Impact of Exposure to a Mixture of Organophosphate Esters on Adrenal Cell Phenotype, Lipidome, and Function

Abstract

Organophosphate esters (OPEs) are used primarily as flame retardants and plasticizers. Previously, we reported that adrenal cells are important targets of individual OPEs. However, real-life exposures are to complex mixtures of these chemicals. To address this, we exposed H295R human adrenal cells to varying dilutions (1/1000K to 1/3K) of a Canadian household dust–based OPE mixture for 48 hours and evaluated effects on phenotypic, lipidomic, and functional parameters. Using a high-content screening approach, we assessed phenotypic markers at mixture concentrations at which there was greater than 70% cell survival; the most striking effect of the OPE mixture was a 2.5-fold increase in the total area of lipid droplets. We then determined the response of specific lipid species to OPE exposures with novel, nontargeted lipidomic analysis of isolated lipid droplets. These data revealed that house dust OPEs induced concentration-dependent alterations in the composition of lipid droplets, particularly affecting the triglyceride, diglyceride, phosphatidylcholine, and cholesterol ester subclasses. The steroid-producing function of adrenal cells in the presence or absence of a steroidogenic stimulus, forskolin, was determined. While the production of 17β-estradiol remained unaffected, a slight decrease in testosterone production was observed after stimulation. Conversely, a 2-fold increase in both basal and stimulated cortisol and aldosterone production was observed. Thus, exposure to a house dust–based mixture of OPEs exerts endocrine-disrupting effects on adrenal cells, highlighting the importance of assessing the effects of environmentally relevant mixtures.

Organophosphate esters (OPEs) are human-made chemicals that are used as flame retardants, plasticizers, hydraulic fluids, and coatings for electronic devices in a wide range of commercial and industrial products (1, 2). Since OPEs are additives, and are thus not chemically bound in products, they can be released readily into the surrounding environment via volatilization, abrasion, or dissolution (1, 3, 4). The concentrations of OPEs in dust samples provide critical information about the potential for exposure to OPEs as a consequence of dust inhalation, ingestion, or dermal contact (5-8). OPEs have been detected in dust worldwide (8-10); up to 13 OPEs were detected in Canadian household dust at levels ranging from 5 to 57 700 ng/g (11, 12). Due to the prevalence of exposures, urinary concentrations of OPEs have increased drastically in samples collected in the United States between 2002 and 2015 (13).

Exposures to OPEs have been associated with adverse outcomes in reproduction (14-17), development (18-22), and thyroid homeostasis (23-26). However, commercially available flame retardants are often mixtures. For instance, Firemaster 550, a proprietary flame retardant mixture, contains 2 brominated compounds, 2-ethylhexyl-2,3,4,5-tetrabromobenzoate and bis(2-ethylhexyl) 2,3,4,5-tetrabromophthalate, and 2 OPEs, triphenyl phosphate (TPHP) and isopropylated triphenolphosphate (IPPP) (1, 27). The few experimental studies that have assessed the effects of mixtures have focused primarily on Firemaster 550. Exposure to this mixture has been associated with alterations in neurodevelopment, metabolism, behavior, and adipogenesis (28-34). Recently, Witchey et al (35) reported lipid dysregulation in the neonatal cortex tissue of rats exposed to Firemaster 550; this effect was sex specific and attributed largely to the OPE components in the mixture.

The adrenal gland produces a variety of hormones that play a vital role in regulating blood pressure, metabolism, and the response to stress (36). Steroid hormones synthesized in the adrenal include glucocorticoids, mineralocorticoids, androgens, and estrogens. Improper functioning of the adrenal gland in producing steroid hormones in conditions such as Addison disease, where insufficient cortisol is produced, may have life-threatening consequences (37). In a previous study with H295R adrenal cells, we identified the adrenal gland as an important target of OPEs (38). Depending on the endpoint examined, individual OPEs had differential, and sometimes contrasting, effects on these cells. However, a consistent finding was that all 6 of the OPEs tested increased the total area of lipid droplets (38). Exposure to IPPP doubled the amount of cortisol produced under basal conditions, yet both TPHP and tris(methylphenyl) phosphate (TMPP) decreased cortisol production. Since the effects of exposures to mixtures are more complex than simple additivity (39, 40) and real-world exposures are to mixtures (11, 12, 41), there is a need to assess the possibly adverse effects of exposure to environmentally relevant OPE mixtures.

The objective of this study was to determine how a Canadian household dust–based OPE mixture affects the phenotype, lipid composition, and steroid production of H295R cells. We are particularly interested in lipids as they have crucial roles in signal transduction as cellular energy sources and as structural components (42). In adrenal cells, lipid droplets also act as reservoirs of the cholesterol esters (CEs) that are essential for steroid hormone synthesis (43, 44). We have employed novel, in-depth, lipidomic analysis in conjunction with a comprehensive evaluation of cell phenotypes and function to identify the effects of an environmentally relevant OPE mixtures in adrenal cells.

Materials and Methods

Organophosphate Ester House Dust Mixture

The OPE house dust mixture, prepared by Dr. Michael G. Wade (Health Canada), contained the 13 OPEs that were detected in over 85% of house dust samples collected from 144 urban Canadian homes between 2007 and 2010 (Fig. 1) (11, 12). The relative proportion of each OPE in this mixture is based on their 95th percentile values from these samples. A list of the dilutions used in this study, and their equivalent concentrations in the culture medium, is provided elsewhere (Table S1 (45)). At the lowest dilution tested in this study (1/1000K), the added level of OPEs is comparable to the urine level of OPEs detected in men living in the greater Montreal area (14). The undiluted OPE mixture contained 0.888 mg/μL of OPEs; this is the equivalent of 5.005 g of house dust.

Figure 1.

Composition of the Canadian household dust–based OPE mixture.

Cell Cultures

The H295R human adrenocortical carcinoma cells were purchased from ATCC (CRL-2128, Manassas, VA, USA). The passage numbers of the H295R cells used in this study did not exceed 10. The cell line tested negative for mycoplasma contamination using the MycoAlert Mycoplasma Detection Kit from Lonza (Rockland, ME, USA). Cells were cultured in Corning 75 cm² U-shaped cell culture flasks at 37 °C with 5% CO₂ in 12 mL of phenol red–free DMEM/F-12 medium (Gibco, Burlington, Ontario, Canada). The culture medium was supplemented with Corning ITS + Premix Universal Culture Supplement, Corning Nu-Serum Growth Medium Supplement, and 0.5% 100× penicillin–streptomycin (Wisent Bioproducts, Montreal, Quebec, Canada). Culture medium was renewed every 2 to 3 days.

High-Content Imaging

H295R cells were seeded (10 000 cells/well) in 96-well black PhenoPlates with optically clear flat bottoms (Perkin Elmer, Waltham, MA, USA) that were precoated with 0.2% collagen 1 (3 mg/mL, rat tail) (Gibco, Burlington, Ontario, Canada). To ensure optimal cell adhesion, there was a 24-hour acclimation period prior to chemical treatments. Subsequently, the cells were exposed to either vehicle control (0.5% dimethyl sulfoxide [DMSO]) or 1 of the mixture dilutions (1/1000K, 1/300K, 1/100K, 1/75K, 1/60K, 1/45K, 1/30K, 1/10K, or 1/3K) for 48 hours. On the experimental day, the stock solutions were diluted with medium to the working concentration. DMSO was added to each of the dilutions to maintain a final concentration of 0.5% DMSO in all treatment groups. To assess the effects of the OPE mixture on cell phenotypes, H295R cells were stained with 1 of 4 different combinations of cell-permeable fluorescent dyes for 30 minutes. Further details regarding the cell-permeable fluorescent dyes, including the combinations used and their respective dilutions, can be found elsewhere (Table S2 (45)).

Live cell imaging was conducted using the Operetta high content imaging system (Perkin Elmer) equipped with a nonconfocal 40×high-NA objective. Twelve fields were screened per well. The acquired images were analyzed using the Columbus Image Data Storage and Analysis System (Perkin Elmer). Detailed information on the specific parameters assessed and the settings used for each analysis are provided elsewhere (Supplementary Methods 1.1 (45)).

Lipid Droplet Isolation

H295R cells were seeded in Corning 175 cm² U-shaped cell culture flasks at a seeding density of 10 000 000 cells/flask (adjusted to ensure that enough lipid droplets were collected for lipidomic analysis). Following a 24-hour acclimation period, cells were exposed to the OPE mixture at 1 of 3 different dilutions: 1/300K, 1/100K, or 1/60K. At the end of the 48 hours chemical exposure period, lipid droplets were isolated from the cells using a lipid droplet isolation kit (Cell Biolabs, San Diego, CA, USA; catalog # MET-5011) according to the manufacturer's protocol. The extracted lipid droplets were then stored at −80 °C until further analysis. Samples were sent to the Metabolomics Innovation Center (TMIC; Edmonton, Alberta, Canada) at the University of Alberta for lipidomic analysis.

Lipidomic Profiling

Nontargeted, high-sensitivity lipidomic analysis of lipid droplets was conducted using nano-liquid chromatography mass spectrometry (LC-MS) as detailed by Buzatto et al (46). The lipid extraction procedure was based on a modified Folch liquid–liquid protocol. In brief, 90 μL of each sample was vortexed with 1.1 μL of an internal standard solution and 479 μL of methanol, followed by extraction with 960 μL of dichloromethane. A clean-up step was conducted using 270 μL of water. The samples were then allowed to equilibrate at room temperature for 10 minutes and centrifuged at 16 000g for 10 minutes at 4 °C. The organic layer was collected and dried with a nitrogen blowdown evaporator. The dried samples were resuspended in 4.6 μL of NovaMT MixB, and the mixture was diluted with 40.4 μL of NovaMT MixA (Nova Medical Testing Inc., Edmonton, Alberta, Canada).

A pooled mixture containing all samples was prepared for quality control. Lipids were identified and quantified using the LC-MS technique. This analysis was performed in both positive and negative ion modes with injection duplicates using the Dionex UltiMate 3000 UHPLC instrument (ThermoFisher Scientific) linked to a Bruker Maxis II QTOF Mass Spectrometer (Bruker Corporation). A 3-tier ID approach based on the MS/MS identification (Tiers 1 and 2) and MS match (Tier 3) was used for lipid identification. Tier 1 identification included lipids with MS/MS scores greater than or equal to 500, precursor m/z error smaller than or equal to 20.0 ppm and 5.0 mDa. Tier 2 included lipids with MS/MS scores smaller than 500, while the remaining criteria stayed the same. Features not identified in tiers 1 and 2 were searched in the LipidMaps database (http://www.lipidmaps.org) for putative identification by mass match with an m/z error smaller than or equal to 20.0 ppm and 5.0 mDa (tier 3). Lipid identification followed the guidelines established by the Lipidomics Standards Initiative (https://lipidomics-standards-initiative.org). Additional information regarding the parameters used for data processing and normalization can be found elsewhere (Supplemental Methods section 1.2 (45)). The identified lipids were class-matched to the most similar internal standard.

Measurements of Basal and Stimulated Production of Steroid Hormones

H295R cells were seeded in 96 well plates precoated with 0.2% rat tail collagen 1 at a seeding density of 10 000 cells/well and were allowed to adhere for 24 hours. The cells were then exposed for 48 hours to 200 μL of culture medium containing the OPE mixture at 1 of 4 different dilutions (1/1000K, 1/100K, 1/60K, or 1/30K) in the presence or absence of 10 μM forskolin (Sigma-Aldrich), a steroidogenic stimulus (47). At the end of the exposure period, culture media from duplicate wells (400 μL in total for each condition) were pooled and stored at −80 °C until further analysis. Hoechst 33342 was used to stain cell nuclei; cells were stained for 30 minutes with no washing afterwards to prevent cell loss. Plates were then screened with the Operetta high content imaging system (all fields were screened; 10× magnification). The Columbus Image Data Storage and Analysis System was used to quantify Hoechst-positive cell counts (Supplementary Methods 1.3 (45)).

To assess the effects of the OPE mixture on the steroid hormone producing function of the cells, enzyme-linked immunosorbent assay kits were used to measure the levels of 17β-estradiol, testosterone, aldosterone, and cortisol (17β-estradiol: Tecan [IBL] catalog # RE52041, RRID:AB_2934323; Testosterone: IBL-America catalog # IB79106, RRID:AB_2814981; Aldosterone: Abcam catalog # ab136933, RRID:AB_2895004; Cortisol: Cayman Chemical catalog # 500370, RRID:AB_2935793) in accordance with the manufacturers’ instructions. The SpectraMax Plus 384 microplate reader (Molecular Devices, San Jose, CA, USA) was used to read the enzyme-linked immunosorbent assay plates at a wavelength of 450 nm. The final concentrations were normalized to the number of cells in each well and are presented as picograms or nanograms per 1 million cells. The average intra-assay and interassay coefficients of variation were 3.2% and 11.3% for the 17β-estradiol assay; 4.5% and 10.2% for the testosterone assay; 3.5% and 12.0% for the aldosterone assay; and 4.8% and 14.2% for the cortisol assay.

Quantitative Real-Time Polymerase Chain Reaction

H295R cells were seeded in 24 well plates precoated with collagen at a density of 60 000 cells/well in 1 mL of complete medium (seeding density was adjusted based on the surface area ratios of the well). Cells were acclimated for 24 hours prior to exposure to the OPE mixture at 1/1000K, 1/100K, 1/60K, or 1/30K for 48 hours, in the presence or absence of 10 μM forskolin (47). Total RNA was extracted using the RNeasy Plus Mini Kit (QIAGEN, Mississauga, Ontario, Canada) following the manufacturer's protocol. A NanoDrop 2000 spectrophotometer (ThermoFisher, Waltham, MA, USA) was used to assess the concentration and purity of the extracted RNA.

All primers were obtained from QuantiTect Primer Assays (QIAGEN): HMG-CoA reductase (HMGCR, QT00004081), steroidogenic acute regulatory protein (STAR, QT00091959), 11-beta-hydroxylase (CYP11B1, QT00028714), aldosterone synthase (CYP11B2, QT00076181), and glyceraldehyde 3-phosphate dehydrogenase (GAPDH, QT00079247). Total RNA was diluted to a working concentration of 2 ng/μL in RNase-free water. For transcripts that had moderate to low expression levels, the concentration was adjusted to 10 ng/μL (CYP11B1 and CYP11B2). The Power SYBR Green RNA-to-CT 1-Step Kit (Applied Biosystems, Foster City, CA, USA) and the Viia7 real-time polymerase chain reaction (PCR) System (Applied Biosystems) were used to quantify transcript levels. Each reaction mix (20 μL) was composed of 0.16 μL of reverse transcriptase, 2 μL of primer, 2.84 μL of RNase-free water, 10 μL of SYBR Green Master Mix, and 5 μL of RNA samples. The following condition was used for PCR experiments: 48 °C for 30 minutes, 95 °C for 10 minutes, 40 cycles at 95 °C for 15 seconds, 55 °C for 30 seconds, and 72 °C for 30 seconds. Each reaction was done in triplicate; outliers (>0.2 C_T values deviating from the average of the other 2) were checked and excluded. The expression levels of transcripts were calculated based on the average of replicates. Averaged expression levels were normalized to the amount of the GAPDH transcripts. The QuantStudio Real Time PCR Software (version 1.3) was used for data analysis.

Statistical Analyses

Data were analyzed using GraphPad Prism (version 9.4.1, GraphPad Software Inc., La Jolla, CA, USA). For high-content imaging data, Holm–Bonferroni–corrected 1-sample t test was used. For hormone production levels, 2-way analysis of variance was used followed by the Dunnett's test. High-content screening experiments were repeated independently 10 times (n = 10). For hormone measurements and lipidomic analyses, experiments were repeated independently 5 times (n = 5). The minimal level of significance was P < .05. For qRT-PCR data, P < .01 was considered statistically significant.

The peak-paired data from lipidomic analyses were uploaded to MetaboAnalyst 5.0 (https://www.metaboanalyst.ca) for multivariate statistical analysis. The parameters used are shown elsewhere (Table S3 (45)).

Results

Effects of the OPE Mixture on the Phenotypic Characteristics of H295R Cells

Cytotoxicity of the OPE mixture was assessed using Calcein-AM, a nonfluorescent cell-permeant dye that is converted to green-fluorescent Calcein in live cells. The numbers of Calcein positive, Hoechst-stained nuclei were quantified and compared with control (Fig. 2A). OPE mixture dilutions ranging from 1/1000K to 1/30K did not induce significant cell death. More than 30% cell death was observed at 1/10K or 1/3K; thus, these dilutions were excluded from subsequent phenotypic assessments.

Figure 2.

Effects of the organophosphate ester (OPE) mixture on (A) cell viability and (B) total area of lipid droplets. Cells were exposed to the OPE mixture for 48 hours, followed by staining with (A) Hoechst 33 342 (blue, nuclei) and Calcein-AM (green, viable cells) or (B) Hoechst 33342 (blue, nuclei) and Nile Red (yellow, lipid droplets) fluorescent dyes and were visualized with the Operetta high content imaging system (40× magnification). Data are shown as percentages relative to controls; values represent means ± SEM; n = 10. **P < .01 and ***P < .001 compared with control. Concentrations that induced >30% cytotoxicity were excluded from the analyses and were not shown for the lipid droplets endpoint.

Nile Red is a dye that fluoresces in lipid-rich environments. Nile Red–stained lipid droplets, shown in the representative figures as yellow dots, were clustered in the cytoplasm of cells (Fig. 2B). The total areas of lipid droplets were increased by OPE mixture exposures in a concentration dependent manner. At 1/300K, the level of lipid droplets was increased by 1.5-fold; it was further upregulated to 2.5-fold at 1/30K, the highest noncytotoxic dilution.

No significant alterations were observed in other phenotypic endpoints, including the levels of reactive oxygen species, lysosomal function, numbers of lysosomes, and the numbers of total and active mitochondria (Fig. S1 (45)).

Effects of the OPE Mixture on H295R Cell Lipid Droplets

Lipidomic analysis was done to assess the specific lipid species affected by the OPE mixture. Using the 3-tier lipid identification method, based on the lipid MS/MS match scores and the precursor m/z errors, a total of 599 lipids were identified with high confidence in tier 1, 68 lipids were identified with high confidence in tier 2, and 2900 lipids were putatively identified in tier 3.

The degree of separation in the lipidome profiles between the control and the 3 OPE mixture treatment groups is visualized in Fig. 3A in a partial least-squares discriminant analysis (PLS-DA) 2D score plot. All treatment groups were fully separated from the control, indicating that exposure to the OPE mixture had a profound impact on lipid composition. There was a slight overlay between treatment groups, suggesting that there were some commonalities in the affected lipid profiles. Indeed, the Venn diagram showed that 6 lipids were commonly affected by all 3 treatment groups; there were 7 to 16 affected lipids shared between any 2 treatment groups (Table S4; Fig S2 (45)). The top 15 lipids that drive the separation observed in the PLS-DA plot are identified in the variable importance in projection (VIP) score graph (Fig. 3B). Notably, lipids from the sphingolipids group displayed the highest VIP scores, accounting for one-third of the top 15 VIP lipids. Additionally, other lipids driving the OPE mixture induced shifts in lipid compositions were from the glycerophospholipids, glycerolipids, sterol lipids (STs), and fatty acyl groups.

Figure 3.

Multivariate analysis showing the degree of separation in lipid profiles between control, 1/300K, 1/100K, and 1/60K dilutions of the OPE mixture. (A) Differences in effects at mixture dilutions are visualized in a supervised PLS-DA plot. (B) PLS-DA Variable importance in projection (VIP) scores of the 15 most important lipids. Lipid category identification is indicated by the box on the left; each color represents a different lipid category.

Exposure to the 1/300K dilution OPE mixture significantly altered the levels of 62 lipids (Fig. 4A); among these, 20 lipids were downregulated and 42 lipids were upregulated. Similarly, exposure to the 1/100K dilution OPE mixture resulted in alterations in the same number of lipids, with an equivalent distribution of significantly upregulated and downregulated lipids. The highest noncytotoxic OPE dilution tested (1/60K) affected a total of 70 lipids; unlike the 1/300K dilution, the majority of these affected lipids were downregulated (50 lipids). The top 10 affected lipids in all treatment groups, ranked by fold change (5 in the downregulated and 5 in the upregulated category), along with their respective lipid subclasses, are labeled in the volcano plots presented in Fig. 4B-4D. Of note, half of the most affected lipids belong to the glycerolipids category. The remaining most affected lipids originated from various categories, including fatty acyls, sphingolipids, STs, glycerophospholipids, and polyketides. Moreover, 56.7% of the significantly affected lipids were oxidized, suggesting a possible correlation between exposure to the OPE mixture and lipid oxidation. A list of the lipids displaying extra oxygen atom(s) in their structures is provided elsewhere (Table S5 (45)).

Figure 4.

(A) Numbers of significantly affected lipids for each mixture dilution. (B) Volcano plot showing significantly altered lipids induced by exposure to a 1/300K dilution of the OPE mixture. (C) Lipids altered after exposure to 1/100K. (D) Lipids altered by the 1/60K dilution of the OPE mixture. Mixture exposures were for 48 hours and are compared with control. Dashed gray lines indicate the cutoffs used (fold change >1.4 or <0.71 and P < .05). Based on these cutoffs, blue dots indicate downregulated lipids and orange dots indicate upregulated lipids. For each dilution, the top 5 affected lipids in both the upregulated and downregulated group were highlighted by their subclass and category.

A heatmap showing the top 50 affected lipids, along with their respective categories, provides an overview of the significantly affected lipids (Fig. 5). OPE exposures at different dilutions showed distinct patterns compared with the controls in terms of the specific lipids affected. For instance, at the 1/100K dilution of the OPE mixture, several lipids from the diglyceride (DG) group were significantly upregulated, whereas no significant alterations in other treatment groups were observed. This distinction is also apparent in the Venn diagram, in which lipids affected by a single treatment constitute more than half of the total affected lipids (Table S6; Fig. S2 (45)).

Figure 5.

Heatmap of the top 50 affected lipids. Blue indicates that lipids were downregulated; red indicates upregulation. Lipids were labeled individually with their subclasses; lipid category is indicated by the box colors.

We next analyzed the relative distribution of significantly affected lipids in each lipid category (Fig. 6A). Glycerolipids represented 24% of the affected lipids at the 1/300K dilution; this percentage increased to 43% at the highest concentration of the OPE mixture tested. While a relatively lower percentage of lipids from the ST group was affected at the 1/300K dilution, exposure to the 1/60K OPE mixture concentration doubled the number of affected lipids from this group (from 5% to 10%). The proportions of the affected lipids in the sphingolipids and glycerophospholipids groups were comparable, with approximately 20% of the affected lipids attributed to each of the treatment groups. The numbers of lipids affected in the sphingolipids group remained relatively stable across treatment groups, whereas exposure to the highest concentration of the OPE mixture affected fewer glycerophospholipids. Lipids from the fatty acyl group showed a notable response to the OPE mixture. At the lowest OPE dilution (1/300K), 19% of the affected lipids were from the fatty acyl group; only 4% from this group were affected with increasing concentrations. The last 2 groups of lipids constituted the smallest proportion of significantly affected lipids. Polyketides initially accounted for 10% of the affected lipids (1/300K OPE mixture) but this was reduced to only 1% at the 1/60K dilution. Prenol lipids represented an even smaller percentage and were affected (2%) only at the 1/300K dilution.

Figure 6.

Relative distribution of significantly altered lipids in each category (A) and subclass (B-F). Color coding represents lipids that are from the same category: glycerolipids (yellow); sterol lipids (red); sphingolipids (green); glycerophospholipids (purple); fatty acyls (orange); polyketides (blue); prenol lipids (teal).

A further analysis of the subclass distribution of the most altered lipids identified the lipid species that were most impacted. In the glycerolipids group, the most affected lipids were DGs and triglycerides (TGs) (Fig. 6B). Within the ST group, 2 subclasses of affected lipids were identified, the CEs and STs (Fig. 6C); the number of CE lipids affected was elevated after exposure to the 1/60 K OPE mixture. In the sphingolipid category, 8 subclasses of lipids were identified; a similar number of affected lipids were present across the treatment groups (Fig. 6D). More lipids from the fatty acyls group (fatty acids, nitrogenated fatty acids, and N-acyl ethanolamines) were affected at lower OPE concentrations (Fig. 6E); however, some lipids (N-acyl ethanolamines, N-acyl amines, and wax esters) were affected primarily at the 1/60K dilution. Among the glycerophospholipids, phosphatidylcholine (PC) accounted for an average of 70% of the affected lipids at both the 1/300K and 1/100K dilutions (Fig. 6F). A proposed disrupted lipid pathway with affected lipid species and their categories is provided in Fig. 7.

Figure 7.

Proposed lipid pathways disrupted by OPE mixture exposures. Colored boxes highlight lipids affected by the OPE mixture; lipids in white boxes were not affected in the current study. Color coding represents lipids from the same category. Abbreviations for lipid subclasses: ceramide (Cer); sphingomyelin (SM); ceramide 1-phosphates (CerP); hexosyl ceramides (HexCer); sulfoglycosphingolipids (SHexCer); galactosylceramide (GalCer); ceramide phosphoinositols (MIPC); phosphatidic acids (PA); phosphatidylinositol (PI); phosphatidylglycerols (PG); glycerophosphoinositol monophosphates (PIP); phosphatidylcholine (PC); phosphatidylethanolamines (PE); phosphatidylserines (PS); lysophosphatidylserines (LPS); n-acyl ethanolamines (NAE); nitrogenated fatty acids (NA); fatty acids (FA); wax esters and diesters (WE); monoacylglycerols (MG); diglyceride (DG); triglycerides (TG); sterol lipids (ST); and cholesterols (Ce).

Effects of the OPE Mixture on H295R Cell Steroidogenesis

Since steroidogenesis is a major function of the adrenal, the effects of exposure to the OPE mixture on the ability of H295R cells to produce 17β-estradiol and testosterone (predominantly gonadal hormones) and aldosterone and cortisol (key adrenal hormones) under basal and stimulated conditions were analyzed (47).

The average basal production of 17β-estradiol was 0.5 ng/10⁶ cells (Fig. 8A). In the presence of stimulation with forskolin, the level of 17β-estradiol produced was increased approximately 7-fold. Exposure to the OPE mixture did not affect basal or stimulated 17β-estradiol production. The production of testosterone was not affected by OPE exposure under basal conditions (Fig. 8B). Forskolin stimulation increased testosterone production from 6.3 ng/10⁶ cells to 9.0 ng/10⁶ cells; exposure to the 1/60K or 1/30K OPE mixture dilutions decreased forskolin-stimulated testosterone production by approximately 19%.

Figure 8.

Effects of the OPE mixture on the steroid-producing function of H295R cells. Cells were exposed for 48 hours. For the stimulated condition, a concentration of 10 μM forskolin was used. Numbers of cells were quantified by Hoechst 33342 staining and high content imaging (10× magnification). Bar graphs show basal (left Y axis, white bars) and stimulated (right Y axis, striped bars) production of (A) 17β-estradiol, (B) testosterone, (C) aldosterone, and (D) cortisol levels. *P < .05, **P < .01, and ***P < .001 compared with control; values represent means ± SEM; n = 5.

The baseline production of aldosterone was 130 pg/10⁶ cells; forskolin stimulation induced a 3-fold increase in aldosterone production (Fig. 8C). Exposure to the OPE mixture strongly upregulated both basal and stimulated production of aldosterone by approximately 2-fold. OPE exposure also increased H295R cell production of both basal and stimulated cortisol (Fig. 8D). The average basal level of cortisol produced was 2.2 pg/10⁶ cells; production was increased by 1.6 times after exposure to 1/100K of the OPE mixture. In forskolin-stimulated cells, cortisol production was increased to an average level of 14.1 pg/10⁶ cells; an approximately 2-fold increase was observed after exposure to OPEs at the 1/1000K, 1/100K, and 1/60K dilutions.

Effects of the OPE Mixture on the Expression of Transcripts Involved in Cholesterol and Steroid Biosynthesis

We further assessed whether exposure to the OPE mixture affects the steroidogenic ability of H295R cells at the transcriptional level. First, we measured the mRNA level of HMGCR, the enzyme responsible for the rate-limiting step in cholesterol biosynthesis (Fig. 9A). Under basal conditions, the OPE mixture at 1/1000K, 1/100K, or 1/60K dilutions induced a slight increase in the level of HMGCR expression. Treatment with the OPE mixture under stimulated conditions induced a 11% increase only at the 1/1000K dilution. The expression level of STAR, a cholesterol transporter, was downregulated at 1/30K of the OPE mixture at both the basal and stimulated levels (Fig. 9B). The expression levels of 2 key transcripts responsible for the last steps in the biosynthesis of cortisol (CYP11B1) and aldosterone (CYP11B2) were assessed. A 2-fold increase in the level of CYP11B1 was observed at the 1/100K dilution; it reached 4.5-fold with exposure to 1/60K of the OPE mixture (Fig. 9C). With forskolin stimulation, an approximately 22% increase was observed with exposure to 1/1000K or 1/100K of the OPE mixture. At 1/60K, the expression level of CYP11B1 was upregulated by 46% compared with control. Similarly, exposure to the OPE mixture upregulated the level of CYP11B2 (Fig. 9D). Under basal conditions, the OPE mixture induced an up to 3-fold increase; a nearly 2-fold increase was observed under stimulated conditions.

Figure 9.

Effects of the OPE mixture on the mRNA expression of key transcripts involved in cholesterol and steroid biosynthesis: (A) HMGCR; (B) STAR; (C) CYP11B1; (D) CYP11B2 in H295R cells under basal and stimulated conditions. Cells were exposed for 48 hours. For the stimulated condition, a concentration of 10 μM forskolin was used. Data represent means and 95% CI, n = 5. **P < .001, ***P < .0001 compared with control.

Discussion

This is the first demonstration that a household dust–based mixture of OPEs affects adrenal cells. Exposure to this OPE mixture, at dilutions as low as 1/300K, induced an accumulation of lipid droplets in adrenal cells and altered the secretion of both basal and stimulated adrenal steroids. A closer look at the lipid species revealed that this OPE mixture dysregulates both the quantity of lipid droplets and their lipid composition.

Previous studies have provided evidence that exposure to individual OPEs may affect the adrenal. In animal studies, neonatal exposure to tris(1,3-dichloro-2-propyl) phosphate (TDCIPP) significantly increased the weight of the adrenal gland in adult male rats (48). Exposure to TMPP or IPPP has been associated with adrenal cortex vacuolization (49-54). Tris(2-butoxyethyl) phosphate (TBOEP), the major component of the OPE mixture, was shown to affect the transcription profiles of genes in the hypothalamus–pituitary–adrenal axis in zebrafish (55). In adrenal cells, exposure to tri-n-butyl phosphate or TMPP decreased cell viability (56). Moreover, exposure to tris(2-chloroethyl) phosphate, tris(1-chloro-2-propyl) phosphate, TDCIPP, TBOEP, TPHP, or TMPP altered the ability of adrenal cells to synthesize steroid hormones by dysregulating the transcription of genes in the steroidogenic pathway (57).

Our high content screening strategy enabled the identification of lipid droplet accumulation as the most prominent phenotypic target affected by a house dust–based OPE mixture (Fig. 2). Interestingly, increases in lipid droplets after OPE exposures have been observed in various cell types, suggesting that this may be a common mechanism of action of OPEs in different organs (31, 32, 58-60). Previously, we reported that exposure to 6 of the individual OPEs present in this mixture induced an increase in the lipid droplet areas in H295R cells. However, in contrast to the effects of individual OPEs, the OPE mixture did not affect oxidative stress, mitochondria, or lysosomes. An accumulation of lipids was also reported in the rat adrenal gland after exposure to TMPP, 1 of the components in the OPE mixture (51). The lipidomics data revealed that the major lipid categories contributing to the increased area of lipid droplets induced by the OPE mixture were glycerolipids and sphingolipids (Fig 5). The STs were affected to a lesser extent. An upregulation in the level of ST was identified in the lipidomics analysis; this was most marked at the 1/300K dilution (Fig. 5). Indeed, exposure to the OPE mixture increased the expression of HMGCR, the rate-limiting enzyme in cholesterol biosynthesis (Fig. 9A). An increase in HMGCR enzyme activity may contribute to the production of cholesterol.

Previous studies have shown that lipid droplets are composed of a hydrophobic core of neutral lipids surrounded by a phospholipid membrane monolayer (42). The neutral core contains mostly TGs, from the glycerolipids category, and CEs, from the STs category. The phospholipid membrane of lipid droplets is enriched with PC, from the glycerophospholipids category (61). Using LC-MS methodology, we identified 3570 lipid species with high sensitivity. Traditional lipidomic analysis typically narrows down to focus on a limited selection of lipid classes in order to simplify the complex lipidome in biological samples. By not limiting ourselves to a specific lipid category, we have shown that our house dust–based OPE mixture significantly altered a total of 142 lipids that belong to 7 lipid categories (Fig. S2 (45); Fig. 6). To the best of our knowledge, comprehensive nontargeted lipidomics analysis has been described previously for only 1 of the steroidogenic tissues, the MA-10 mouse Leydig cells (62, 63, 64). At the functional resting state of these Leydig cells, the main types of neutral lipids were TGs (97.1%), followed by 2.0% CEs (62). In the current study, TGs and DGs, were also 2 of the most prominently affected lipid classes (Fig. 6). Effects of OPEs on these 2 lipids have been characterized previously in liver cells (65, 66). Ethylhexyl diphenyl phosphate (EHDPHP), 1 of the components in our OPE mixture, significantly increased TG levels and the ratio between TGs and DGs produced by HepG2 spheroids (65). Increased intracellular levels of TGs have also been reported in HepG2 cells after exposure to other components of our OPE mixture; these include tris(2-chloroethyl) phosphate, tris(1-chloro-2-propyl) phosphate, and TDCIPP (66). In animal models, exposure to TPHP has been associated with serum hypertriglyceridemia as a consequence of inhibition of carboxylesterase 1G, a gene coding for the enzyme responsible for fatty acyl metabolism (67).

Exposure to all 3 dilutions of the OPE mixture significantly downregulated the level of CEs in adrenal cells; at increasing concentrations, more lipids from the CE subclass were affected (Fig. 4; Fig. 6). CEs play important roles in steroid-secreting cells as they are the stored form of cholesterol, the precursor for all steroid hormones (68). A downregulation of the levels of CE was observed in HepG2 cells after exposure to 10 μM EHDPHP (65); these investigators also observed a decrease in the level of total cholesterol when they analyzed the total lipidome. In contrast, exposure to TMPP was reported to upregulate the level of CEs in the adrenal glands of F344 rats; this effect was attributed to disruption in the process by which the CEs stored in lipid droplets are converted to free cholesterol (51).

The subclass of lipids most affected by the OPE mixture in the glycerophospholipids category was PC, the main component of lipid droplet membranes (Fig. 6). PC was upregulated by the OPE mixture, especially at the 1/300K dilution. A decreased level of PC was reported in murine macrophage cells after exposure to TPHP; this was accompanied by downregulation of the expression of lysophosphatidylcholine acyltransferase 3 (69). Exposure to the OPE mixture also increased ceramide and hexosyl ceramides (Fig. 4). Interestingly, a previous study reported that the levels of these 2 lipids were downregulated in liver cells after EHDPHP exposure (65), suggesting that there are chemical and cell line specific effects on this category of lipids.

Prenol lipids and polyketides were also affected by the OPE mixtures, although to a lesser extent (Fig. 6). Prenol lipids, more specifically the isoprenoids that were affected by the OPE mixture, serve as precursors and intermediates in the synthesis of STs (70). Several OPEs, including IPPP, TPHP, TMPP, TBOEP, and TDCIPP, have been reported to affect isoprenoid biosynthesis by targeting the rate limiting enzyme HMG-CoA reductase (HMGCR) and the master regulator of the pathway, sterol regulatory element binding protein (SREBP) (38, 66, 71, 72). The functional importance and the effects of OPEs on polyketides remains largely unknown; however, some polyketides appear to serve primarily as means of chemical defense for cells (73).

Few studies have assessed the effect of mixtures on the lipidome. In Wistar rats, exposure to Firemaster 550 affected ceramides, sphingomyelins, and TGs in the neonatal cortex (35). In preadipocytes and murine 3T3-L1 preadipocytes, Firemaster 550 targeted the peroxisome proliferator activated receptor and liver X receptor (31, 32). These 2 receptors are important regulators of TGs, fatty acids, and CE homeostasis (74-77); disruptions in their functions may explain the changes in the levels of lipid species we have observed.

It is perhaps not surprising that the effects of our house dust OPE mixture on the adrenal cell lipidome are accompanied by effects on steroidogenesis. This OPE mixture significantly decreased testosterone production but increased the levels of aldosterone and cortisol (Fig. 8). Luo et al (78) reported that the decreased concentrations of testosterone in urine samples collected from 6- to 19-year-old children and adolescents were associated with increased levels of dibutyl phosphate, a metabolite of tri-n-butyl phosphate, and of dibutyl phosphate, a metabolite of IPPP and TPHP. Increased urinary levels of dibutyl phosphate, bis(1-chloro-2 propyl) phosphate (a metabolite of TDCIPP), and bis(2-butoxylethyl) phosphate (a metabolite of TBOEP) were also associated with an increase in serum cortisol levels (79). Limited human data are available on the effects of OPE exposures on aldosterone levels. However, Zhang et al (80) reported an increase in aldosterone production in H295R cells exposed to 5 μM TPHP.

The mechanism(s) behind the observed increase in the area of lipid droplets and disruptions in the hormone production levels were examined by assessing the effect of the OPE mixture on the expression level of key transcripts involved in cholesterol and steroid biosynthesis (Fig. 9). In line with the increased level of ST reported in the lipidomics analysis, the level of HMGCR was upregulated by the OPE mixture (Fig 9A). Similarly, exposure to the OPE mixture strongly upregulated the levels of CYP11B1 and CYP11B2 (Fig. 9C and 9D). This effect could contribute to the increased levels of cortisol and aldosterone observed. In contrast, the level of STAR was downregulated by the OPE mixture at the highest dilution tested (Fig 9B). It was reported that in STAR knockout mouse MA-10 Leydig cells, the composition of lipid droplets was altered (64). Thus, it is possible that changes in STAR expression contribute to the observed disruptions in lipid compositions. Previously, we investigated the effects of 3 OPEs present in the OPE mixture, IPPP, TMPP, and TPHP, and reported similar results of these chemicals on the levels of HMGCR, CYP11B1, and CYP11B2 (38). However, exposure to the mixture downregulated STAR expression, whereas it was upregulated by IPPP and downregulated by TMPP or TPHP exposures. Thus, assessing the effects of environmentally relevant mixtures is imperative, as relying on evaluating individual chemicals provides only a partial representation of effects.

Our data provide compelling evidence that the adrenal gland is an important endpoint to consider in assessing the effects of OPEs. Furthermore, it is important to assess the impact of environmentally relevant mixtures. Together, our studies demonstrate that exposure to a house dust based OPE mixture disrupts human adrenal cell lipid homeostasis and steroid production.

Abbreviations

CE

cholesterol ester

DG

diglyceride

DMSO

dimethyl sulfoxide

EHDPHP

ethylhexyl diphenyl phosphate

IPPP

isopropylated triphenolphosphate

LC-MS

liquid chromatography mass spectrometry

OPE

organophosphate ester

PC

phosphatidylcholine

PLS-DA

partial least-squares discriminant analysis

qRT-PCR

quantitative real-time polymerase chain reaction

ST

sterol lipid

TBOEP

tris(2-butoxyethyl) phosphate

TDCIPP

tris(1,3-dichloro-2-propyl) phosphate

TG

triglyceride

TMPP

tris(methylphenyl) phosphate

TPHP

triphenyl phosphate

VIP

variable importance in projection

Funding

Canadian Institutes of Health Research (CIHR) Institute of Population and Public Health team Grant (FRN IP3-150711), Canadian Institutes of Health Research (CIHR) Project Grant FRN 156239, and McGill University. Z.L. is the recipient of training awards from McGill University and the Centre for Research in Reproduction and Development (CRRD). B.F.H. and B.R. are James McGill professors.

Author Contributions

Z.L., B.F.H., and B.R. were responsible for the experimental design, data interpretation, and manuscript preparation. Z.L. was responsible for data acquisition and analyses. All authors approved the final version of the article.

Disclosures

The authors have nothing to disclose.

Data Availability

All data are available upon request.