structural transitions and covalent coupling and report four binary mixtures eliciting graded synergistic activities. Last, we demonstrate that the robust activity obtained with two synergizing PXR ligands can be enhanced further in the presence of RXR environmental ligands. Our study reveals insights as to how low-dose EDC mixtures may alter physiology through interaction with RXR-PXR and potentially several other nuclear receptor heterodimers.

Keywords: cocktail effect; endocrine disruptor; low dose; mixture; synergy.

Fig. 1.

Compound activity in transactivation assays using the HG5LN GAL4-PXR-LBD cell line. (A) Cells were exposed to different concentrations of test compounds. Assays were performed in quadruplicate in at least three independent experiments, and data are expressed as mean (±SEM). Note that, because of toxicity issues, CLO could not be used at concentrations above 3 µM. (B–F) Cells were exposed to compounds in binary mixtures as indicated. Black dashed lines represent the theoretical activation curves obtained for the additive combination of individual compound activities calculated using the Bliss independence model (20). Assays were performed in quadruplicate in at least three independent experiments, and data are expressed as mean (±SEM).

Fig. 2.

Chemicals bind to PXR with varied binding mechanisms and pocket occupancies. (A–I) Close-up view of the LBP of PXR bound to the various test compounds. The chemicals (color code for carbon atoms as in Fig. 1A) and residues belonging to the aromatic cage (gray) are shown as sticks. Other residues are displayed as lines. Oxygen, nitrogen, sulfur, and chlorine atoms are colored in red, blue, yellow, and green, respectively. Residues and secondary structural elements discussed in the text are labeled.

Fig. 3.

Differential occupancy of the PXR LBP. The four PXR LBP subpockets defined in this study are displayed and labeled. One compound representative of each group is shown within its subpocket. Both the ligands and their pockets are colored following the code used in Fig. 1A.

Fig. 4.

Analysis of ligand-binding cooperativity by native MS. (A) Native ESI-MS was used to characterize PXR LBD in the presence of six binary mixtures as indicated. Asterisks indicate acetate adducts. (B) Relative abundance distributions of unliganded and liganded PXR (L1, EE2 or E2; L2, TNC, END, CC, HEP, CLO, or ZEA) derived from native MS analyses of PXR LBD (5 µM) in the presence of 2 molar equivalents of EE2/TNC, E2/END, E2/CC, E2/HEP, E2/CLO, and E2/ZEA ligand mixtures (10 µM each).

Fig. 5.

PXR-driven transactivation is synergistically activated in vivo. Transient transactivation of human PXR LBD. Somatic transgenesis in tadpole tail muscle of two GFP–reporter constructs (CMV-Gal4 DBD-PXR LBD + 5UAS-GFP) was done at day 0. After 8 d of treatment with daily renewal, GFP expression was measured from pictures using ImageJ. (A–J) Magnification (40×) of representative tail skeletal muscle fibers (one per group) after treatment with (A) solvent, (B) PXR synthetic agonist SR12813 1 µM, (C) 17α-ethinylestradiol EE2 1 µM, (D) 17β-estradiol E2 1 µM, (E) TNC 1 µM, (F) TNC + EE2 1 µM each, (G) CC 0.1 µM, (H) E2 1 µM + CC 0.1 µM, (I) END 5 nM, and (J) E2 1 µM + END 5 nM. (K and L) Quantification of fluorescence and statistical analysis. Experiments were performed at least three times (n > 8 for each condition), providing similar results (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

Fig. 6.

The binding modes of binary mixtures as revealed by X-ray crystallography. Close-up view of the LBP of PXR bound to 17β-estradiol (E2) and (A) CC, (B) HEP, (C) END, and (D) CLO. The compounds (color code for carbon atoms as in Fig. 1A) and residues belonging to the aromatic cage (gray) are shown as sticks. Other residues are displayed as lines. Oxygen, nitrogen, sulfur, and chlorine atoms are colored in red, blue, yellow, and green, respectively. Residues and secondary structural elements discussed in the text are labeled. (Insets) Comparison of the binding modes of CC, END, and CLO in the presence and absence of E2.

Fig. 7.

Synergistic activation of the RXR–PXR heterodimer by mixtures of RXR and PXR ligands. (A) LS174T-PXR 3A4 luciferase cells were treated by compounds either alone or in combination as indicated. Assays were performed in triplicate in at least three independent experiments, and data are expressed as mean (±SEM). The black dashed line represents the theoretical activation curve obtained for the additive combination of EE2, TNC, and TBT activities calculated using the Bliss independence model (20). Note that, in the EE2/TNC/TBT combination, TBT is 100-fold less concentrated than EE2 and TNC. (B) RT-qPCR analysis of CYP3A4 mRNA expression in control or PXR-overexpressing LS174T cells treated for 48 h by solvent (0.1% DMSO) or the indicated ligands (TNC, EE2, and SR12813 at 3 μM; TBT at 30 nM). Results were obtained from three separate experiments performed in duplicates. Data are expressed as mean (±SEM) compared to DMSO-treated cells. (C) Fluorescence anisotropy analysis showing the relative affinity of the fluorescein-labeled SRC-1 NID for RXR–PXR LBD heterodimer in the presence of saturating concentrations of reference and test compounds alone or in mixture.

Fig. 8.

Model for synergistic activation of the PXR signaling pathway by ternary mixtures of EDCs. Together with previous ones, our study shows that the LBP of PXR displays several specific structural features accounting for its role as a sensor responding to a wide variety of chemicals. They comprise an aromatic π-trap (π trap), two reactive cysteine residues (C207, C284) available for covalent coupling or highly dynamic and conformable secondary structural elements (Dyn. SSE). Upon binding of an environmental ligand to PXR, transcriptional coactivators are recruited by the DNA-bound RXR–PXR heterodimer via the interaction of one of their LxxLL binding motifs (gray ovals) with the coactivator binding site (CBS) of PXR, thus inducing the transcription of target genes. However, rather than single molecules, human exposure involves a broad mix of chemicals, which may act in a synergistic manner, possibly through the two converging mechanisms identified in this work. Both rely on the binding cooperativity of: 1) a second compound that physically assembles with the first one into the PXR LBP to form a supramolecular ligand with improved functional properties in regard to those of its individual components, or 2) a second coactivator LxxLL motif upon RXR activation insuring a robust interaction of the coactivator with the RXR–PXR heterodimer (highlighted by red ovals).