2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) Alters Hepatic Polyunsaturated Fatty Acid Metabolism and Eicosanoid Biosynthesis in Female Sprague-Dawley Rats

Abstract

2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) is a potent aryl hydrocarbon receptor (AhR) agonist that elicits a broad spectrum of dose-dependent hepatic effects including lipid accumulation, inflammation, and fibrosis. To determine the role of inflammatory lipid mediators in TCDD-mediated hepatotoxicity, eicosanoid metabolism was investigated. Female Sprague-Dawley (SD) rats were orally gavaged with sesame oil vehicle or 0.01–10 μg/kg TCDD every 4 days for 28 days. Hepatic RNA-Seq data was integrated with untargeted metabolomics of liver, serum, and urine, revealing dose-dependent changes in linoleic acid (LA) and arachidonic acid (AA) metabolism. TCDD also elicited dose-dependent differential gene expression associated with the cyclooxygenase, lipoxygenase, and cytochrome P450 epoxidation/hydroxylation pathways with corresponding changes in ω-6 (e.g. AA and LA) and ω-3 polyunsaturated fatty acids (PUFAs), as well as associated eicosanoid metabolites. Overall, TCDD increased the ratio of ω-6 to ω-3 PUFAs. Phospholipase A2 (Pla2g12a) was induced consistent with increased AA metabolism, while AA utilization by induced lipoxygenases Alox5 and Alox15 increased leukotrienes (LTs). More specifically, TCDD increased pro-inflammatory eicosanoids including leukotriene LTB₄, and LTB₃, known to recruit neutrophils to damaged tissue. Dose-response modeling suggests the cytochrome P450 hydroxylase/epoxygenase and lipoxygenase pathways are more sensitive to TCDD than the cyclooxygenase pathway. Hepatic AhR ChIP-Seq analysis found little enrichment within the regulatory regions of differentially expressed genes (DEGs) involved in eicosanoid biosynthesis, suggesting TCDD-elicited dysregulation of eicosanoid metabolism is a downstream effect of AhR activation. Overall, these results suggest alterations in eicosanoid metabolism may play a key role in TCDD-elicited hepatotoxicity associated with the progression of steatosis to steatohepatitis.

1. INTRODUCTION

2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) is a persistent environmental contaminant that elicits a broad spectrum of biochemical and toxic effects. In rodent models, TCDD induces dose-dependent hepatic lipid accumulation (steatosis), inflammation (steatohepatitis), and fibrosis (Boverhof et al.,2005; Kopec et al., 2010; Angrish et al., 2013; Pierre et al., 2014), which resemble features of Non-Alcoholic Fatty Liver Disease (NAFLD) in humans (Smith and Adams, 2011; Collier, 2006). The current model of NAFLD progression proposes multiple “hits” involving interactions between environmental factors, host genetics, and the gut microflora (Wree et al., 2013). Although hepatic steatosis is reversible and typically benign, the burden of accumulating damage from multiple “hits” drives the subsequent progression to steatohepatitis and fibrosis (Smith and Adams, 2011; Collier, 2006).

TCDD is the prototypical ligand and most potent agonist of the aryl hydrocarbon receptor (AhR) (Denison et al., 2011). Upon binding, the cytoplasmic AhR dissociates from aryl hydrocarbon receptor-interacting protein (AIP), p23, and heat-shock protein 90 (Hsp90) chaperone proteins and translocates to the nucleus where it heterodimerizes with the AhR nuclear translocator (ARNT). The AhR/ARNT transcription factor complex causes differential expression of target genes upon binding to dioxin response elements (DRE), although AhR-mediated gene expression changes independent of DREs have also been reported (Beischlag et al., 2008; Dere et al., 2011a; Huang and Elferink, 2012; Watson et al., 2014). Despite the conserved structure and signaling pathway of the AhR, TCDD-elicited differential gene expression is not conserved between species, suggesting underlying mechanistic differences are responsible for the species-specific effects of TCDD (Fader and Zacharewski, 2017).

Following a single dose of TCDD, Sprague-Dawley (SD) rats exhibit centriacinar hepatocellular hypertrophy, where the cytoplasm of the enlarged hepatocytes is more granular and eosinophilic compared to controls (Boverhof et al., 2006; Fletcher et al., 2005). Chronic 2-year National Toxicology Program (NTP) studies report dose-dependent increases in hepatocyte hypertrophy, multinucleated hepatocytes, eosinophilic matrix, inflammation, diffuse fatty change, portal fibrosis, and nodular hyperplasia in female SD rats gavaged with corn oil or 3, 10, 22, 46, or 100 ng/kg TCDD 5 days a week for 105 weeks (NTP, 1994). Similar findings were reported in an independent 2-year TCDD study in which female SD rats were given chow supplying 0.001, 0.01, or 0.1 μg TCDD/kg daily (Kociba et al., 1978; Walker et al., 2006). Despite differences in the incidence of cholangiocarcinoma, both studies reported the dose-dependent induction of features indicative of hepatotoxicity (Walker et al., 2006).

Polyunsaturated fatty acids (PUFAs) are major components of phospholipids (e.g. phosphatidylethanolamine, phosphatidylcholine, and phosphatidylinositide) that make up cellular membranes and serve other important cellular functions (James et al., 2000). Depending on their structure, PUFAs also exhibit inflammatory activities that may impact NAFLD progression. ω-6 PUFAs, such as linoleic acid (LA) and arachidonic acid (AA), are precursors to potent pro-inflammatory eicosanoids, while ω-3 PUFAs, including α-linolenic acid (ALA), eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA), have anti-inflammatory properties (James et al., 2000). For example, decreased lipogenesis, increased fatty acid oxidation, and ultimately reduced hepatic lipid accumulation has been attributed to ω-3 PUFAs (Siriwardhana et al., 2012). Lipidomic analyses consistently report higher serum levels of ω-6 compared to ω-3 PUFAs in NAFLD and NASH patients that correlate with disease severity (Wree et al., 2013).

ω-6 PUFAs serve as the primary source of eicosanoids. They represent the major metabolically active lipids, exerting effects through receptor binding and intracellular signaling cascades (Needleman et al., 1986). Eicosanoids are involved in several homeostatic processes including maintaining vasoconstriction/vasodilation, thrombotic/antithrombotic, and inflammation/anti-inflammation balance (Needleman et al., 1986). Their biosynthesis requires phospholipase A2 (PLA2) activation, which releases AA from membrane phospholipids. AA can then be metabolized by: (i) cyclooxygenases, (ii) cytochrome P450 monooxygenases, and (iii) lipoxygenases (Needleman et al., 1986). In the liver, Kupffer cells are the primary source of eicosanoids (Brouwer et al., 1995; Kolios et al., 2006). In response to injury, viral invasion, or insult of any etiology, the liver initiates a localized inflammatory response to minimize the damage and protect healthy tissue from further injury. While initial inflammatory responses act to resolve damage, excessive or chronic inflammation can exacerbate injury, that can cause scar formation, and fibrosis/cirrhosis that ultimately compromises function (Dennis and Norris, 2015). Although the temporal pattern and specific cellular events responsible for inflammation differ between acute and chronic liver injury, both share common inflammatory pathways and mediators.

In this report, we investigated PUFA metabolism and eicosanoid biosynthesis in female SD rats following treatment with TCDD. Although the cytochrome P450 hydroxylase/epoxygenase and the lipoxygenase pathways were most sensitive to TCDD, ChIP-Seq analysis suggests they are downstream effects rather than direct AhR targets.

2. METHODS

2.1. Animal Treatment

Female SD rats with body weights (BW) within 10 % of the average BW were received on post-natal day (PND) 25 (Charles River Laboratories). Animals were housed in Innovive Innocages® pre-bedded with corn cob in a 23°C HEPA-filtered environment with 30–40% humidity and a 12 h light/dark cycle. Animals were allowed free access to Aquavive® water and Harlan Teklad 22/5 Rodent Diet 8640 (Madison, WI). On PND29, rats (n=8–12 per treatment group) were orally gavaged with 0.1 mL sesame oil vehicle or 0.01, 0.03, 0.1, 0.3, 1, 3, 10 μg/kg TCDD (Dow Chemical Company, Midland, MI) every 4 d for 28 d. The selected doses compensate for the short duration of the study relative to the potential lifelong cumulative human burden from diverse AhR ligands, the bioaccumulative characteristics of halogenated AhR ligands (while acknowledging potential non-additive interactions), and the half-life differences of TCDD in humans and rodents (7.2–8.7 years in humans (Dilberto et al., 2001) vs. 8–12 days in mice (Kopec et al., 2013) and 20 days in female SD rats (Geyer et al., 2002; Li et al., 1995)). Only 10 μg/kg TCDD was used, rather than 30 μg/kg as in past C57BL/6 mouse studies, due to the greater sensitivity of SD rats to TCDD toxicity (Bickel, 1982; Vos et al., 1974). Doses of 0.01 to 10 μg/kg TCDD also captures the full range of responses facilitating dose-response modeling. Animals and food intake were weighed daily. Urine was collected on day 26 and stored at −80°C. Blood was collected by submandibular vein puncture into microtainer tubes with serum separator, centrifuged, and stored at −80°C. Animals were sacrificed by cervical dislocation and tissue samples were removed, weighed, flash-frozen in liquid nitrogen and stored at −80°C. Each right liver lobe was sectioned and fixed in 10% neutral buffered formalin (NBF, Sigma) for histological analysis. All procedures were approved by the Michigan State University (MSU) Institutional Animal Care and Use Committee (IACUC), in accordance with ethical guidelines and regulations.

2.2. Clinical Chemistry and Histology

Alanine aminotransferase (ALT) (n =4) was measured in serum from four independent animals using the Infinity ALT(GPT) liquid stable reagent according to the manufacturer’s protocol (ThermoFisher Scientific, Waltham, MA).

Formalin (10%) fixed hepatic tissues were sectioned and processed sequentially in ethanol, xylene, and paraffin using a Thermo Electron Excelsior (Waltham, MA). Tissues were embedded in paraffin using a Miles Tissue Tek II embedding center. Paraffin blocks were sectioned at 5 microns at the Michigan State University Histology Laboratory (https://humanpathology.natsci.msu.edu). Sections were placed on glass microscope slides, dried, and stained with hematoxylin and eosin.

2.3. Quantification of Hepatic TCDD Levels

Liver samples were processed in parallel with laboratory blanks and a reference or background sample at Wellington Laboratories, Inc. (Guelph, ON, Canada) following a USEPA standardized method (USEPA, 2007). Samples were weighed, spiked with ¹³C₁₂-2,3,7,8-TCDD, digested with sulfuric acid, and extracted. Extracts were cleaned, concentrated, and spiked with ¹³C₁₂-1,2,3,4-TCDD as an injection standard. Analysis was performed using high resolution gas chromatography (HRGC) coupled with high resolution mass spectrometry (HRMS). A Hewlett Packard 6890 GC interfaced to a Waters Autospec Ultima HRMS was utilized. The HRMS was operated in positive electron ionization/selective ion recording mode at 10,000 resolution. A 60-m DB₅ column (J&W Scientific, Folsom, CA) with an internal diameter of 0.25 mm and film thickness of 0.25 mm was employed. Injection volumes were 1 or 2 μL splitless injections with helium as the carrier gas and a constant flow of 1 mL min⁻¹.

2.4. Metabolite Extraction

Liver (25 mg) and serum (30 μL) metabolites were extracted from independent biological samples (n = 5) using chloroform/water/methanol from rats orally gavaged with sesame oil vehicle, 1, 3, and 10 μg/kg TCDD. Briefly, frozen liver sections were homogenized (Polytron PT2100, Kinematica) or vortexed (serum) in HPLC grade methanol/water (62/38) in Pyrex glass tubes. HPLC-grade chloroform was added following homogenization, vortexed, and shaken. The aqueous phase was collected and dried under nitrogen gas at room temperature and resuspended in HPLC-grade water. Hepatic extract and serum protein layers were dried. The bicinchoninic acid assay (Sigma-Aldrich) on a Tecan Infinite 200 microplate reader (Mannedorf, Switzerland) was used to normalize signal in liver and serum. Urine samples were centrifuged at 10,000 x g to remove any solid debris and diluted 1:2 with HPLC-grade water prior to analysis.

2.5. Untargeted Metabolomic Analysis

UPLC separation was coupled with negative-mode electrospray ionization (ESI) to a Xevo G2-XS Quadropole Time of Flight (QTOF) mass spectrometer (Waters) run in MS^E continuum mode to analyze liver extracts, serum, and urine samples. The LC parameters were as follows: autosampler temperature, 10°C; injection volume, 10 μL; and flow rate, 200 μL min⁻¹. The LC solvents were Solvent A: 10 mM tributylamine and 15 mM acetic acid in 97:3 water: methanol (pH 4.95); and Solvent B: methanol. Elution from the column (C18, 2.7 μm particle size, 5 cm × 2.1 mm, Supelco) was performed over 11 min gradient. Data analysis used Progenesis QI software (Waters) and compound ions were annotated with the human metabolome database (HMDB) and ChemSpider. Metabolites were normalized to total ion chromatogram.

2.6. Untargeted Eicosanoid and Docosanoid Analysis

Frozen liver tissue (50 mg) was extracted from independent biological replicates (n = 5) in 60% methanol containing ethylenediaminetetraacetic acid, butylhydroxytoluene, triphenylphosphine, and indomethacin (Mattmiller et al., 2014). Deuterated internal standards including d4-LTB4, d8-arachidonic acid, d5-Resolvin D2, d4-PGE2, and d4-TBXB2 (Cayman Chemicals) were added at final concentrations of 100 pg/μL. Extract supernatants underwent solid phase extraction using Phenomenex Strata-X 33 micron SPE Columns (200 mg/6 ml) (Mattmiller et al., 2014) to remove biological matrix components. Eluates were reconstituted in 200 μL of methanol containing 0.1% butylated hydroxytoluene. Fatty acids and their oxygenated derivatives were analyzed by high resolution/accurate mass (HRAM)-LC-MS using an Agilent 1260 HPLC coupled to a Thermo Scientific LTQ-Orbitrap Velos mass spectrometer. Extracts (5 μL) were injected into water/acetonitrile/acetic acid (98/2/0.02) at a flow rate of 3 μL per min and loaded onto an inline peptide Opti-Trap (Optimize Technologies) for 3 min for analyte concentration. Samples were then eluted onto a ProntoSil C18 AQ 200 mm × 50 mm, 3 μm column (nanoLCMS Solutions) using a gradient of Solvent A: acetontrile/water/acetic acid (45/55/0.02) and Solvent B: acetontrile/water/acetic acid (90/10/0.02) from 40% to 50% B over 15 minutes, followed by 90% B for 3 minutes, at a flow rate of 3 μL/min. The column eluent was introduced to the mass spectrometer by an Advance nano-ESI source (Michrom BioResources) at a spray voltage of −1.7 kV. High resolution (R = 100,000 at 400 m/z) negative ion mass spectra were collected over 200–900 m/z with top 3 data-dependent MS/MS at a resolution of 60,000 using the HCD cell at a normalized collision energy of 75. Chromatographic peak alignment, feature detection, and quantitation were performed using Elements for Metabolomics (Proteome Software) and SECD-LIMSA software (Haimi et al., 2009).

2.7. RNA isolation

Frozen liver samples (~100 mg) were homogenized in 1.3 mL TRIzol (Life Technologies, Carlsbad, CA) using a Mixer Mill 300 tissue homogenizer (Retsch, Germany). Total RNA was isolated according to manufacturer’s protocol with an additional phenol:chloroform extraction (Sigma-Aldrich, St. Louis, MO). Isolated RNA was resuspended in RNA storage solution (Life Technologies). Total RNA was quantified and assessed for purity by nanodrop (Thermo Scientific, Waltham, MA), Qubit (Life Technologies), Bioanalyzer (Agilent Technologies, Santa Clara, CA), A₂₆₀/A₂₈₀ ratio and by visual inspection on a denaturing gel.

2.8. Hepatic Gene Expression

Gene expression were examined using RNA-sequencing (Michigan State University Research Technology Support Facility (RTSF) Genomics Core, https://rtsf.natsci.msu.edu/genomics). Libraries from independent biological replicates (n = 3) were prepared using Illumina TruSeq RNA Sample Preparation Kit (Illumina, San Diego, CA). Libraries were quantified and sequenced (1 × 50 bp), using a read depth of ~30 M per sample (Nault et al., 2015). Read quality was assessed using FASTQC v0.11.2 (www.bioinformatics.babraham.ac.uk/projects/fastqc/), adaptor sequences removed using Cutadapt v.1.4.1 (Martin, 2011), and low-complexity reads cleaned using FASTX v0.0.14 (http://hannonlab.cshl.edu/fastx_toolkit/index.html). Reads were mapped to the rat reference genome (RGSC 5.0 release 74) using Bowtie 1.0.0 and TopHat v1.4.1 (Langmead et al., 2009). Hepatic RNA-Seq analysis of TCDD-treated female C57BL/6 mice was previously published (GSE62902) (Nault et al., 2015). RNA-Seq data sets for female SD rat liver were deposited in the Gene Expression Omnibus (GEO; accession number GSE110293).

Genes were considered differentially expressed if |fold-change| ≥ 1.5 and posterior probability (P1(t)) value ≥ 0.8 at one or more doses (Glaus et al., 2012). Estimates of benchmark dose (BMD) and BMD lower confidence limit (BMDL) were calculated using BMDExpress (Yang et al., 2007). A benchmark response factor of 1.349 was chosen to represent 10% change in transcript levels compared to vehicle controls. Normalized linear RNA-Seq fold-changes for each feature were fit to Hill, linear (1° polynomial), 2° polynomial, or power dose-response models and the best globally fitting model with the least complexity was selected following the criteria previously published by U.S. EPA (Bhat et al., 2013).

2.9. pDRE Distribution

Putative dioxin response elements (pDREs) in the rat genome (Rn5) were previously identified (Dere et al., 2011b). Matrix similarity scores calculated using an updated position weight matrix (Nault et al., 2015). pDREs UCSC genome browser tracks are available at http://dbzach.fst.msu.edu/index.php/supplementarydata.html.

2.10. Hepatic AhR ChIP-Seq

PND29 female SD rat liver samples collected at 2 h following a single oral gavage with 10 μg/kg TCDD. Cross-linked DNA was immunoprecipitated with either rabbit IgG or rabbit IgG and anti-AhR (Dere et al., 2011a; Lo and Matthews, 2012). Libraries prepared using the MicroPlex kit (Diagenode) were pooled and sequenced to a depth of ~30M on an Illumina HiSeq 2500 at the MSU RTSF. Read processing and analysis was preformed using the MSU High Performance Computing Center. Quality was determined using FASTQC v0.11.2 and adaptor sequences removed using Cutadaptv1.4.1 while low-complexity reads were cleaned using FastX v0.0.14. Reads were mapped to the rat (assembly v3.4) reference genome using Bowtie 2.0.0 and alignments were converted to SAM format using SAMTools v.0.1.19. Normalization and peak calling was performed using CisGenome (Ji et al., 2008) by comparing IgG control and AhR enriched samples (n = 5) using a bin size (−b) of 25 and boundary refinement resolution (−bw) of 1 with default parameters. A false discover rate (FDR) cutoff of 0.05 was used in evaluating AhR enrichment. ChIP-Seq data sets for female SD rat liver were deposited in the Gene Expression Omnibus (GEO; accession number GSE110282).

2.11. Functional enrichment analysis

Functional enrichment of differentially expressed orthologs in mouse and rat was assessed using the database for annotation, visualization, and integrated discovery (DAVID) v6.8 (Huang et al., 2009) using only KEGG pathways. Categories with enrichment scores ≥ 1.3 (equivalent to −log of the geometric mean P value of 0.05) were considered statistically significant.

2.12. Statistical Analysis

Statistical analysis was performed using GraphPad Prism. Specific tests for each analysis are noted in the figure captions. Differences between treatment groups were considered significant when p < 0.05.

2.13. Data Availability

Female SD rat liver RNA-Seq and ChIP-Seq data sets are available in the Gene Expression Omnibus (GEO; accession number GSE110293 and GSE110282, respectively).

3. RESULTS

3.1. Hepatic TCDD Levels and Effects on Body and Tissue Weights

Hepatic TCDD levels in female SD rats dose-dependently increased with treatment. The increases were comparable to levels reported in female C57BL/6 mice using the same treatment regimen, with slightly higher accumulations at doses below 10 μg/kg TCDD (Supplementary Figure S1) (Fader et al., 2015). TCDD is sequestered in the liver by CYP1A2, a classic AhR battery gene that enhances the persistence of its effects (Voorman and Aust, 1989; Diliberto et al., 1997; De Vito et al., 1998). Cyp1a2 was induced 1.7-fold higher in female SD rats compared to female C57BL/6 mice (Nault et al., 2015), potentially contributing to the greater hepatic TCDD accumulation and increased sensitivity of rats compared to mice.

Terminal body weights at 10 μg/kg TCDD were 12.9% lower than controls (Figure 1A). This was not due to decreased food consumption as differences in daily food intake was not observed until day 27, while body weight differences were significant by day 10 (Supplementary Figure S2). Relative liver weights were increased 1.4-, 1.6-, and 1.6-fold at 1, 3, and 10 μg/kg TCDD, respectively (Figure 1B), as previously reported (Boverhof et al., 2006), while relative gonadal white adipose tissue (gWAT) weights decreased 2.0-fold at 10 μg/kg TCDD (Figure 1C). Likewise, serum ALT levels were elevated 1.8-fold at 10 μg/kg TCDD (Figure 1D) (Fletcher et al., 2005; Harvey et al., 2016).

Figure 1. Effects of TCDD on body and tissue weights.

Female Sprague-Dawley (SD) rats were orally gavaged with sesame oil vehicle or 0.01–10 μg/kg TCDD every 4 days for 28 days. (A) Terminal body weights were recorded on day 28 prior to euthanasia. (B) Livers and (C) gonadal white adipose tissue (gWAT) were weighed and normalized to terminal body weight. (D) Alanine aminotransferase (ALT) activity in the serum was measured using the Infinity ALT(GPT) liquid stable reagent. Bars represent the average of ≥ 4 rats + standard error of the mean (SEM). Statistical significance (*p ≤ 0.05) was determined using a one-way ANOVA analysis followed by Dunnett’s post hoc test.

3.2. Histopathology

TCDD-treated female SD rats exhibited moderate hepatocellular hypertrophy. The cytoplasm of enlarged hepatocytes was more granular and eosinophilic with increased vacuolization compared to controls (Figure 2A–D). While observable at 1 and 3 μg/kg TCDD, these observations were most prominent at 10 μg/kg TCDD.

Figure 2. Hematoxylin and eosin (H&E)-stained liver histopathology following TCDD treatment.

Female Sprague-Dawley rats were orally gavaged with (A) sesame oil vehicle, (B) 1 μg/kg TCDD, (C) 3 μg/kg TCDD, or (D) 10μg/kg TCDD every 4 days for 28 days. Bars = 50 μm.

Past assessments of TCDD-treated female C57BL/6 mice using the same dosing regimen (every 4 days for 28 days) elicited dose-dependent increases in centrilobular hepatocyte vacuolation, inflammation, and fibrosis (Nault et al., 2016). Inflammation and fibrosis were only observed at 10 μg/kg and 30 μg/kg, respectively (Nault et al., 2016). In comparison, female SD rats showed less vacuolation, and negligible inflammation or fibrosis.

3.3. Cross-Species Comparison of TCDD-Elicited Differential Gene Expression

RNA-Seq detected a total of 14,450 unique hepatic genes expressed in SD rats, of which 4,062 were differentially expressed (|fold-change| ≥ 1.5, P1(t) ≥ 0.8), comparable to the 3,599 genes differentially expressed in the TCDD-treated female mouse liver (GSE62902) (Nault et al., 2015) (Figure 3). Differentially expressed rat genes included AhR battery members including Cyp1a1 (induced 680-fold), Cyp1b1 (induced 335-fold), and Cyp1a2 (induced 27.2-fold).

Figure 3. Cross-species comparison of TCDD-elicited differential gene expression in the liver.

RNA-Seq analysis detected 14,450 unique genes expressed in the female Sprague-Dawley rat liver (GSE110293) and 17,188 unique genes expressed in the female C57BL/6 mouse liver (GSE62902). HomoloGene IDs (www.ncbi.nlm.nih.gov/homologene) identified 6,650 orthologous genes expressed in both species, of which 1,862 rat and 1,445 mouse orthologs were differentially expressed (|fold-change ≥ 1.5| and P1(t) ≥ 0.8) in response to TCDD. DAVID (v6.8) analysis was used to identify functions associated with the 564 differentially expressed orthologs common to the two species. Categories with enrichment scores ≥ 1.3 (equivalent to −log of the geometric mean p value of 0.05) were considered significant.

Using HomoloGene (ncbi.nlm.nih.gov/HomoloGene/), 6,650 orthologous genes were identified in the livers of both species (Figure 3), of which 1,862 and 1,445 were differentially expressed by TCDD in the rat and mouse, respectively. Comparison of the differentially expressed orthologs identified 564 genes (|fold-change| ≥ 1.5, P1(t) ≥ 0.8) in common between the two species, while 1,298 (70%) and 881(61%) were specific to the rat and mouse, respectively (Figure 3). Although relaxation of the filtering criteria (|fold-change| ≥ 1.4, P1(t) ≥ 0.75) increased the overlap of differentially expressed orthologs, the results are consistent with TCDD-elicited species-specific gene expression profiles. Several orthologs in common between the two species exhibited divergent regulation (e.g., induced in rat, repressed in mouse).

Functional analysis of the 564 common DEGs using DAVID identified over-represented functions associated with cytochrome P450-mediated xenobiotic metabolism, chemical carcinogenesis, purine metabolism, and tryptophan metabolism (Figure 3). The 1,298 rat-specific responses were enriched for fatty acid metabolism and degradation, glycan biosynthesis, lipid metabolism, primary bile acid biosynthesis, and PPAR signaling pathways. Mouse-specific responses were associated with retinol metabolism, ascorbate and aldarate metabolism, porphyrin metabolism, and pentose/glucoronate interconversions.

3.4. TCDD-Elicited Effects on the Rat Metabolome

Untargeted metabolomics of female SD rat urine (Figure 4A), liver extract (Figure 4B), and serum (Figure 4C) detected 941 (487 increased; 454 decreased), 432 (112 increased; 320 decreased), and 183 (93 increased; 90 decreased) metabolite changes, respectively, between vehicle and TCDD-treated samples. MetaboAnalyst associated these changes with 18 different pathways including several involved in eicosanoid metabolism (Figure 4D) (Xia and Wishart, 2016). More specifically, genes and metabolites involved in the metabolism of arachidonic acid (AA) and linoleic acid (LA), two PUFAs associated with eicosanoid biosynthesis, were highly enriched. This, in addition to the known role of inflammation in NAFLD progression, prompted further investigation into TCDD-elicited dysregulation of PUFA and eicosanoid metabolism.

Figure 4. Untargeted metabolomics analysis of urine, liver, and serum.

Female Sprague-Dawley (SD) rats were orally gavaged with sesame oil vehicle or 0.01–10 μg/kg TCDD every 4 days for 28 days. An untargeted metabolomics approach was used to assess TCDD-elicited changes in the (A) urine, (B) liver, and (C) serum metabolome. Detected peaks were matched with potential metabolite identifications corresponding to specific retention times and mass/charge (m/z) ratios when compared to the human metabolome database (HMDB). Following manual curation and removal of redundancies, the number of altered metabolites was determined by applying a statistical criterion of p ≤ 0.05. (D) Hepatic RNA-Seq was integrated with the untargeted metabolomics analysis of urine using MetaboAnalyst, that identified 18 significantly enriched pathways. Several enriched pathways were associated with eicosanoid biosynthesis, including linoleic acid metabolism, arachidonic acid metabolism, and glutamate metabolism. Categories with enrichment scores ≥ 1.3 (equivalent to −log of the geometric mean p value of 0.05) were considered significant.

3.4.1. Omega-6 and Omega-3 PUFAs

TCDD treatment increased total pro-inflammatory ω-6 PUFAs (e.g. LA and AA) while decreasing anti-inflammatory ω-3 PUFAs (e.g. α-linolenic acid (ALA), eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA)), resulting in a 1.6-fold increase in the hepatic ω-6/ω-3 PUFA ratio (Table 1). ω-6 PUFAs are precursors to potent pro-inflammatory eicosanoids, while ω-3 PUFAs give rise to anti-inflammatory mediators. Dietary ω-3 PUFAs exert their beneficial effects by replacing ω-6 PUFAs at the sn-2 position of glycerophospholipids producing anti-inflammatory prostaglandins and leukotrienes (Smith, 1989). Eicosanoids derived from the ω-6 PUFA AA, including prostaglandin E2 and leukotriene B4 (Figure 5), are more potent mediators of thrombosis and inflammation compared to those derived from ω-3 PUFAs (i.e. prostaglandin E3 and leukotriene B5 from EPA) (Simopoulos, 2016). Lipidomic studies report NAFLD and NASH patients also have higher ω-6/ω-3 ratios in both blood and liver extracts (Wree et al., 2013).

Table 1.

The Effects of TCDD on Hepatic ω-3 and ω-6 Polyunsaturated Fatty Acids.

	TCDD (μg/kg)
	0	1	3	10
ω-3 PUFAs
α-Linolenic acid (ALA)	4.73a ± 1.84b	2.52a ± 0.08b	1.65a ± 0.31b	2.58a ± 0.34b
Eicosapentaenoic acid (EPA)	2.93a ± 1.09b*	0.81a ± 0.09b*	0.51a ± 0.16b*	0.63a ± 0.12b*
Eicosatrienoic acid (ETA)	0.21a ± 0.03b	0.28a ± 0.04b	0.19a ± 0.03b	0.43a ± 0.09b*
Docosahexanoic acid (DHA)	2.50a ± 0.72b	2.98a ± 0.41b	1.92a ± 0.46b	2.39a ± 0.33b
Sum of ω-3 PUFAs	10.37 ± 3.63	6.59 ± 0.50	4.27 ± 0.94	6.03 ± 0.80
ω-6 PUFAs
Linoleic acid (LA)	6.93a ± 2.13b	7.04a ± 1.34b	5.84a ± 1.60b	9.05a ± 1.39b
Arachidonic acid (AA)	4.99a ± 1.59b	5.09a ± 1.03b	3.00a ± 0.98b	4.05a ± 0.73b
Docosapentanoic acid (DPA)	0.34a ± 0.07b	0.47a ± 1.03b	0.36a ± 0.09b	0.54a ± 0.04b
Docosatetraenoic acid (DTA)	0.15a ± 0.02b	0.37a ± 0.05b	0.39a ± 0.15b	0.65a ± 0.06b*
Sum of ω-6 PUFAs	12.41 ± 3.71	12.97 ± 2.38	9.61 ± 2.43	14.30 ± 2.08
Ratio of ω-6 to ω-3 PUFAs	1.47 ± 0.31	1.97 ± 0.25	2.25 ± 0.31	2.37 ± 0.11*

^aNormalized abundance of PUFA detected by Mass Spectrometry.

^bStandard error of the mean (SEM) of N = 5.

^*p < 0.05 by a one-way ANOVA with Dunnett’s post hoc test

Figure 5. Arachidonic acid metabolism and eicosanoid biosynthesis.

(1) Eicosanoid biosynthesis is initiated upon activation of phospholipase A2 enzymes (PLA₂), releasing arachidonic acid (AA) from membrane phospholipids. AA is converted into biologically active eicosanoids via the (2) cyclooxygenase (COX), (3) cytochrome P450 monooxygenase (CYP450) and (4) 5-lipoxygenase (ALOX5) pathways. Hepatic gene expression data from RNA-Seq and liver metabolite levels from the untargeted metabolomics analysis were integrated to demonstrate dysregulation of leukotriene synthesis from PUFAs following exposure to 10 μg/kg TCDD.

AA is an abundant ω-6 PUFA that is esterified within cell membranes. AA is released from membranes by activated phospholipases such as Pla2g12a, which was induced 5.8-fold by TCDD (Figure 6A). AA then undergoes oxidative metabolism via cytochrome P450 epoxygenases/hydroxylases, cyclooxygenases, and lipoxygenases. Of the 63 hepatic genes involved in AA metabolism, 31 were differentially expressed by TCDD. Only 4 showed AhR enrichment despite the majority possessing putative dioxin response elements (pDREs) (Figure 6A–D). Total eicosanoids including prostanoids (i.e. prostaglandins, prostacyclins, and thromboxanes), leukotrienes (LTs), and resolvins were increased by TCDD (Figure 7 & 8).

Figure 6. TCDD-elicited differential expression of genes involved in hepatic eicosanoid biosynthesis.

(A) Hepatic phospholipases that initiate eicosanoid biosynthesis through the release of PUFAs from membrane phospholipids. Hepatic genes involved in the following eicosanoid biosynthesis pathways: (B) cytochrome P450 hydroxylation/epoxidation, (C) cyclooxygenase, and (D) lipoxygenase. 31 of 63 genes directly involved in eicosanoid biosynthesis exhibited differential expression (|fold change| ≥ 1.5 and P1(t) ≥ 0.8. Color scale represents the log₂(fold-change) for gene expression as determined by RNA-Seq analysis (n = 3). The presence of pDREs (matrix similarity score ≥ 0.856) and AhR enrichment peaks (FDR ≤ 0.05) at 2 h are shown as green boxes. Relative gene expression level indicating the number of aligned reads is also shown, where yellow represents lower expression (≤ 500 reads) and pink represents higher expression (≥10,000 reads). Benchmark dose (BMD) and the lower 95% confidence limit (BMDL) were determined using BMDExpress software.

Figure 7. Effects of TCDD on total hepatic eicosanoids.

Female Sprague-Dawley (SD) rats were orally gavaged with sesame oil vehicle or 1–10 μg/kg TCDD every 4 days for 28 days. Untargeted metabolomics of extracted livers identified increases in total hepatic: (A) eicosanoids, (B) leukotrienes, and (C) prostanoids following treatment with TCDD. Bars represent the average of 5 rats + standard error of the mean. Statistical significance (*p ≤ 0.05) was determined using a one-way ANOVA analysis followed by Dunnett’s post hoc test.

Figure 8. The effects of TCDD on polyunsaturated fatty acids (PUFAs) and eicosanoids.

Female Sprague-Dawley (SD) rats were orally gavaged with sesame oil vehicle or 1–10 μg/kg TCDD every 4 days for 28 days. Hepatic PUFA and eicosanoid levels were measured via untargeted metabolomics. Results are presented as the average fold-change (n = 5) with red indicating increased metabolites and blue indicating decreased metabolites. Point of departure (POD = BMD in μg/kg) was calculated using BMDExpress.

3.4.2. Cytochrome P450 Epoxidation/Hydroxylation Pathway

Multiple mammalian cytochrome P450 subfamilies metabolize AA to different eicosanoids with varying catalytic efficiencies (Zeldin, 2001). Allylic oxidation forms midchain conjugated dienols (5-,8-,9-,11-,12-, and 15-hydroxyeicosatetraenoic acid (HETEs)) (Zeldin, 2001), and ω-terminal hydroxylation produces C₁₆-C₂₀ alcohols (16-, 17-, 18-, 19-, and 20-HETEs) (Zeldin, 2001), while olefin epoxidation (i.e. epoxygenase reaction) results in cisepoxyeicosatrienoic acids (EETs) (14,15-, 11,12-, 8,9-, and 5,6-EETs) (Figure 5) (Zeldin, 2001). For example, CYP1A1, CYP1A2, and CYP1B1 catalyze AA hydroxylation/epoxidation to produce midchain HETEs, terminal HETEs, and EETs (Choudhary et al., 2004; Schwarz et al., 2004). Hepatic Cyp1a1 (680.4-fold), Cyp1a2 (27.2-fold), and Cyp1b1 (335.1-fold) were dose-dependently induced by TCDD with BMD values of 0.002, 0.007 and 0.027 μg/kg, respectively (Figure 6B). Although eicosanoid products from cytochrome P450 hydroxylation/epoxidation were not measured in this study, TCDD-elicited increases in dihydroeicosatrienoic acid metabolites (DHETs) and HETEs have previously been attributed to the induction of Cyp1a1, 1a2 and 1b1 (Bui et al., 2012). Moreover, previous TCDD studies demonstrate increased hepatic EET biosynthesis, likely through induction of cytochrome P450s (Schlezinger et al., 1998; Gilday et al., 1998).

Other cytochrome P450s with epoxidase activity required for EET formation were generally repressed including Cyp2c22 (1.8-fold), Cyp2c13 (14.3-fold), Cyp2c12 (1.8-fold), Cyp2b3 (2.0-fold), Cyp2c6v1 (2.8-fold), Cyp2b2 (3.4-fold), and Cyp2j4 (1.7-fold), while Cyp2c24 was dose-dependently induced 3.3-fold (Figure 6B). Microsomal epoxide hydrolase (Ephx1) was also induced 3.3-fold with AhR enrichment, while the soluble hydrolase (Ephx2) was negligibly repressed (1.3-fold). EPHX1 and EPHX2 metabolize EETs into DHETs, with EPHX2 being the preferred reaction (Figure 5). EETs have potent effects on peptide hormone secretion, vascular and smooth muscle tone, ionic transport, regulating cell proliferation, inflammation, homeostasis, and signaling pathways (Smith, 1989).

The cytochrome P450 4A and F subfamilies also produce HETEs. Cyp4a8 was induced 2.1-fold, whereas Cyp4a2 (2.0-fold), Cyp4a1 (2.3-fold), Cyp2j4 (1.7-fold), and Cyp2e1 (3.4-fold) were repressed by TCDD, all in the absence of AhR enrichment (Figure 6B).

3.4.3. Cyclooxygenase Pathway

The prostaglandin endoperoxide H synthases (PTGS1 and PTGS2) are cyclooxygenase isozymes which cyclize and oxygenate AA to form prostaglandin G2 (PGG₂). The hydroperoxyl group of PGG₂ is then reduced to form prostaglandin H2 (PGH₂), a highly unstable endoperoxide that is rapidly converted by specific synthases to prostaglandins (PG), prostacyclins, and thromboxanes (TX) (Figure 5) (Simmons et al., 2004). PTGS1 is constitutively expressed in most cells, while PTGS2 is induced by inflammatory stimuli, hormones, and growth factors, contributing to prostanoid formation during inflammation and proliferative diseases (i.e. cancer) (Ricciotti and FitzGerald, 2011). TCDD induced Ptgs2 2.1-fold, while Ptgs1 was repressed 1.4-fold (Figure 6C). Our dose-dependent induction of Ptgs2 in the absence of AhR enrichment is consistent with results in canine MDCK and Hepa1c1c7 cells (Abran et al., 1997).

Downstream isomerases and oxidoreductases metabolize PGH₂ to yield other bioactive prostaglandin species (i.e. PGE₂, PGF₂, PGD₂, PGI₂, or thromboxane A₂) (Figure 5). Ptgis, which catalyzes the conversion of PGH₂ to PGI₂ (prostacyclin), was repressed 5.3-fold in the absence of AhR enrichment (Figure 6c). Although Ptges expression was not changed, levels of PGE₂, a pro-inflammatory lipid mediator which acts as a vasodilator and fever-inducer, were increased 1.8-fold at 1 μg/kg TCDD (Figure 8) (Abran et al., 1997; Yashiro and Ohhashi et al., 1997; Tilley et al., 2001). At higher TCDD doses, several PGE₂ metabolites were increased including bicyclo PGE₂ (4.1-fold), a stable breakdown product of 13,14-dihydro-15-keto-PGE₂ (Granstrom et al., 1980; Hamberg and Samuelsson, 1971). Bicyclo PGE₂ is considered an indicator of PGE₂ biosynthesis and disposition in vivo due to its greater stability (PGE₂ t_1/2 = 2 min; 13,14-dihydro-15-keto PGE₂ t_1/2 = 9 min) (Granstrom et al., 1980; Hamberg and Samuelsson, 1971) (Figure 8). PGE₂ also undergoes dehydration and isomerization, producing PGA₂ and PGB₂ (Cattan et al., 2000). PGB₂, which was increased 2.6-fold, induces IL-2 production, IL-20052α expression, and nuclear translocation of NF-κB, consistent with pro-inflammatory activity (Cattan et al., 2000).

Thromboxane A₂ (TXA₂) is an unstable AA metabolite synthesized from PGH₂ by TBXAS1 (Figure 5). TXA₂ is rapidly and non-enzymatically degraded to the biologically inactive form, TXB₂. Although Tbxas1 was unchanged by TCDD, TXB₂ (1.5-fold) and its metabolites (TXB₁, 1.6-fold; 11-dehydro-TXB₂, 1.5-fold) were increased at 1 μg/kg TCDD (Figure 8).

Prostanoids exert their biological effects by activating a subfamily of seven transmembrane spanning G protein-coupled receptors (GPCRs) including: E prostanoid subtype receptors of PGE₂ receptor (EP1, EP2, EP3, EP4), PGD receptor (DPI), PGF receptor (FP), PGI receptor (IP), and TX receptor (TP) (Figure 5). Ptgir (1.7-fold; encodes IP), Ptger2/3/4 (1.2-, 1.8-, and 1.7-fold; encodes EP2–4), and Tbxa2r (2.1-fold; encodes TP) were induced at 10 μg/kg TCDD, while Ptgfr (1.6-fold; encodes the PGF2α receptor FP) was repressed (Figure 5, Figure 6C). Depending on the concentration and structure of the prostanoids, activation of these receptors can induce a variety of intracellular signaling pathways and cellular effects. For example, EP2, EP4, IP, and DP1 receptors activate adenylyl cyclase, increasing intracellular cyclic adenosine monophosphate (cAMP) (Ricciotti and FitzGerald, 2011). TP receptor activation mediates platelet adhesion and aggregation, smooth muscle contraction and proliferation, and endothelial inflammatory responses (Nakahata, 2008). Although TXA₂ is the preferred TP receptor ligand, other prostanoids (e.g. PGH₂), isoprostanes, and HETEs are also known activators (Audoly et al., 2000; Behm et al., 2009).

3.4.4. Lipoxygenase Pathway

Lipoxygenases metabolize AA to unstable hydroperoxy intermediates that form leukotrienes (LTs), hydroxyeicosatetraenoic acids (HETEs), and lipoxins (Figure 5). Of the eicosanoid species identified, LTs were the most abundant species detected in both vehicle-and TCDD-treated rats and also the most affected by TCDD. ALOX5 activation following interaction with ALOX5AP (ALOX5 activating protein) metabolizes AA to 5-hydroperoxyeicosatetraenoic acid (5-HPETE), which is then converted to several leukotriene species. While Alox5ap expression was unchanged by TCDD, Alox5 was dose-dependently induced 2.2-fold in the absence of AhR enrichment (Figure 6D). Likewise, TCDD dose-dependently induced Alox15 9.7-fold and Alox12 1.2-fold. 5-HPETE can be further reduced to 5-HETE or to the unstable LTA₄ epoxide. Hepatic LTA hydrolase (LTA4H) metabolizes LTA₄ to the dihydroxy acid metabolites LTB₃ and LTB₄, which were the most abundant LTs detected in both vehicle- and TCDD-treated SD rats. Although Lta4h was only induced 1.2-fold, LTB₃ and LTB₄ levels were increased 4.6- and 3.0-fold, respectively (Figure 8). LTs increase pro-inflammatory cytokine production and microvascular permeability, and are potent eosinophil, neutrophil, and monocyte chemotactic agents. Specifically, LTB₄ and LTB₃ recruit neutrophils and promote inflammatory cytokine production by infiltrating immune cells. Superoxide dismutase (Sod1), which detoxifies superoxides produced during inflammation, was induced 1.7-fold by TCDD.

LTA₄ can be conjugated to glutathione by LTC4 synthase (LTC4S) to yield LTC₄, which was increased 1.3-fold (Figure 8). Outside of the cell, LTC₄ can be converted by gamma-glutamyl transferase (GGT1) to form LTD₄, which was decreased 12.5-fold at 10 μg/kg TCDD. LTD₄ can also be metabolized by DPEP1 to produce LTE₄ (Figure 5). Although Dpep1 was induced 3.3-fold, there was no change in LTE₄ hepatic levels. However, N-acetyl LTE₄, a major inactive LTE₄ biliary metabolite (Stene and Murphy, 1987), was increased 2.1-fold by TCDD (Figure 8). LTE₄ can also be metabolized to LTF₄, which was decreased 2.3-fold (Figure 8). LTC₄, D₄, E₄, F₄, and N-acetyl LTE₄ are cysteinyl leukotrienes that increase vascular permeability and induce smooth muscle contraction (Izumi et al., 2002). Although these were the least abundant species detected, their levels were dose-dependently decreased by TCDD (Figure 8).

LTs bind to four G-protein coupled receptors: LTB₄ activates BLT1 and BLT2, while cysteinyl LTs act on CysLT1 and CysLT2 (Takeda et al., 2017) (Figure 5). Takeda et al. recently showed that TCDD increases LTB₄ levels and neutrophil infiltration in mouse and rat hepatic tissues, while these changes are not observed in BLT1-null animals (Takeda et al., 2017). LTB₄ has also been shown to enter the nucleus and activate peroxisomal proliferator-activated receptor α (PPARα) (Fielder et al., 2001), which was induced 2.1-fold by TCDD (Figure 6D). PPARα regulates gene expression associated with lipid metabolism including β-oxidation of fatty acids (Latruffe et al., 1997). It has been proposed that LTB₄ regulates the duration of inflammation through negative feedback, where activation of PPARα induces its own metabolism.

Resolvins (Rv) D1 and D2 are potent anti-inflammatory and pro-resolving lipid mediators arising from ω-3 PUFAs (i.e. DHA and EPA) (Serhan et al., 2007a; Serhan, 2007b). TCDD increased RvD1 and RvD2 6.1- and 1.8-fold, respectively, at 1 μg/kg (Figure 8). This anti-inflammatory response was attenuated at higher doses of TCDD.

4. DISCUSSION

TCDD and related AhR agonists elicit changes in gene expression, leading to hepatotoxic responses. The induction of cytochrome P450s, xanthine dehydrogenase/xanthine oxidase, and NADPH oxidases increase reactive oxygen species (ROS) levels, eicosanoid production, activation of pro-inflammatory cytokines, and the infiltration of immune cells into sites of damage. Although initially a protective or healing response, persistent inflammation can become deleterious. Recruited immune cells express cytokines and chemokines that intensify the inflammatory response, with chronic inflammation promoting the development of complex metabolic diseases (Hardwick et al., 2013). In this study, metabolomic analyses were integrated with differential gene expression and genome-wide chromatin immunoprecipitation studies to investigate the dose-dependent effects of TCDD on AA metabolism and eicosanoid biosynthesis in the female SD rat liver. TCDD increased total eicosanoid biosynthesis, consistent with the differential expression of genes associated with ω-6 PUFA metabolism.

Eicosanoid biosynthesis begins with the release of AA from damaged membranes by phospholipases. TCDD induced the expression of several secretory phospholipases, providing substrates for CYP1A-mediated production of AA metabolites such as EETs and 20-HETE. These AA metabolites are known to alter K⁺ and Ca²⁺ channel function, dysregulate intermediary metabolism, and contribute to wasting syndrome (Gilday et al., 1998; Rifkind, 2006). With the exception of the 1A family, most hepatic CYPs were repressed in TCDD-treated rats, consistent with previous studies in C57BL/6 mice (Nault et al., 2017). Therefore, induction of Cyp1a1 and 1a2 (and to a lesser extent Cyp2c24 based on expression levels) is likely responsible for most of the eicosanoid production via cytochrome P450 metabolism.

Although the cytochrome P450 hydroxylation/epoxidation pathway has been studied, few investigated the effects of TCDD on the lipoxygenase and cyclooxygenase pathways. Levels of eicosanoids produced by the lipoxygenase and cyclooxygenase pathways were altered by TCDD including increases in PGE₂ and its metabolites bicyclo-PGE₂ and PGB₂. While changes were consistent with altered gene expression, there was no evidence of AhR enrichment. Among the eicosanoid species identified, LTs were most significantly affected, consistent with induction of lipoxygenases (Alox5, Alox12, and Alox15) and downstream LT biosynthesis enzymes, while the cysteinyl LTs were decreased. Interestingly, several LTA₄ metabolites including 5,6-DiHETE isomers activate the AhR (Chiaro et al., 2008), suggesting increased LT levels could have broader implications in enhancing hepatotoxicity. Furthermore, studies have shown that treating with the leukotriene inhibitor, Montelukast, decreased TCDD toxicity (Bentli et al., 2016).

Recently, Takeda et al. showed similar AhR-dependent increases in hepatic LTB₄ in male rats and mice (Takeda et al., 2017). Although the majority of lipoxygenase and cyclooxygenase genes contain pDREs, our ChIP-Seq analysis suggests dysregulation does not involve direct AhR binding. Furthermore, the increase in total leukotrienes, and more specifically LTB₃ and LTB₄, may be due to changes in the ω-6/ω-3 PUFA ratio. Higher ω-6/ω-3 ratios correlated with NAFLD severity (Araya et al., 2004; Puri et al., 2007). TCDD decreased ω-3 PUFAs including eicosapentaenoic acid (EPA), leading to an increase in the ω-6/ω-3 ratio. Studies in rat primary hepatocytes show EPA protects against TCDD-induced hepatotoxicity (Turkez et al., 2012). Similarly, the ω-3 PUFA docosahexaenoic acid (DHA) has also been shown to be protective (Turkez et al., 2016). Moreover, dietary supplementation of ω-3 PUFAs in NAFLD patients reduced hepatic fat accumulation and liver enzyme levels (Parker et al., 2012), suggesting decreased ω-3 PUFAs may exacerbate TCDD-induced hepatotoxicity.

Integration of differential gene expression with complementary untargeted metabolomics demonstrated that PUFA metabolism and eicosanoid biosynthesis were disrupted by TCDD. These changes increased inflammatory lipid mediators which may be contributing to increased inflammation in chronically exposed rats (Kociba et al., 1978; NTP, 1994; Walker et al., 2006). Based on BMD values for differential gene expression with correlative changes in metabolites, the cytochrome P450 hydroxylation/epoxidation pathway and the lipoxygenase pathways were more sensitive to TCDD than the cyclooxygenase pathway. Although EETs are implicated in TCDD-elicited hepatotoxicity, this suggests LTs also play a key role in AhR-mediated inflammation. Overall, this study provides further evidence that dysregulation of eicosanoid metabolism promotes the progression of TCDD-elicited hepatic steatosis to steatohepatitis with fibrosis.

HIGHLIGHTS.

• Metabolomics identified perturbation of eicosanoid metabolism in TCDD treated rats.

• TCDD increased the ratio of pro- and anti-inflammatory ω-6 to ω-3 PUFAs.

• TCDD altered lipoxygenase and cyclooxygenase pathway produced eicosanoids.

• Absence of AhR binding at eicosanoid metabolism genes indicates secondary effects.