Synergistic activation of human pregnane X receptor by binary cocktails of pharmaceutical and environmental compounds

Humans are chronically exposed to multiple exogenous substances, including environmental pollutants, drugs and dietary components. Many of these compounds are suspected to impact human health, and their combination in complex mixtures could exacerbate their harmful effects. Here we demonstrate that a pharmaceutical oestrogen and a persistent organochlorine pesticide, both exhibiting low efficacy when studied separately, cooperatively bind to the pregnane X receptor, leading to synergistic activation. Biophysical analysis shows that each ligand enhances the binding affinity of the other, so the binary mixture induces a substantial biological response at doses at which each chemical individually is inactive. High-resolution crystal structures reveal the structural basis for the observed cooperativity. Our results suggest that the formation of ‘supramolecular ligands' within the ligand-binding pocket of nuclear receptors contributes to the synergistic toxic effect of chemical mixtures, which may have broad implications for the fields of endocrine disruption, toxicology and chemical risk assessment.

E xternal compounds (xenobiotics) to which humans are continuously exposed include environmental pollutants, drugs or dietary components. Many of them belong to the structurally heterogeneous group of endocrine-disrupting chemicals (EDCs) that trigger adverse health effects by mimicking or antagonizing the action of endogenous signalling molecules [1][2][3] . More than 20 years of experimental and epidemiological studies have highlighted the pivotal role of nuclear receptors (NRs) in transducing many of the harmful effects of EDCs 4,5 . NRs belong to a large family of evolutionarily related transcription factors that control complex gene networks, resulting in profound physiological changes 6 . They contain a ligand-binding domain (LBD) that responds to a wide variety of endogenous hormonal and metabolic ligands. The endocrine-disrupting action of chemicals relies mostly on their ability to substitute for natural ligands and deregulate NR signalling, causing reproductive, proliferative and metabolic disorders [7][8][9] . In addition, human exposure to mixtures of xenobiotics can induce unpredictable additive, antagonistic or synergistic adverse effects 10 . Yet, the molecular mechanisms underlying these cocktail effects are largely unknown.
To explore the outcome of combined exposure to chemicals and establish a detailed mechanistic understanding of this emerging paradigm for EDC action, we focused our attention on the xenoreceptor PXR (pregnane X receptor; NR1I2) which has been identified by the US Environmental Protection Agency ToxCast's program as a major front-line target of chemicals. This NR is a key regulator of the body's defense against foreign substances. It forms heterodimers with the retinoid X receptor (RXR) and binds to PXR responsive elements (PXRE) in the regulatory regions of target genes. Upon activation by xenobiotics (for example, bisphenol-A, organophosphate pesticides, alkylphenols, rifampicin), PXR interacts with coactivators, such as the steroid receptor coactivator-1 (SRC-1), and transcriptionally upregulates major detoxification genes such as the phase I cytochrome P450 enzyme CYP3A4 (ref. 11), which metabolizes more than half of all drugs in clinical use. On the other hand, the interaction of PXR with EDCs has been linked to an increased risk of cardiovascular 12 and metabolic 13 diseases.
Here, using compound screening followed by extensive functional analysis, we demonstrate that the combined use of the pesticide trans-nonachlor (TNC) and the active component of contraceptive pills, 17a-ethinylestradiol (EE2), produces synergistic effects on PXR activation and expression of its endogenous target gene CYP3A4. Biophysical characterization reveals that EE2 and TNC bind cooperatively to PXR and that the binary mixture has considerably improved functional properties over each of the compounds alone. Crystallographic analysis shows that reciprocal stabilization of the compounds in the ligand-binding pocket (LBP) of the receptor accounts for the enhanced efficacy and potency of the chemical mixture. We therefore propose the concept of a 'supramolecular ligand' that defines a molecular assembly consisting of two or more compounds that interact with each other inside the LBP of a receptor, resulting in the creation of a new entity with improved functional characteristics in regard to those of its individual components.

Results
Synergistic activation of PXR by EE2 and TNC. Using mediumthroughput ligand screening and a mammalian (HeLa) cell-based activation assay (HG5LN GAL4-PXR-LBD reporter cell line) 14,15 , we monitored the agonistic potential of 40 chemicals either alone or in binary mixtures (Supplementary Table 1). Most combinations exhibited additive effects, inducing 50-60% of the transactivation seen with the cholesterol-lowering drug SR12813 (Fig. 1), a potent and well-characterized PXR full agonist [16][17][18] (EC 50 in the 100-200 nM range) used as a reference in all our experiments. However, we observed that the combined use of the organochlorine pesticide TNC and the synthetic oestrogen 17a-ethinylestradiol (EE2), the active component of contraceptive pills, produced more than an additive effect with an induction level of 90% ( Fig. 1 and Supplementary Fig. 1). This preliminary observation was confirmed by dose-response experiments. The combination of TNC and EE2 led to a shift of the corresponding activation curve by one order of magnitude towards the low concentrations (Fig. 2a). We next examined the effect of TNC, EE2, and their combination in human liver (HepG2) and colon (LS174T) carcinoma cell lines containing the full-length PXR and the CYP3A4-XREM luciferase reporter plasmids. Again, we found that co-treatment yielded to much stronger activation of PXR as monitored by the transactivation of the CYP3A4 reporter gene (Fig. 2b,c). When used simultaneously, the two compounds activated PXR in a synergistic fashion as illustrated by the theoretical activation curve obtained for the additive combination of EE2 and TNC activities (Fig. 2a-c, red dashed lines calculated using the Bliss independence model 19 ). Note that synergism was also observed with other steroidal and organochlorine compound combinations. Supplementary Fig. 2 shows two representative examples associating either EE2 and cis-chlordane, or TNC with the natural hormone 17b-estradiol ( Fig. 1).
Cocktail effect on CYP3A4 expression and activity. Consistent with the reporter gene assays, the effectiveness of the individual compounds at increasing the level of endogenous CYP3A4 mRNA was drastically enhanced by the addition of the second compound in LS174T cells expressing PXR but not in control cells (Fig. 2d). We then compared the ability of EE2 and TNC, alone or in combination, to increase CYP3A4 gene expression in freshly isolated primary human hepatocytes (PHHs) in culture, the most biologically relevant model regarding PXR function. As shown in Fig. 2e, the CYP3A4 mRNA expression was considerably augmented when both EE2 and TNC were used. Accordingly, the enhanced induction of the CYP3A4 protein by the binary mixture (Fig. 2f, upper panel and Supplementary  Fig. 3) closely correlated with higher CYP3A4 enzymatic activity (Fig. 2f, lower panel).
As a whole, cell-based assays clearly show that EE2 and TNC act as poor PXR agonists when used separately whereas their combination triggers PXR activation nearly as efficiently as the reference agonist SR12813. Notably, synergism could be observed in various cellular contexts and with different compound combinations, including the natural hormone 17b-estradiol.  Coactivator recruitment by PXR upon co-treatment. In order to decipher the molecular mechanism involved in the synergistic activation of PXR-mediated transcription by EE2 and TNC, we characterized their impact on coactivator recruitment. For this purpose, we used fluorescence anisotropy assays with the purified PXR/RXR LBD heterodimers and the fluorescein-labelled NR interaction domain (NID) of SRC-1. We found that, as expected, the PXR agonist SR12813 efficiently enhanced SRC-1 recruitment (Fig. 3a). Interestingly, EE2 and TNC had modest effects on their own but their combination produced a strong increase in the coactivator recruitment, similar to that observed with SR12813. Comparable results were obtained from mammalian two-hybrid experiments (Fig. 3b), suggesting that the active form of the receptor is highly stabilized in presence of both ligands.   Simultaneous binding of EE2 and TNC to PXR. To characterize further the interaction of PXR with EE2 and TNC, we used electrospray ionization mass spectrometry (ESI-MS) under native conditions. Analysed separately, TNC and EE2 were shown to interact with PXR with affinities in the low micromolar range; about 80 and 70% complex were detected when PXR (10 mM) was incubated with 2 molar equivalent excess of TNC or EE2, respectively ( Fig. 4a-c). TNC bound to PXR in a 1:1 molar ratio, while EE2 was found to interact with the receptor with 1:1 and 1:2 binding stoichiometries. Binding specificity of the two compounds for PXR ligand-binding pocket was assessed using competition experiments against SR12813. As shown in Supplementary Fig. 4, SR12813 was able to compete efficiently for EE2 and TNC. No non-specific binding was detected with TNC, whereas some residual EE2 was observed in presence of SR12813, suggesting that part of the secondary EE2 binding sites could reside outside the PXR LBP. Analysis of PXR after incubation with the binary cocktail showed a large predominance of a ternary complex corresponding to PXR interacting with EE2 and TNC in a 1:1:1 molar ratio; the global range of affinity of the two compounds binding to PXR was sub-micromolar ( Fig. 4d). Together, these data showed that PXR can accommodate EE2 and TNC simultaneously and suggested greater binding affinity of the binary mixture compared to individual compounds.
EE2 and TNC bind cooperatively to PXR. We next assessed the binding characteristics of EE2 and TNC, alone or in mixture, to PXR. Competitive binding assays using time resolved fluorescence resonance energy transfer between a fluorescent PXR ligand and purified human PXR-LBD (LanthaScreen TR-FRET PXR Competitive Binding Assay) showed that the binary mixture binds 10-to 30-fold more avidly to PXR than TNC and EE2 alone (Fig. 5a). To define the receptor-ligand interactions by a direct method, we next measured the binding affinity constants of various EE2 and TNC combinations by isothermal titration calorimetry (ITC). Before ITC experiments, dose-dependent aggregation of the compounds in working buffer was assessed using dynamic light scattering. Critical aggregation concentration (CAC) was estimated to be in the 50-100 mM range when compounds were diluted serially from 500 mM down to 3 mM in Tris-HCl 20 mM, pH 8.5, NaCl 200 mM, TCEP 1 mM, Tween20 0.05%, DMSO 5%. Accordingly, we cannot rule out that when used at 200-300 mM in some experimental conditions, a fraction of the compounds undergo aggregation despite the dilution process occurring during titration experiments (see Methods). Nevertheless, ITC analyses clearly converged towards similar conclusion of cooperative binding of EE2 and TNC to PXR resulting in increased global binding affinity. ITC data showed much better affinities of EE2 and TNC for PXR when the receptor was pre-incubated with TNC or EE2, respectively (K d values in  ARTICLE the high-nanomolar range compared to the mid-micromolar range interaction of the compounds tested independently; Fig. 5b-e). Simultaneous titration of EE2 and TNC against PXR was associated with a binding affinity of the binary mixture that was substantially higher than those of the compounds tested alone (Fig. 5f), corroborating the observations derived from ESI-MS and LanthaScreen experiments. Taken altogether, these in vitro data performed with purified material strongly support the notion that both the cooperative binding and synergism observed with EE2 and TNC rely on direct interactions with PXR and not on other cellular mechanisms such as cellular influx/efflux, metabolism, or binding to other cellular targets.
Structural basis for supramolecular ligand activity. To gain structure-based insight into the binding mode of EE2 and TNC to PXR, we solved the crystal structures of PXR-LBD in complex with EE2, TNC, or both at 2.00 Å, 2.55 Å, and 2.25 Å resolution, respectively (Table 1 and Supplementary Fig. 5). In all cases, the PXR-LBD adopts the canonical active conformation, with the C-terminal activation helix H12 capping the LBP (Fig. 6a and Supplementary Fig. 6a,b). Whereas EE2 could be precisely placed in the electron densities obtained for both the PXR-EE2 and PXR-TNC-EE2 complexes, TNC could be positioned unambiguously in the ternary complex only (Fig. 6b and Supplementary Fig. 6c,d). The poorly defined electron density of TNC in the PXR-TNC complex likely reflects high mobility of this ligand in the pocket. In contrast to EE2 which is engaged in several hydrogen bonds maintaining the ligand in a defined location, TNC has no polar group and likely adopts an ensemble of different positions and/or orientations in the LBP.
EE2 binds closely adjacent to H12 with a binding mode that is reminiscent of that seen for 17b-estradiol 20 ( Supplementary  Fig. 7a), both in presence or absence of TNC. The 3-hydroxyl group on the A-ring of EE2 forms an hydrogen bond with S247 (helix H3), whereas the 17b-hydroxyl group on the D-ring is hydrogen bonded with R410 in helix H11 and the main chain oxygen atom of D205 from helix H2' (Fig. 6c). The position of R410 is further stabilized by a network of hydrogen bonds involving S208 (helix H2') and E321 from the loop preceding H7. The remaining contacts between EE2 and the protein involve van der Waals interactions with several hydrophobic residues. The particular position of EE2 in a restricted region of PXR LBP leaves a significant portion of the pocket unoccupied and available for additional interactions (Supplementary Fig. 6a,c). As seen in the ternary complex, this empty region can accommodate one molecule of TNC which is stabilized in a well-defined position via a number of interactions with both EE2 and the protein. As illustrated in Fig. 6c, eight van der Waals contacts of 3.7 to 4.5 Å in length could be measured between EE2 and TNC. These inter-ligand contacts generate a mutual stabilization of the compounds in the LBP and account for the enhanced binding affinity of the binary mixture (Fig. 5). On the protein side, TNC forms essentially nonpolar interactions with ten residues, including F281, F288, W299, Y306, M323 (Fig. 6c) and two weak halogen bonds with Q285 and C284 from helix H5.
In keeping with the robust agonistic activity of the binary mixture, superposition of the supramolecular ligand-bound structure with that of PXR in complex with SR12813 (ref. 21) or rifampicin 22 reveals that the binding sites of EE2 and TNC overlap those of the two PXR agonists (Supplementary Fig. 7b,c). Hence, the molecular assembly of EE2 and TNC into the LBP of PXR can be regarded as a supramolecular ligand whose functional properties rely on intermolecular interactions and differ from those of the individual components.

Discussion
Most current knowledge of EDCs action is derived from data sets that use single molecule exposure, with few studies taking into account the more realistic situation where humans are simultaneously and chronically exposed to low doses of multiple EDCs. Indeed, a growing number of studies indicate that human risk assessment approaches based on single molecule exposure underestimate the risk for adverse effects of chemicals 10 . The evaluation of mixture effects by regulatory bodies has been mainly hampered by the huge numbers of pollutants and potential combinations, but also by the lack of knowledge of the molecular pathways involved. This study provides both the first detailed mechanistic explanation and a proof of concept for the synergistic action of two compounds via their simultaneous interaction with the LBP of a NR. Our results provide not only new insight as to how low doses of EDCs or drugs may affect physiology and homeostasis, but also suggest that the concomitant binding of chemicals stabilizing each other in NR LBPs likely corresponds to one of the possible mechanisms accounting for the cocktail effect by which compounds toxicity is exacerbated.
Reports describing the simultaneous binding of two or more compounds to a common protein binding site are very few in number. One of the rare examples is that of the Staphylococcus aureus multidrug-binding transcription repressor QacR which was shown to bind concomitantly to ethidium and proflavin. However no cooperative binding mechanism was observed in this case 23 . Other examples include the cytochrome CYP3A4 bound to two molecules of ketoconazole 24 or the peroxisome proliferator activated receptor gamma (PPARg) which can accommodate two copies of FMOC-L-Leucine 25 , or endogenous fatty acids 26 . However, in both cases the bound molecules were of the same type and the possibility of synergism between the two ligands was not addressed. Besides the interaction of several compounds with a unique binding site, two studies reported on the oestrogen antagonist 4-hydroxytamoxifen and a synthetic ligand binding to two alternate sites of the oestrogen receptor b and PPARg LBDs, respectively 27,28 . Again, no cooperativity was reported between these dual binding sites. Our study shows that individually, EE2 and TNC are too small to make all the necessary interactions ensuring high binding affinity and effective stabilization of the active conformation of the receptor. In contrast, when associated in a binary mixture, EE2 and TNC form a supramolecular ligand that fills a larger fraction of the PXR LBP, and displays apparent functional properties (e.g. activity and binding affinity) comparable to those of the full PXR agonists, SR12813 and rifampicin.
Cloning of PXR orthologues from human, rabbit, rat and mouse has shown that the ligand-binding domain has diverged considerably between the different species, leading to specific ligand-binding and activation profiles 29,30 . This species-specific induction pattern of PXR is possibly an adaptive response to the environment and a need to adjust toxicological responses to endogenously produced substances. Therefore, additional studies should be performed to test the potential EE2/TNC synergy in vivo in these species. Notably future studies will be completed in mice, along with the use of PXR-knock-out models and long term exposures with these compounds, to confirm the role of PXR and the physiological relevance of the EE2/TNC synergism on drug and bile acid metabolism, hepatic steatosis, or liver regeneration for instance.
Large LBPs such as those of PXR or PPARg are obviously predisposed to accommodate several molecules at the same time. However, in contrast to the perception that most NRs possess a well-defined pocket to account for the specific binding of a unique endogenous ligand, several structural studies have revealed that their LBPs exhibit a greater conformational flexibility than previously thought. Importantly, this structural adaptability allows the oestrogen (ERa) 31 , thyroid (TRb) 32 , or glucocorticoid (GR) 33 receptors to expand their binding pocket and accommodate much larger ligands 34 . Another interesting example is the oestrogen-related receptor gamma (ERRg) whose LBP appears too small to bind any ligand, yet can expand it to accommodate rather large compounds 35 . Considering the huge chemical and size diversity of xenobiotics and the high structural plasticity of NR LBPs, the mechanism defined for PXR is likely to occur also with other members of the NR superfamily, with broad reaching implications in the fields of endocrine disruption, chemicals risk assessment and toxicology. The development of novel supramolecular chemistry-based therapeutic options could also benefit from the discovery reported here.

Methods
Ligands. TNC, EE2, E2 and CC were purchased from Sigma-Aldrich. SR12813 was purchased from Tocris Bioscience. The provenance of compounds used for medium-throughput screening is described in Supplementary Table 1. All compound stock solutions were prepared at 10 mM in DMSO. ARTICLE Plasmids. The [ À 7.8/ À 7.2] XREM-CYP3A4 (XREM-CYP3A4 pGL3b) luciferase plasmid has been described previously 36 . The 17Mx5-Glob-LUC containing five GAL4 binding sites upstream of the luciferase reporter gene and the VP16 (activation domain)-TIF2 (amino acids 624-869 containing three LXXLL motifs) are gifts from Hinrich Gronemeyer (IGBMC, Illkirch, France). The pM-hPXR expression vector was generated by inserting a PCR fragment corresponding to the full-length hPXR in the pM vector (CLONTECH).
Medium-throughput screening. HG5LN GAL4-PXR reporter cell lines (100 ml) were seeded at a density of 25,000 cells per well in 96-well white opaque tissue culture plates (Greiner CellStar). 24 h later, negative (DMSO 0,1%) and positive (SR12813 3 mM) controls, and the 40 tested compounds (50 ml) were added into the wells as indicated in Supplementary Table 1.
Then the ligand to be tested in combination with the different ligands was added (50 ml). Cells were incubated at 37°C for 16 h. At the end of the incubation period, culture medium was replaced with medium containing 3.10 À 4 M luciferin. Luciferase activity was measured for 2 s in intact living cells using a plate reader (PerkinElmer Luminometer).
Transactivation assays. HG5LN GAL4-PXR, HEPG2-and LS174T-PXR reporter cell lines were seeded at a density of 25,000 cells per well in 96-well white opaque tissue culture plates (Greiner CellStar). Compounds to be tested were added 24 h later, and cells were incubated at 37°C for 16 h. At the end of the incubation period, culture medium was replaced with medium containing 3.10 À 4 M luciferin.
Luciferase activity was measured for 2 s in intact living cells using a plate reader (PerkinElmer Luminometer). EC 50 values were measured using GraphPad Prism (GraphPad Software Inc.).
Preparation of primary human hepatocytes. Liver samples were obtained from liver resections performed in adult patients for medical reasons unrelated to our research program or from donors when the liver was considered unsuitable for organ transplantation. The use of human specimens for scientific purposes was approved by the French National Ethics Committee. Written or oral informed consent was obtained from each patient or family prior to surgery. The clinical characteristics of the liver donors are presented in Supplementary Table 2. Hepatocytes were isolated by using a two-step perfusion protocol and cultured as described previously 39 . Briefly, several veins apparent on the cut edge of the lobectomy were used for sequential perfusions with a washing buffer (10 mM HEPES, 136 mM NaCl, 5 mM KCl, 0.5% glucose, pH 7.6), with a calcium chelating buffer (washing buffer complemented with 0.5 mM EGTA), with the washing buffer and then with a collagenase IV solution (washing buffer supplemented with collagenase IV, 200 U ml À 1 , Sigma) for 20 min. After gentle disruption of the tissue and filtration through a 250 mm mesh, the post-collagenase homogenate was centrifuged at low speed (50 g for 5 min) to pellet the hepatocytes. Hepatocytes were seeded in collagen type-I coated dishes (Becton Dickinson) at 1.7 10 5 cells per cm 2 in a hormonally and chemically defined medium for short term culture consisting of a mixture of William's E and Ham's F-12 (1:1 in volume) and additives as previously described 40 supplemented with 2% heat inactivated fetal calf serum (FCS). After overnight attachment, the medium and unattached cells were removed and fresh medium without FCS was added. Hepatocytes were treated with the molecules at day 2 post-seeding for 24 or 48 h.
Reverse transcription-PCR and real-time quantitative PCR. Total RNA was extracted from 200,000 cells using the RNeasy mini kit (Qiagen) or Trizol reagent (Invitrogen) and treated with DNAse-1. The first strand cDNA was synthetized using Superscript II (Invitrogen) or MMLV (Invitrogen) and random hexamers. Primer sequences are provided in Supplementary Table 3. Gene expression was normalized with GAPDH and the level of expression was compared with the mean level of the corresponding gene expression in untreated cells and expressed as n-fold ratio. The relative amount of RNA was calculated with the 2 DD CT method.
Measurement of CYP3A4 activity. Primary human hepatocytes were seeded in 96-well plates and treated with the test compounds or vehicle for 48 h. CYP3A4 activity was detected using the P450-Glo CYP3A4 luciferin-IPA Enzyme Activity Kit (Promega) according to the manufacturer's instructions. Cell number was normalized using CellTiter-Glo Luminescent Cell Viability Assay (Promega).
Statistical analyses. For the analysis of the correlation between parametric data, Stutent's t-test was used, while the Mann-Whitney U-test was used for nonparametric data. Differences were considered statistically significant when P-values were ***Po0.001 **Po0.01 *Po0.05.
Preparation of PXR/RXR for fluorescence anisotropy assays. The human PXR-LBD (130-434; pDB-His-MBP vector) and the human RXR-LBD (223-462; pET-3a vector) were overproduced in Escherichia coli BL21(DE3). Cells were grown at 37°C in LB medium supplemented with the appropriate antibiotic until OD 600 reached 0.6. Expression of T7 polymerase was induced by addition of isopropyl-b-D-thiogalactoside to a final concentration of 0.5 mM. After 16 h of incubation at 20°C, cell cultures were harvested by centrifugation at 6,000g for 15 min. Cell pellets from 3 L of His 6 -MBP-PXR-LBD culture and from 1 l of nontagged RXR-LBD culture were resuspended in 150 ml buffer A (20 mM HEPES pH 7.5, 250 mM NaCl, 15 mM imidazole, 5% (v/v) glycerol) supplemented with a protease inhibitor cocktail tablet (cOmplete, EDTA-free, Roche). The suspension was then lysed by sonication and centrifuged at 18,000g at 4°C for 30 min. The supernatant was loaded onto a Ni 2 þ -affinity column (HisTrap 5 ml; GE Healthcare) equilibrated with buffer A, using the Ä KTA pure system (GE Healthcare).
The column was washed with 20 volumes of buffer A and 20 volumes of buffer A containing 50 mM imidazole. The His 6 -MBP-PXR/RXR heterodimer was eluted with buffer A containing 100 mM imidazole. The fractions containing the heterodimer were pooled and incubated for 48 h at 4°C with the tobacco etch virus protease to cleave the His 6 -MBP tag, and then further reloaded onto the Ni 2 þaffinity column to eliminate the tag. The flow-through containing the PXR/RXR heterodimer was then purified using a gel filtration column (Superdex 200 16/60; GE Healthcare) equilibrated with buffer B (10 mM Tris-HCl pH 7.5, 250 mM NaCl, 1 mM DTT and 5% (v/v) glycerol). The fluorescein-labelled SRC-1 fragment (residues 570-780) was prepared using the protocol previously described 41 . Briefly, the SRC-1fragment was expressed in E. coli BL21(DE3) as a fusion with an inducible self-splicing intein (Sce VMA) and a chitin-binding domain using the vector pTYB1 (New England Biolabs). The cells were grown at 37°C to 0.6 at OD 600 and then at 17°C overnight. The fusion protein was purified using chitin resin (New England Biolabs). Intein cleavage was induced using 2 mM cys-fluor with 50 mM MESNA, releasing C-terminally labelled SRC-1. Excess cys-fluor was removed using a phenyl sepharose resin (Amersham). SRC-1 was further purified using size-exclusion chromatography. Thin-layer chromatography was used to confirm that there was no free fluorescein label in the purified samples.
Steady-state fluorescence anisotropy. Measure of the binding affinities of the coactivator fragment for the PXR/RXR heterodimer in the absence and presence of various ligands was performed using a Safire2 microplate reader (TECAN). The excitation wavelength was set at 470 nm and emission measured at 530 nm for the fluorescein-tagged fragment. Assays were carried out in the following buffer solution 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM TCEP and 5% (v/v) glycerol. We initiated the measurements at the highest concentration of protein (20 mM) and diluted the protein sample twofold successively with the buffer solution. For each point of the titration curve, the protein sample was mixed with 5 nM of fluorescent fragment and a 3 molar excess of ligand (60 mM final concentrations). Binding data were fitted using a sigmoidal dose-response model using GraphPad Prism (GraphPad Software Inc.).
Mammalian two-hybrid experiments. Gal4-hPXR and VP16-TIF2 interaction was monitored on 17Mx5-Glob-LUC reporter construct. Transient transfections assays were performed in U2OS cells using Jet-PEI (Ozyme) according to manufacturer's instructions. Luciferase assays were performed with the Promega dual-reporter kit, according to the manufacturer's instructions. Renilla luciferase encoded by the normalization vector phRLTK (Promega) was used as internal control for firefly luciferase normalization.