Metabolomics Reveals that Aryl Hydrocarbon Receptor Activation by Environmental Chemicals Induces Systemic Metabolic Dysfunction in Mice

Abstract

Environmental exposure to dioxins and dioxin-like compounds poses a significant health risk for human health. Developing a better understanding of the mechanisms of toxicity through activation of the aryl hydrocarbon receptor (AHR) is likely to improve the reliability of risk assessment. In this study, the AHR-dependent metabolic response of mice exposed to 2,3,7,8-tetrachlorodibenzofuran (TCDF) were assessed using global ¹H nuclear magnetic resonance (NMR)-based metabolomics and targeted metabolic profiling of extracts obtained from serum and liver. ¹H NMR analyses revealed that TCDF exposure suppressed gluconeogenesis and glycogenolysis, stimulated lipogenesis, and triggered inflammatory gene expression in an Ahr-dependent manner. Targeted analyses using gas chromatography mass spectrometry showed TCDF treatment altered the ratio of unsaturated/saturated fatty acids. Consistent with this observation, an increase in hepatic expression of stearoyl coenzyme A desaturase 1 was also observed. In addition, TCDF exposure resulted in inhibition of de novo fatty acid biosynthesis manifested by down-regulation of acetyl-CoA, malonyl-CoA and palmitoyl-CoA metabolites and related mRNA levels. In contrast, no significant changes in the levels of glucose and lipid were observed in serum and liver obtained from Ahr-null mice following TCDF treatment, thus strongly supporting the important role of the AHR in mediating the metabolic effects seen following TCDF exposure.

INTRODUCTION

2, 3, 7, 8-tetrachlorodibenzofuran (TCDF), 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) and many structurally related halogenated aromatic hydrocarbons (HAHs) and polynuclear aromatic hydrocarbons (PAHs) are environmentally persistent and toxic compounds commonly generated as byproducts of municipal waste combustion.¹ These environmental contaminates can exert diverse biological and toxic effects.^2–4

The molecular mechanisms of TCDD toxicity have been extensively investigated by a number of microarray analyses^5,6 and metabolomic approaches^7–9 using rodent models. For example, the AHR target gene TiPARP has been reported to be associated with suppression of hepatic gluconeogenesis by TCDD and confirmed by decreased expression of phosphoenolpyruvate carboxykinase (Pepck), glucokinase (Gck), and glucose 6-phosphatase (G6pase) genes, which encode for enzymes controlling gluconeogenic flux.¹⁰ Interestingly, TCDD exposure in rats results in tissue-specific differences in the regulation of PEPCK activity including a significant decrease in liver PEPCK activity yet kidney and brown adipose tissue show a significant increase in PEPCK activity.¹¹ Serum metabolomics revealed that down-regulation of hepatic carboxylesterase 3 was related to an inflammatory cascade involved in steatohepatitis induced by dioxin exposure without activation of the AHR signaling pathway.⁸ Aside from hepatoxicity, TCDD-induced immunotoxicity and inflammation has also been shown to be associated with Foxp3+ nTreg cells and their proportional increase in abundance at sites of inflammation through the activation of AHR.^12,13 Hence, a holistic systems approach, facilitated through metabolomics approaches in similar fashion to previously published studies with TCDD, ^7–9,50 may be useful to further our understanding of the dynamic metabolic effects of TCDF in vivo.^14–17

In the current study, we analyzed TCDF-induced changes in serum and liver extracts of wild-type and Ahr-null mice using a combination of global ¹H NMR and targeted GC/LC-MS based metabolomic approaches. The objectives were to investigate the metabolic responses of mice to TCDF exposure and to further our understanding of the role the AHR plays in mediating the response to these environmental chemicals.

MATERIALS AND METHODS

Chemicals

Sodium chloride, K₂HPO₄ and NaH₂PO₄ (all analytical grade) were obtained from Sigma-Aldrich Chemical Co. Ltd. (St Louis, MO). Sodium 3-trimethylsilyl [2,2,3,3-d₄] propionate (TSP-d₄) and D₂O (99.9% in D) and 2, 3, 7, 8-tetrachlorodibenzofuran (TCDF, 99.4%) were purchased from Cambridge Isotope Laboratories (Miami, FL).

Animal Experiment and Samples Collection

Animal experimental procedures were performed according to the National Institutes of Health (NIH) guidelines and approved by the Pennsylvania State University Institutional Animal Care and Use Committee. Male C57BL/6J mice (6 weeks old) were purchased from the Jackson Laboratory (Bar Harbor, Maine). A total of 12 mice were housed in temperature- and a light-controlled room and given water and food ad libitum. After acclimatization for one week, mice were randomly divided into two groups of six and trained to eat dough pills (Bio-Serve, Frenchtown, NJ) without TCDF for another week. The dough pills with TCDF (acetone solution as vehicle) were prepared with tablet molds and uniformly contained 0.15 µg TCDF for one mouse per pill, thus providing final dosage of 5 µg/kg body weight, respectively. In the TCDF studies, mice were fed with the dough pills containing TCDF continuously for five days (one pill per mouse per day). Male Ahr^−/− mice (C57BL/6J congenic mice) were obtained from Dr. Christopher Bradfield (University of Wisconsin, Madison WI) and bred in house. The TCDF dose used in the Ahr^−/− mice was 24 µg/kg body weight and the treatment procedure was the same as described above with the dough pills containing TCDF continuously for five days (one pill per mouse per day). Urine and feces were collected every other day (i.e., day 2, 4, and 6) over the TCDF treatment period. Blood, liver and intestinal tissue samples were collected immediately following CO₂ asphyxiation on day 7. All samples were stored at −80 °C until analysis.

Histopathology

Formalin fixed liver tissue was embedded in paraffin wax, sectioned (3–4 µm), and stained with hematoxylin and eosin (H&E) followed with examination under a microscope.

RNA Isolation and Quantitative Gene Expression Analysis

RNA was extracted from frozen liver and ileum tissue (50 mg) using TRIzol reagent (Invitrogen, Carlsbad, CA). After extraction, the concentration of RNA was determined by NanoDrop (ND-1000, V 3.3) and its quality was confirmed by the 260/280 and 260/230 ratios. cDNA was synthesized from 1 µg of total RNA using Superscript II reverse transcriptase (Invitrogen). Gene-specific primers were used in each reaction and all results were normalized to β-actin mRNA. Gene expression was performed on an ABI Prism 7900HT Fast Real-Time PCR sequence detection system (Applied Biosystems, Foster City, CA).

Sample Preparation for NMR Spectroscopy

Serum samples were prepared by mixing 200 µL serum with 400 µL saline solution containing 30% D₂O; 550 µL samples was transferred into 5 mm NMR tubes after vortexing and centrifugation (11180 x g, 10 min, 4 °C). Liver tissues (~50 mg) were extracted three times with 600 µL of precooled methanol-water mixture (2/1, v/v) using the PreCellys Tissue Homogenizer (Bertin Technologies, Rockville, MD). After centrifugation at 11180 x g for 10 min at 4 °C, the combined supernatants were dried. Each of the aqueous extracts was separately reconstituted into 600 µL phosphate buffer (K₂HPO₄/NaH₂PO₄, 0.1 M, pH 7.4, 50% v/v D₂O) containing 0.005% sodium 3-trimethylsilyl [2,2,3,3-d4] propionate (TSP-d₄) as chemical shift reference. Following centrifugation, 550 µL of each extract was transferred into 5 mm NMR tube for NMR analysis.

¹H NMR Spectroscopy

¹H NMR spectra of serum and aqueous liver extracts were acquired at 298 K on a Bruker Avance III 600 MHz spectrometer (operating at 600.08 MHz for ¹H and at 150.93 MHz for ¹³C) equipped with a Bruker inverse cryogenic probe (Bruker Biospin, Germany). For aqueous liver extracts, typical one-dimensional NMR spectrum was acquired for each of all samples employing the first increment of NOESY pulse sequence (NOESYPR1D). For serum, water-presaturated Carr-Purcell-Meiboom-Gill (CPMG) CPMG pulse sequence (recycle delay-90°-(τ-180°-τ)_n-acquisition) was employed to attenuate NMR signals from macromolecules whereas diffusion-edited spectra can be acquired to obtain only signals of macromolecules such as lipid, lipoprotein and long chain fatty acids. To suppress the water signal, a weak continuous wave irradiation in CPMG method was applied to the water peak during recycle delay (2 s) and mixing time (100 ms). Diffusion-edited spectra were acquired as with diffusion time (Δ) of 200 ms, a duration of the magnetic field pulse gradients (δ) of 1000 us and pulse-field gradient strength of 31.2 G/cm. The 90° pulse length was adjusted to approximately 10 µs for each sample and 64 transients were collected into 32 k data points for each spectrum with spectral width of 20 ppm. To facilitate NMR signal assignments, a range of 2D NMR spectra were acquired and processed for selected samples including ¹H-¹H correlation spectroscopy (COSY), ¹H-¹H total correlation spectroscopy (TOCSY), ¹H-¹³C heteronuclear single quantum correlation (HSQC), and ¹H-¹³C heteronuclear multiple bond correlation spectra (HMBC).

Spectral Data Processing and Multivariate Data Analysis

All free induction decays (FID) were multiplied by an exponential function with a 1 Hz line broadening factor prior to Fourier transformation. The spectra were referenced to TSP-d₄ at δ 0.00 when TSP-d₄ was present in liver extracts. Otherwise, the chemical shift of anomeric proton signal of α-glucose (δ 5.233) was used as chemical shift reference for serum. ¹H NMR spectra were corrected manually for phase and baseline distortions and spectral region δ 0.50–9.50 was integrated into regions with equal width of 0.004 ppm (2.4 Hz) using AMIX software package (V3.8, Bruker-Biospin). Region δ 4.60–5.15 was discarded by imperfect water saturation. Each bucketed region was then normalized to the total sum of the spectral integrals to compensate for the overall concentration differences prior to statistical data analysis.

Multivariate data analysis was carried out with SIMCAP+ software (version 13.0, Umetrics, Sweden) as described.^32,48 Briefly, Principal Component Analysis (PCA) and Orthogonal Projection to Latent Structures with Discriminant Analysis (OPLS-DA) were conducted on the NMR data. The OPLS-DA models were validated using a 7-fold cross validation method and the quality of the model was described by the parameters R²X and Q² values (Figures 1, 2, 3 and 4).¹⁸ After back-transformation of the loadings generated from the OPLS-DA, color-coded correlation coefficient loading plots (MATLAB, The Mathworks Inc.; Natick, MA) were employed to indicate the significance of the metabolite contribution to the class separation with a “hot” color (e.g., red) being more significant than a “cold” color (e.g., blue). In this study, a cutoff value of |r| > 0.707 (r > 0.707 and r < −0.707) was chosen for correlation coefficient as significant based on the discrimination significance (p ≤ 0.05).

Figure 1.

Representative 600 MHz ¹H NMR spectra of liver extracts from the vehicle (A) and TCDF-treated mice (B). The regions of δ 6.10–9.20 and δ 0.60–3.10 in the liver spectra was vertically expanded 16 times and 4 times compared with the region of δ 3.10–4.70, respectively. O-PLS-DA scores (left) and coefficient-coded loadings plots (right) for the models obtained from the NMR data of liver aqueous extracts (C). The model was cross-validated with CV-ANOVA, p = 2.33 × 10⁻³ corresponding to liver extracts of mice after vehicle and TCDF exposure. Metabolite assignments are shown in Figure 1 and Table S1.

Figure 2.

Representative 600 MHz ¹H NMR CPMG spectra of serum from the vehicle (A) and TCDF-treated mice (B). The regions of δ 6.00–8.50 and δ 5.00–5.50 in the liver spectra was vertically expanded 16 times and 4 times compared with the region of δ 0.50–4.40, respectively. O-PLS-DA scores (left) and coefficient-coded loadings plots (right) for the models obtained from the NMR data (C). The model was cross-validated with CV-ANOVA, p = 0.035 corresponding to serum of mice after vehicle and TCDF exposure, respectively. Metabolite keys are shown in Figure 2 and Table S1.

Figure 3.

Representative 600 MHz ¹H NMR diffusion-edited spectra of serum from vehicle (A) and TCDF-treated mice (B) and O-PLS-DA scores (left) and coefficient-coded loadings plots (right) for the models obtained from the NMR data (C). The model was cross-validated with CV-ANOVA, p = 2.06 × 10⁻³ corresponding to serum of mice after vehicle and TCDF exposure, respectively. Metabolite keys are shown in Figure 3 and Table S1.

Figure 4.

O-PLS-DA scores (left) and coefficient-coded loadings plots (right) for the models obtained from the CPMG (A) and diffusion-edited (B) NMR data acquired from serum of Ahr^−/− control and TCDF-treated Ahr^−/− mice. The model was cross-validated with CV-ANOVA, p = 0.571 and 1.392 corresponding to serum of Ahr^−/− mice after vehicle and TCDF exposure, respectively. Metabolite keys are shown in Figure 3 and Table S1.

Fatty Acid Profiling by GC-MS

Liver tissues (~50 mg) were mixed with 1 mL of methanol-chloroform (2/1, v/v) with addition of 5 µL internal standards (50 µM C15:0 free acid and the methyl ester of C17:0) and then homogenized using the Precellys Tissue Homogenizer (Bertin Technologies, Rockville, MD). After centrifugation (20187 x g, 4 °C) for 15 min, the supernatant was collected. 500 µL saline (0.9 %) was added to the liver extracts and serum (50 µL). After vortexing for 5 min and centrifugation (20187 x g, 4 °C) for 15 min, the mixture was separated into two layers. The solution in the bottom layer was transferred into 10 ml glass tube and dried with nitrogen gas. After adding 1 mL MeOH/HCl (41.5ml/9.7ml) and vortexing for 5 min, the solution was incubated overnight at 60°C. The resultant mixture was combined with 5 mL hexane and 5 mL saline. Following vortexing for 5 min, the top layer was collected and dried down with nitrogen gas. The resultant residues were re-dissolved in 200 µl hexane and then transferred to an autosampler vial for GC-MS analysis. Fatty acid composition was measured on an Agilent 7890A-5975C GC-MS system (Agilent Technologies, Santa Clara, CA). A HP-5MS (Agilent Technologies) capillary column (30 m, 0.25 mm ID, 0.25 µm film thickness) was employed with helium as a carrier gas at flow rate of 1mL/min. Sample injection volume was 0.5 µL with a pressure pulsed split ratio (1:10 split, 10 psi). The injection port and detector temperatures were 230 and 250 °C, respectively. The initial column temperature was 80 °C where it was held for 1 min and then increased to 205 °C at a rate of 20 °C/min, and then increased to 220 °C at a rate of 2 °C/min, and then increased to 310 °C at a rate of 15 °C /min, where it was held for 2 min. Fatty acids were quantified by comparing integrated peak areas following normalization to the internal standards.

Coenzyme A profiling by UPLC-TQS-MS

Targeted analysis of coenzyme A was performed using an Acquity UPLC system coupled to a Waters Xevo TQS MS with a C18 BEH (2.1 × 100 mm, 1.7 µm) UPLC column (All Waters, Milford, MA). The internal standard palmitoyl-¹³C₁₆ coenzyme A was obtained from Sigma-Aldrich. Liver tissues (50 mg) were extracted with 750 µL of methanol containing 1mM EDTA by homogenization and then mixed with 375 µL chloroform with vortexing for 15 sec. Following centrifugation (18000 x g, 4 °C) for 5 min, the top layer was collected. After repeating the above extraction process twice with 500 µl water/methanol/chloroform (45:50:5 v/v/v), the combined supernatants were dried down and later dissolved with 400 µL of 10% aqueous acetonitrile containing internal standard (1.0 µM final concentration of palmitoyl-¹³C₁₆ coenzyme A) by sonication. Following vortex and centrifugation, 300 µL supernatants were transferred to an autosampler vial. Analytes were detected by selected reaction monitoring (SRM) or multiple reactions monitoring (MRM) and quantified by measuring the integrated peak areas, and normalized to the response for palmitoyl-¹³C₁₆. Results were calculated according to individual standard curve established with the response (area_analyte/area_{internal std.}).

Serum lysophosphatidylcholine profiling by UHPLC-QTOF-MS

Serum LPCs were extracted using the Folch method.⁵¹ Briefly, 50 µL serum was mixed with 100 µL chloroform and methanol (2:1 v/v). The samples were incubated at room temperature for 5 min after vortexing. Following centrifugation (18000 x g, 4 °C) for 20 min, the bottom layer was collected and then dried down under a gentle stream of nitrogen gas. The samples were dissolved with 125 µL of solution containing isopropanol, acetonitrile and water (2:1:1 v/v/v). Following vortex and centrifugation, 100 µL supernatants were transferred to an autosampler vial. Samples (5 ul) were separated by reverse phase UHPLC using a Prominence 20 UFLCXR system (Shimadzu, Columbia MD) with a Waters (Milford, MA) CSH C18 column (100mm x 2.1mm 1.7 um particle size) maintained at 55C and a 20 minute aqueous/acetonitrile/isopropanol gradient, at a flow rate of 225 uL/min. Solvent A was 40% water, 60% acetonitrile with 10 mM ammonium formate and 0.1% formic acid, and Solvent B was 90% isopropanol, 10% acetonitrile with 10 mM ammonium formate and 0.1% formic acid. The initial condition were 60% A and 40 % B, increasing to 43% B at 2 min, 50% B at 2.1 min., 54% B at 12 min, 70% B at 12.1 min and 99% B at 18 min., held at 99% B until 20.0 min before returning to the initial conditions. The eluate was delivered into a 5600 (QTOF) TripleTOF using a Duospray™ ion source (all AB Sciex, Framingham, MA).

Data Analysis

All the experimental values are presented as mean ± s.d. Graphical illustrations and statistical analysis were performed with GraphPad Prism version 6.0 (GraphPad). P-values < 0.05 were considered significant.

RESULTS

Metabolic Profiling by ¹H NMR Analysis

¹H NMR spectra of aqueous liver extracts and serum were obtained from vehicle mice and mice treated with TCDF (Figures 1, 2, and 3). The metabolite resonances were assigned on the basis of previous publications^19–21 and confirmed by a range of 2D NMR experiments (Table S1). Liver extracts spectra (Figure 1) were dominated by amino acids, glucose, glycogen, choline-containing metabolites, organic acids, lipid moieties, and nucleoside metabolite signals. Serum CPMG spectra (Figure 2) contained peaks from glucose, amino acids, carboxylic acids such as lactate, fumarate and D-3-hydroxybutyrate (3-HB), and choline-containing metabolites. Serum diffusion-edited spectra (Figure 3) were comprised of lipoproteins, glycoproteins, triglycerides (TG), phospholipids and long chain fatty acids (LCFA).

Dietary TCDF-Induced Metabolic Changes

PCA was initially performed on the normalized NMR data collected from serum and liver extracts to visualize intrinsic similarities/dissimilarities between the samples and for detection of abnormal data points. Comparative OPLS-DA was subsequently conducted between the NMR data from the TCDF-treated and vehicle-treated mice for the different biological matrices. The cross-validated scores plots indicate separations between the metabolic profiles of vehicle and corresponding TCDF-dosed mice, whereas the OPLS-DA correlation coefficient plots indicate the relative contribution of the metabolites. Here, the model quality indicators (R²X and Q²) clearly showed that the metabolite profiles obtained from the different treatment groups were distinctive. These observations were further supported by results from the model evaluation with CV-ANOVA and permutation tests for the models (Figure S1 and Table S2). The metabolites with statistically significant contributions to group discrimination between TCDF-treated and vehicle-treated groups are labeled in the corresponding color-coded coefficient plots (Figure 1C, 2C, and 3C).

Compared with vehicle-treated mice, TCDF exposure resulted in significant alterations of metabolic profiles of liver extracts as illustrated in the cross-validated OPLS-DA coefficient plot (Figure 1C). Dietary TCDF significantly elevated the levels of lipid moieties, unsaturated fatty acid (UFA), and glycogen but significantly decreased the levels of hepatic glucose and lactate. The same OPLS-DA strategy was further applied to the metabolic profiles of serum obtained from TCDF-treated and vehicle-treated mice (Figure 2C, 3C and 4). Compared with the vehicle-treated mice, higher levels of lipid, N-acetyl glycoproteins (NAG), O-acetyl glycoproteins (OAG), phospholipid and LDL/VLDL and lower levels of alanine, citrate, UFA, phosphocholine (PC) and glycerophosphocholine (GPC) were also observed in the serum of TCDF-treated mice. However, the Q² values (0.29, 0.26 and 0.11 for the models of liver, serum with CPMG, and diffusion-edited NMR data, respectively) of the OPLS-DA models of data from liver and serum of Ahr^−/− mice revealed no significant differences between the TCDF-treated and control (Figure 4, Table S2).

Targeted fatty acid compositional analysis revealed that TCDF exposure led to significant decreases in saturated fatty acids (C18:0 and C20:0) and elevation of mono-saturated fatty acids (C18:1n9) and polyunsaturated fatty acids (C18:2n6, C20:3n6, C20:3n3 and C20:5n3) in the liver (Figure 5A). A significant decrease in saturated fatty acids (C18:0 and C22:0) and an elevation of polyunsaturated fatty acids (C18:2n6) were also found in the serum of TCDF-treated mice (Figure 5B). TCDF exposure also significantly decreased serum lysophosphatidylcholine (LPC) metabolites including C16:0-LPC, C18:0-LPC, C16:1-LPC, C18:2-LPC, and C20:4-LPC (Figure 5D). In addition, significant decreases of coenzymes including acetyl-CoA, butyryl CoA, malonyl CoA, hexanoyl CoA, myristoyl CoA, palmitoyl CoA and stearoyl CoA together with significant elevation of lineleoyl CoA in the livers of TCDF-treated mice were observed (Figure 6A).

Figure 5.

Fatty acid composition in liver (A) and serum (B) of vehicle-treated (blue bars) and 5 µg/kg TCDF-treated mice (green bars). Fatty acids were extracted by the Folch method and quantified by GC-MS. QPCR analysis of Scd1, Fasn and Acaca mRNA expression in the liver (C) and lysophosphatidylcholine (LPC) composition in serum (D) by LC-MS. The primers used in this study are listed in Table S2. Data are presented as mean ± s. d, n = 6 per group; *p < 0.05, **p < 0.01, ***p < 0.001, two-tailed Student’s t-test.

Figure 6.

Coenzymes CoA composition in liver (A) from vehicle-treated (blue bars) and 5 µg/kg TCDF-treated mice (green bars). QPCR analysis of mRNA levels of Cyp1a1, G6pase, Glut2 and Pepck in the liver (B) and inflammatory cytokine (Lcn-2, IL-1β, TNF-α and Saa1) expression in the ileum (C). The primers used in this study are listed in Table S2. Data are presented as mean ± s. d., n = 6 per group; *p < 0.05, **p < 0.01, ***p < 0.001, two-tailed Student’s t-test.

Dietary TCDF Alters Expression Profile of Genes Associated with Lipid and Glucose Metabolic Pathways

In order to monitor the transcriptional (and/or changes in mRNA stability/turnover) responses associated with TCDF treatment, we measured the expression of those genes involved in xenobiotic responses, inflammatory signaling, lipid, and glucose metabolism using QPCR. One of the transcriptional targets of AHR, Cyp1a1, which is commonly regarded as the most important marker of xenobiotic response, was found to be significantly induced in the liver and ileum of mice after TCDF exposure (Figures 6B and C). TCDF exposure caused significant increase in gene expression of stearoyl coenzyme A desaturase-1 (Scd1), the key enzyme controlling desaturation of saturated fatty acid, in TCDF-treated mouse livers (Figure 5C). Significant depletion in the expression of fatty acid synthase (Fasn) and acetyl-CoA carboxylase 1 (Acaca) mRNAs, encoding important enzymes involved in the process of de novo fatty acid biosynthesis, in the livers of TCDF-treated mice was also noted (Figure 5C). Furthermore, dietary TCDF exposure caused significant down-regulation of G6pase, Glut2, and Pepck expression in the liver (Figure 6B), and up-regulation of Lcn-2, Il-1β, Tnf-α and Saa1 expression in the ileum tissues (Figure 6C) involved in glucose metabolism and inflammatory signaling, respectively.

DISCUSSION

Dioxins and dioxin-like compounds such as TCDD and TCDF exert their toxic effects through activation of the AHR.^22,23 Activation of AHR by these potent agonists leads to a multitude of adverse effects such as liver toxicity and may be associated with the development or exacerbation of obesity.^24,25 The objective of the current study was to investigate the metabolic consequences of TCDF exposure that was not associated with overt liver toxicity. We examined metabolic alterations in serum and liver after 5 days of continuous TCDF exposure in both Ahr^+/+ and Ahr^−/− mice using a global ¹H NMR metabolomics approach. The effects of dietary TCDF on fatty acid metabolism and related CoA levels were further investigated by targeted GC/LC-MS analyses. QPCR analyses were also employed to validate the relevant gene expression. The results indicated that TCDF exposure caused comprehensive metabolic alterations in serum and liver, which were found to be AHR-dependent (Figure 7).

Figure 7.

A summary of the significant changes found in the host metabolome of mice exposed to TCDF. Dietary TCDF altered many metabolic pathways involved in hepatic lipogenesis, gluconeogenesis, glycogenolysis, TCA cycle, amino acids, and lipid metabolism and inflammatory signaling pathways. The metabolites or mRNAs in red or blue represent a higher or lower level in the respective biological matrices (liver and serum) obtained from TCDF-treated mice compared to vehicle.

Hepatotoxicity Induced by Low TCDF

Based on allometric scaling calculations, the TCDF dose (5 µg/kg body weight) used in this study is equivalent to 600 ng/kg body weight in humans and lower than that used in our previous study (24 µg/kg body weight); however this calculation does not consider total body burden.⁴⁸ In this study, liver histopathology showed that dietary TCDF (5 µg/kg body weight) exposure in mice induced mild hepatotoxicity (Figure S2). In addition, dietary TCDF exposure induced significant changes in xenobiotic and energy metabolism based on QPCR and global ¹H NMR metabolomic analyses. The metabolic pathway related to AHR activation was altered, manifested by higher expression of Cyp1a1, which is involved in the detoxification of xenobiotics in the liver and intestinal tract of TCDF-treated mice. Hepatic cytochromes p450s are useful biomarkers of xenobiotic toxicity and have been widely used to evaluate the toxic effects of environmental pollutants.^26,27,50 Our previous study showed that Ahr^−/− mice exhibited no significant change histopathologically and had no change in the liver enzymes ALT, ALP and CYP1A1,⁴⁸ suggesting TCDF mediated mild liver injury through activation of AHR. Further, previous study showed that the Ahr^−/− mice were relatively unaffected by doses of TCDD (2000 µg/kg) that were tenfold higher than those found to induce severe toxic and pathologic effects in wild-type mice.⁵²

The depletion of citrate in serum and fumarate in liver induced by dietary TCDF exposure (Figure 1 and 2) suggests an alteration of TCA cycle (Figure 7), leading to an imbalance of the global energy supply. Citrate and fumarate are important intermediates for the TCA cycle, which is crucial for energy metabolism. Previous studies indicated TCDD-induced hepatotoxicity was initiated by an inhibition of mitochondrial respiratory function,²⁸ which inevitably led to changes in the TCA cycle intermediates in the liver and serum profiles. This can also be confirmed by significant depletion of acetyl-CoA level, which is regarded as the “Hub of Metabolism”²⁹ and is involved in many biochemical reactions, in the liver of TCDF-treated mice. Its main function is to convey the carbon atoms within the acetyl group to the TCA cycle to be oxidized for energy production.³⁰ Similar observations have been reported in the cases of hepatotoxicity induced by acetaminophen,³¹ aflatoxin-B1 (AFB1)³² and allyl formate.³³ Further, the decrease in alanine levels also indicated perturbations of energy metabolism induced by dietary TCDF.³⁴

Lipid Metabolism

The liver is an essential organ in metabolic homeostasis. In the current study, significantly elevated levels of lipid and unsaturated fatty acid (UFA) in the liver of TCDF-treated mice (Figure 1C) indicated hepatic lipogenesis which appears to be a common cellular response to a number of liver toxicants.^31–33 Significant elevation of LDL/VLDL and depletion of HDL, PUFA, and UFA was also observed in the serum of TCDF-treated mice (Figure 3C). The hepatic lipidosis caused by TCDF exposure may also be due to the reduced alterations in lipid transport from the liver to the serum as in the case of the hepatic lipidosis induced by allyl formate,³³ AFB1³² and methionine.³⁵ Indeed, markedly decreased expression and activity of Cd36, Apob48, and Apob100 enzymes related to transport of lipid out of the hepatic cells have been observed, in the liver of TCDD-treated rats from previous genomic studies.^36,37 However, no significant changes in the levels of lipoprotein, lipid, and fatty acids were observed in serum between TCDF-treated and vehicle-treated Ahr^−/− mice, showing TCDF exposure impacts host lipid export and utilization in an AHR-dependent manner.

GC-MS fatty acid analysis revealed that TCDF-induced disruption in lipid metabolism with a significant elevation in UFA levels and depletion of saturated fatty acids (Figure 5A), indicating a decrease in lipid oxidation with concurrent increases in the biosynthesis of UFAs via desaturation pathways. This implied TCDF-treatment promoted biotransformation from saturated fatty acids to UFA likely via activation of Scd1, which was in agreement with previous observation.⁵⁰ TCDF decreased stearate (18:0) and increased oleate (18:1n9) which is consistent with the QPCR results that Scd1 mRNA levels in TCDF-treated mouse liver were higher than in vehicle-treated mice liver. Additionally, others have reported that TCDD induced the up-regulation of Elovl5 which plays a key role in elongation of long-chain polyunsaturated fatty acids.^37,38

TCDF-treatment induced significant depletion in the expression of hepatic Fasn and Acaca mRNAs, encoding important enzymes involved in the process of de novo fatty acid biosynthesis, suggesting that TCDF suppressed de novo fatty acid biosynthesis. This was supported by the observation that TCDF induced the down-regulation of acetyl-CoA, palmitoyl CoA, malonyl-CoA, and stearoyl CoA, the key metabolites involved in fatty acid biosynthesis. Consistently, previous studies also indicated TCDD induced inhibition of de novo fatty acid biosynthesis and β-oxidation.^39,40 Interestingly, activation of AHR by single injection of β-naphthoflavone (BNF) into mouse and primary human hepatocytes was found to attenuate the expression of lipogenic genes including Scd1, Acaca, and Fasn), through a DRE-independent mechanism, indicating that AHR plays a key role in the inhibition of fatty acid synthesis in mice and humans.⁴¹

Serum metabolomics analysis in the present study revealed that dietary TCDF exposure was associated with decreased levels of phosphatidylcholine (PC), glycerolphosphocholine (GPC) (Figure 3C) and lysophosphatidylcholine (LPC) metabolites including C16:0-LPC, C18:0-LPC, C16:1-LPC, C18:2-LPC, and C20:4-LPC (Figure 5D). LPC is a precursor of PC and both PC and GPC are known to be important endogenous compounds associated with bile acid excretion and maintenance of choline homeostasis.⁴² Moreover, PC and GPC are not only essential components of cell biomembranes, but also protect cells and their organelles from oxidative stress and lipotoxicity.⁴³ Our previous study reported that TCDF exposure induced disruption of bile acid excretion such as elevation of tauro-β-muricholate (TβMCA) and deoxycholic acid (DCA).⁴⁸ In agreement with our findings, previous investigations reported similar changes in serum metabolites and related mRNA levels including increased expression of hepatic Lpcat1–4 and Abcc1/4 and reduced expression of Slc10a1 and Slco1a1 in methionine- and choline-deficient diet-induced nonalcoholic steatohepatitis and in GalN-injected ob/ob mice models.⁴⁴ Collectively, these observations suggested dietary TCDF disrupted cell structural integrity and triggered robust inflammation, which subsequently may lead to steatohepatitis at higher doses.⁴⁹

Glucose and Glycogen Metabolism

Compared with vehicle, mice administered dietary TCDF exhibited significantly reduced glucose and lactate and elevated glycogen levels in liver (Figure 1C), suggesting that one of the consequences of TCDF exposure is inhibition of gluconeogenesis and glycogenolysis. Similar observations have been reported in the case of suppression of hepatic gluconeogenesis by TCDD through the AHR target gene TiPARP.¹⁰ Further, such depletion of hepatic glucose has also been observed in human HCC and rats exposed to perfluorododecanoic acid²¹ and mycotoxin.³² Consistently, as reported here, markedly decreased expression of Pepck, G6pase, and Glut2, which encode enzymes important for controlling gluconeogenic flux, in the liver of TCDF-treated mice (Figure 6B). Moreover, a previous microarray study also reported that several enzymes involved in carbohydrate metabolism such as glucokinase (Gck) and pyruvate carboxylase (Pc) were also down-regulated after TCDD exposure.⁵ Taken together, disruption of lipid and carbohydrate metabolism further indicated hepatotoxicity and subsequent perturbations of energy metabolism induced by dietary TCDF exposure.

Gut Responses and Inflammation

Aside from hepatotoxicity, the gastrointestinal tract was affected by dietary TCDF exposure, which was manifested by up-regulation of Cyp1a1 and Lcn-2, Il-1β, Tnf-α and Saa1 in the ileum tissues, indicating significant AHR activation and inflammation in the gut (Figure 6C). One possible explanation for these results is that the AHR plays a direct role in the regulation of inflammatory gene expression. This notion is supported by recent studies that have shown that there are functional dioxin response elements in the Il-6 promoter that are capable of synergistically increasing transcription in the presence of an inflammatory signal and an AHR ligand.⁴⁵ A previous study showed that LCN-2 was a sensitive and dynamic non-invasive biomarker for intestinal inflammation; LCN-2 levels are dramatically increased upon TCDF exposure.⁴⁶ In addition, dietary TCDF exposure induced markedly elevated levels of OAG and NAG in serum. High levels of OAG and NAG in mouse serum, which are regarded as “acute-phase” glycoproteins related to inflammation of injured tissue,⁴⁷ further confirmed inflammatory conditions in mice after TCDF exposure. Our previous study showed that Ahr^−/− mice exhibited no significant changes in expression of Il-1β and Tnf-α, indicating that the inflammation triggered by TCDF was dependent on AHR.⁴⁸

In conclusion, dietary TCDF exposure induced liver dysfunction and significant alterations in several metabolic pathways, including hepatic lipogenesis, perturbed TCA cycle, disrupted carbohydrate and amino acid metabolism as well as inhibition of de novo fatty acid biosynthesis (Figure 6). Such dysregulation may also cause gut dysfunction and inflammation. Interestingly, the mild liver injury and the subsequent metabolic disorders induced by TCDF exposure were found to be AHR-dependent. Our results have highlighted that an integrated global ¹H NMR and targeted GC/LC-MS analyses of multiple biological matrices provided a systemic method for elucidating mechanisms underlying a toxic insult.