Characterizing the Peroxisome Proliferator-Activated Receptor (PPARγ) Ligand Binding Potential of Several Major Flame Retardants, Their Metabolites, and Chemical Mixtures in House Dust

Background: Accumulating evidence has shown that some environmental contaminants can alter adipogenesis and act as obesogens. Many of these contaminants act via the activation of the peroxisome proliferator-activated receptor γ (PPARγ) nuclear receptor. Objectives: Our goal was to determine the PPARγ ligand binding potency of several major flame retardants, including polybrominated diphenyl ethers (PBDEs), halogenated phenols and bisphenols, and their metabolites. Ligand binding activity of indoor dust and its bioactivated extracts were also investigated. Methods: We used a commercially available fluorescence polarization ligand binding assay to investigate the binding potency of flame retardants and dust extracts to human PPARγ ligand-binding domain. Rosiglitazone was used as a positive control. Results: Most of the tested compounds exhibited dose-dependent binding to PPARγ. Mono(2-ethylhexyl) tetrabromophthalate, halogenated bisphenols and phenols, and hydroxylated PBDEs were found to be potent PPARγ ligands. The most potent compound was 3-OH-BDE-47, with an IC50 (concentration required to reduce effect by 50%) of 0.24 μM. The extent of halogenation and the position of the hydroxyl group strongly affected binding. In the dust samples, 21 of the 24 samples tested showed significant binding potency at a concentration of 3 mg dust equivalent (DEQ)/mL. A 3–16% increase in PPARγ binding potency was observed following bioactivation of the dust using rat hepatic S9 fractions. Conclusion: Our results suggest that several flame retardants are potential PPARγ ligands and that metabolism may lead to increased binding affinity. The PPARγ binding activity of house dust extracts at levels comparable to human exposure warrants further studies into agonistic or antagonistic activities and their potential health effects. Citation: Fang M, Webster TF, Ferguson PL, Stapleton HM. 2015. Characterizing the peroxisome proliferator-activated receptor (PPARγ) ligand binding potential of several major flame retardants, their metabolites, and chemical mixtures in house dust. Environ Health Perspect 123:166–172; http://dx.doi.org/10.1289/ehp.1408522


PPARγ competitive binding assay
Fluorescence polarization (FP) assays allow ligand binding to be quantified in a homogeneous format without perturbing equilibrium by physical separation of bound vs. free ligand, which is advantageous for the measurement of low-affinity interactions (Rossi and Taylor 2011). In this study, a commercially available high-throughput ligand binding assay (PolarScreen TM PPARγcompetitor assay kit, Invitrogen) was used to investigate the binding potency of tested compounds to PPARγ LBD. The kit uses the human-derived recombinant PPARγ-LBD tagged with a N-terminal GST-tag and a selective fluorescent PPARγ ligand (PPARγ Green) in a 384well plate (Corning Thermowell GOLD 3756). The test compounds were dissolved in DMSO and dosed into each well with a final volume of 40 µL containing 38 nM PPARγ LBD and 1.25 nM PPAR-Green as recommended by the protocol. We used 3% DMSO in the final incubation mixture to reduce potential effects on polarization by the solvent. The well plate was mixed and incubated for 3 hours at room temperature to reach equilibrium. A SpectraMax M5 plate reader was used in polarization mode with 485 nm excitation and 535 nm emission wavelenth. To measure ligand binding, we quantified polarization (mP) of the bound protein using the following equation: mP = 10 3 *(I p -I s )/(I p +I s ) [1] where I p and I s are the fluorescence intensity of emissions that are parallel (P) and perpendicular (S) to the excitation light; respectively (Rossi and Taylor 2011).

Quality assurance/quality control
For each dose level of tested compounds, triplicate samples were prepared and each well was read five times in the plate reader. DMSO and rosiglitazone were run alongside each batch as a control and positive control, respectively. For the dust samples, a procedural blank was prepared and preceded alongside the dust extracts to examine the background contamination. Since variability of FP was observed across the batches, normalization to DMSO control was conducted when comparing between batches. The potent ligands such as 3-OH-BDEs were tested three times at different days with different well plates.

Operation of Gel Permeation Chromatography
The flow rate was set to 10 mL/min and DCM was used as the mobile phase. To test the elution profile, several compounds ranging from small MW halogenated phenols to large MW TBPH were loaded onto GPC and fractions were collected at 0-10 min, and then 2 min for each fraction until 30 mins. As shown in Supplemental Material, Table S1, most of the chemicals were eluted after 14 mins except TBPH. To recover the highest amount from the mixture, a fraction from 12-28 min was collected, concentrated, and solvent exchanged into DMSO.

Bioactivation of Dust Samples
The incubation method was modified based on previous studies (Montaño et al. 2012) and the flow chart was shown in Figure S3. To increase the amount of the metabolites, incubation was up-scaled and performed in a borosilicate glass tube in 3 mL phosphate buffer (PB, 100 mM, pH 7.4), S9 fraction (1 mg protein/mL), and dust extract in ~30 µL DMSO (~33 mg dust/mL PB). 10 mM DTT and 6 mM magnesium chloride were added. After 5 minutes pre-incubation in a shaking water bath at 37℃, the reaction was initiated with the addition of 50 µL of 60 mM NADPH in PB. Additional 50 µL of 60 mM NADPH in PB were added after 60 mins.
Metabolism was stopped after 120 minutes by denaturation of microsomal protein with 150 µL of ice-cold 6M HCl. An additional sample for each dust extract was incubated with inactive S9 by 150 µL of ice-cold 6M HCl before incubation and run alongside as the comparison control.
The challenges in the bioactivation of dust samples for the PPARγ binding assay includes the low metabolic rate, interference from coextracts in S9 fraction, and FB of dust (Montaño et al. 2013). In our preliminary study, we found that S9 extract could interfere with the PPARγ polarization assay [see Supplemental Material, Figure 2 (b), 2 (c), 2 (d), and Figure S9], which might be caused by the lipids which can work as natural ligands for PPARγ. Therefore, further cleanup was needed to remove the coextracts that interfere with PPARγ binding. In this study, the extraction and cleanup method for the metabolites was modified according to a recently Metabolites were extracted with 2 × 2 mL ethyl acetate and 1× 2 mL hexane : methyl tert-butyl ether (1:1, v/v). TBBA, TBBPA and MEHP were tested for the recovery of dextran and LRA cleanup. Good recovery (> 80%) was observed in the dextran assisted extraction and no elution of any compound in the LRA. Therefore, only dextran extraction was used in the study.
The low metabolic rate does not generate sufficient metabolites to activate the PPARγ binding and further concentration of the metabolites was needed. However, the dust matrix will be concentrated and cause fairly high fluorescence background to interfere with the PPARγ binding.
Thus, further cleanup of the dust matrix was still needed. In our previous studies, we found that the phenolic extraction could retain most of the dust matrix in the organic solvent. Furthermore, most of the known PPARγ agonists or antagonists are the chemicals with polar groups such as -OH, -COOH, and -NH 2 . If we hypothesize that charged chemicals including phenolic or carboxylic metabolites were the major metabolites, it would be possible to reduce the dust matrix without losing most of the metabolites as a tradeoff. In this study, the extracts from dextran assisted liquid-liquid extraction was blown down to near dryness under nitrogen gas and the sample was reconstituted in 1 mL DCM, which was then extracted with 3×1 mL deionized water with a pH of ~13. The pool of the 3 mL water was then acidified with 6 M HCl to pH < 3 and extracted with 2 × 2 mL ethyl acetate and 1 × 2 mL hexane : methyl tert-butyl ether (1:1, v/v).
The final extracts were dried and reconstituted with 200 µL DMSO. As shown in Figure S2 (b), the phenolic extraction can reduce the matrix background of the dust extracts to the level which was only slightly above the DMSO control and would not greatly interfere with the binding assay. The recoveries of tested compounds such as TBBPA, MEHP and TBBA were > 85% using the above-mentioned method. Therefore, the combination of dextran-assisted extraction and phenolic extraction make the PPARγ binding assay feasible for the bioactivated dust extracts.

Performance of the Bioactivation of Dust
The challenges in the bioactivation of dust samples for the PPARγ binding assay includes the low metabolic rate, interference from coextracts in S9 fraction, and FB of dust. As shown in Supplemental Material Figure S9, the natural ligands such as fatty acid in the S9 fraction can competitively inhibit ~20% of the binding between PPARγ-LBD and PPARγ Green at a concentration of 250 µg protein/mL. Dextran was used to selectively remove the lipid during extraction (Montaño et al. 2012). The result showed the dextran assisted extraction could partially remove the interference from the coextracts, which was close to 90% of the DMSO control. The application of incubation with inactive S9 running alongside as control could correct 6 this interference. Phenolic extraction could further reduce the FP background by approximately five times than that of the raw extract, leaving the FP of dust with a concentration of 6 mg DEQ/mL close to DMSO control (see Supplemental Material Figure S2 (b)). As shown in Supplemental Material Figure S3, MEHP could be formed by the incubation of dust with S9 fraction and the formation rate was approximately 70 pmol/mg protein/min, which was slightly higher than that of 25 µM DEHP pure chemical. However, the formation rate was decreased when the dosing amount of dust was 100 mg/mL, which might due to the decrease of enzyme activity caused by the impurities in dust extracts. Therefore, we used ~33 mg DEQ/mL throughout the incubation experiment in this study to maintain a high metabolic rate.