Exposure to low-dose perfluorooctanoic acid promotes hepatic steatosis and disrupts the hepatic transcriptome in mice

Objective Perfluoroalkyl substances (PFAS) are man-made chemicals with demonstrated endocrine-disrupting properties. Exposure to perfluorooctanoic acid (PFOA) has been linked to disturbed metabolism via the liver, although the exact mechanism is not clear. Moreover, information on the metabolic effects of the new PFAS alternative GenX is limited. We examined whether exposure to low-dose PFOA and GenX induces metabolic disturbances in mice, including NAFLD, dyslipidemia, and glucose tolerance, and studied the involvement of PPARα. Methods Male C57BL/6J wildtype and PPARα−/− mice were given 0.05 or 0.3 mg/kg body weight/day PFOA, or 0.3 mg/kg body weight/day GenX while being fed a high-fat diet for 20 weeks. Glucose and insulin tolerance tests were performed after 18 and 19 weeks. Plasma metabolite levels were measured next to a detailed assessment of the liver phenotype, including lipid content and RNA sequencing. Results Exposure to high-dose PFOA decreased body weight and increased liver weight in wildtype and PPARα−/− mice. High-dose but not low-dose PFOA reduced plasma triglycerides and cholesterol, which for triglycerides was dependent on PPARα. PFOA and GenX increased hepatic triglycerides in a PPARα-dependent manner. RNA sequencing showed that the effects of GenX on hepatic gene expression were entirely dependent on PPARα, while the effects of PFOA were mostly dependent on PPARα. In the absence of PPARα, the involvement of PXR and CAR became more prominent. Conclusion Overall, we show that long-term and low-dose exposure to PFOA and GenX disrupts hepatic lipid metabolism in mice. Whereas the effects of PFOA are mediated by multiple nuclear receptors, the effects of GenX are entirely mediated by PPARα. Our data underscore the potential of PFAS to disrupt metabolism by altering signaling pathways in the liver.


INTRODUCTION
Obesity and related metabolic disorders such as non-alcoholic fatty liver disease (NAFLD), type 2 diabetes, and dyslipidemia cause an ever-growing burden on our society [1]. With estimations of more than 39% of the human population being overweight worldwide [2], delineating the causes of these disorders is becoming of utmost importance. While diet, lifestyle, and genetics are well known to contribute to the development of obesity and related metabolic disorders [3], there are growing indications that exposure to certain chemicals in the environment may also play a role. These environmental chemicals originally gained attention due to their capability to interfere with normal endocrine function, hence labeling them as endocrine-disrupting chemicals (EDCs) [4]. Later on, a specific group of EDCs has been defined based on their impact on metabolism [5]. These substances, now commonly defined as obesogens or metabolism-disrupting chemicals, are known to affect metabolic processes within the body and may thereby contribute to obesity, NALFD, and type 2 diabetes [6]. A group of chemicals that has been associated with metabolic disturbances is perfluoroalkyl substances (PFAS). PFAS are man-made chemicals that are extensively used in industrial products due to their high-temperature resistance and water-and dirt-repellent properties. Accordingly, PFAS are present in a wide array of consumer products, including non-stick coatings, food packaging, and firefighting foams. The general structure of PFAS consists of a hydrophobic fluorinated alkyl chain of variable length joined to a hydrophilic end group [7]. PFAS generally exhibit long half-lives, causing them to accumulate both in the environment and the human body [7]. Perfluorooctanoic acid (PFOA), a well-known PFAS, has an estimated serum half-live of 2e4 years in humans [8,9]. Studies have shown that an important molecular target of PFAS, including PFOA, is the nuclear receptor Peroxisome Proliferator-Activated Receptor a (PPARa) [10e12]. PPARs form a group of nuclear receptors that play essential roles in the transcriptional regulation of lipid homeostasis. Three subtypes can be distinguished, consisting of PPARa, PPARb/d, and PPARg, each of which is characterized by a different expression pattern and function. PPARa is particularly important in the liver [13], where it transcriptionally regulates numerous enzymes and factors involved in nearly every branch of lipid metabolism, including fatty acid oxidation, fatty acid uptake, and triglyceride turnover [14,15]. PPARa can be activated by endogenous ligands such as fatty acids and their eicosanoid derivatives, as well as by synthetic agonists such as fibrates [16e18]. PPARa agonists are used in the treatment of dyslipidemia and are being examined for their ability to ameliorate NAFLD, illustrating the importance of PPARa in lipid metabolism [19]. Besides fibrates, PFOA, which structurally resembles fatty acids, also potently activates mouse and human PPARa [10,20,21]. In mouse liver and human HepaRG cells, PFOA induces the expression of numerous PPARa target genes [22e24]. Although there is no doubt that PPARa is a key molecular target of PFOA, there is evidence that PFOA impacts hepatic lipid accumulation in the absence of PPARa, suggesting that additional molecular mechanisms likely play a role in the lipid disturbances triggered by PFOA [22,25]. In line with its ability to activate PPARa and its target genes, rodent data indicate that PFOA influences lipid homeostasis in the liver, which is the major target organ of PFOA [23,26e30]. Specifically, exposure to PFOA has been repeatedly shown to promote hepatic lipid accumulation in vivo [24,28,31,32]. In human HepaRG liver cells, exposure to PFOA also resulted in triglyceride accumulation and caused the downregulation of genes related to cholesterol biosynthesis [33,34]. Despite extensive research, the effects of PFOA on the development of obesity, glucose intolerance, and NAFLD have not yet been investigated. Because of its suspected effects on human health, major efforts have been made to phase out the use of many PFAS. As a result, in 2019, more than 180 countries agreed to ban the production and use of PFOA. However, in response to the ban on PFOA, new replacement chemicals have been produced. 2,3,3,3-Tetrafluoro-2-(heptafluoropropoxy)propanoic acid (HFPO-DA or GenX, referring to the namesake technology) is an example of such a replacement [35]. Currently, there is very limited information on the metabolic effects of GenX in vivo. Recently, it was shown that the treatment of mice with GenX leads to the upregulation of many PPARa targets in the liver [36,37]. However, the overall impact of GenX on the development of obesity, glucose intolerance, and NAFLD has not been investigated. Also, the importance of PPARa in mediating the in vivo effects of GenX remains unclear. Accordingly, the present study aimed to examine the effect of PFOA and GenX on the development of obesity, glucose intolerance, and NAFLD. To that end, we used a model of diet-induced obesity in which mice were given a high-fat diet for 20 weeks concurrent with the provision of PFOA or GenX via drinking water. To investigate the role of PPARa, the experiments were run in parallel in wildtype C57BL/6J mice and their PPARa À/À counterparts.

Animals
Male and female wildtype and PPARa À/À mice that had been backcrossed on a pure C57BL/6J background for more than 10 generations were acquired from Jackson Laboratories (no. 000664 and 008154, respectively). The mice were further bred at the animal facility of Wageningen University under specific pathogen-free conditions to generate the number of mice necessary for the experiments. Animals were housed on a 12 h lightedark cycle with normal bedding and cage enrichment and held at the animal facility of Wageningen University. At 9e11 weeks of age, male wildtype and PPARa À/À mice received either PFOA or GenX via the drinking water while being fed a high-fat diet for 20 weeks (45% kcal fat; D12451, Research Diets, New Brunswick). PFOA (Perfluorooctanoic acid, CAS no. 335-67-1; purity 95%) was purchased from SigmaeAldrich and GenX (2,3,3,3-Tetrafluoro-2-(heptafluoropropoxy)propanoic acid, CAS no. 13252-13-6; purity 97%) was purchased from Synquest laboratories (Alachua FL, US). PFOA was added to the drinking water to a concentration that was calculated to lead to an exposure of 0.05 or 0.3 mg/kg body weight/day. For GenX, a single concentration in the drinking water was used that was calculated to lead to an exposure of 0.3 mg/kg body weight/day. Three different treatment groups were thus included per genotype next to the control group, each containing 12 mice per group, leading to a total of 48 mice per genotype. The mice had ad libitum access to food and drinking water. Body weights, food intake, and water intake were assessed weekly. After 20 weeks of exposure, mice were euthanized in the fed state at Zeitgeber time (ZT)2.5 e ZT4.5. Mice were anesthetized with isoflurane followed by blood collection via orbital puncture. Lean and fat mass was subsequently measured using EchoMRI 100 V (EchoMedical Systems, Houston, TX, USA). Immediately thereafter, the mice were euthanized by cervical dislocation, and tissues were collected. Tissues were weighed, prepared for histological analyses, or snap frozen in liquid nitrogen and subsequently stored at À80 C. The animal study was approved by the central committee on animal experimentation and the local animal welfare committee of Wageningen University (AVD104002015236, 2016.W-0093.020).

Intraperitoneal glucose and insulin tolerance test
Intraperitoneal glucose and insulin tolerance tests were performed after 18 and 19 weeks of treatment, respectively. For the glucose tolerance test, the mice fasted for 5 h, after which blood was collected via tail bleeding for baseline blood glucose measurement (t ¼ 0). Next, the mice received an intraperitoneal injection of 0.8 mg/kg body weight glucose (ThermoFisher Scientific, MA, USA), followed by blood collection via tail bleeding at 15,30,45,60,90,120, and 150 min. For the insulin tolerance test, the mice fasted for 5 h, after which blood was collected via tail bleeding for baseline blood glucose measurement (t ¼ 0). Next, the mice received an intraperitoneal injection of 0.75 U/ kg body weight insulin (Actrapid; Novo Nordisk A/S, Denmark), followed by blood collection at 15, 30, 45, 60, and 90 min. Glucose levels in blood were measured with GLUCOFIX Tech glucometer and glucose sensor test strips (Menarini Diagnostics, Valkenswaard, The Netherlands).

Liver triglycerides and glycogen
For measurement of liver triglycerides, 5% liver homogenates were made in a buffer containing sucrose (250 mM), EDTA (2 mM), Trise base (10 mM) at pH 7.5. Triglycerides were subsequently measured using a commercially available kit (Liquicolor mono) according to the instruction of the manufacturers. To measure glycogen, liver pieces (approximately 50 mg) were dissolved in 10 volumes of 1 M NaOH and incubated at 55 C for 1e2 h. Afterward, an equal volume of 1 M HCL was added, followed by a 5 min centrifugation step at 3,000 RPM. Next, amyloglucosidase (1000 U/ml in 0.2 M sodium acetate 4.8 pH) was added to the sample (1:10 ratio) in order to break down glycogen into glucose. The mixture was incubated for 2 h at 42 C while shaking at 700 RPM, after which the samples were centrifuged shortly. Glucose levels were subsequently measured by the use of a commercially available kit (Glucose GOD FS 10 0 , DiaSys).

Liver histology
Fresh liver tissue was fixed in 4% paraformaldehyde, dehydrated, and embedded in paraffin. Thin sections of the samples were prepared at 5 mm using a microtome and placed onto glass slides followed by overnight incubation at 37 C. Liver sections were stained with hematoxylin & eosin (H&E). To this end, liver tissues were stained in Mayer hematoxylin solution for 10 min and eosin for 10 s at room temperature with intermediate washing in ethanol. The tissues were allowed to dry at room temperature and subsequently imaged using light microscope.

HepaRG experiments
The human hepatic cell line HepaRG was obtained from Biopredic International (Rennes, France) and cultured in growth medium consisting of William's E Medium þ GlutaMAXÔ (Thermofisher Scientific, Landsmeer, The Netherlands) supplemented with 10% Good Forte filtrated bovine serum (FBS; PANÔ Biotech, Aidenbach, Germany), 1% PS (100 U/ml penicillin, 100 mg/ml streptomycin; Capricorn Scientific, Ebsdorfergrund, Germany), 50 mM hydrocortisone hemisuccinate (sodium salt) (SigmaeAldrich), and 5 mg/ml human insulin (PANÔ Biotech). HepaRG cells were seeded in 24-well plates (Corning, Corning, NY; 55,000 cells per well in 500 ml) according to the HepaRG instruction manual from Biopredic International. After 2 weeks on growth medium, cells were cultured for two days in growth medium supplemented with 0.85% DMSO to induce differentiation. Subsequently, cells were cultured for 12 days in growth medium supplemented with 1.7% DMSO (differentiation medium) for final differentiation. At this stage, cells were ready to be used for toxicity studies. Cell cultures were maintained in an incubator (humidified atmosphere with 5% CO2 at 37 C) and the medium was refreshed every 2e3 days during culturing. Prior to the toxicity studies, differentiated HepaRG cells were incubated for 24 h in assay medium (growth medium-containing 2% FBS) supplemented with 0.5% DMSO. Differentiated HepaRG cells were subsequently exposed for 24 h to PFOA and GenX in different concentrations up to 400 mM.
2.6. RNA isolation and quantitative PCR To isolate RNA from the liver, tissues were homogenized using TRIzol reagent (Life Technologies, Bleiswijk, The Netherlands). To isolate RNA from human HepaRG cells, RLT buffer was used. RNA was subsequently isolated and purified by using the RNeasy mini kit from Qiagen (Venlo, The Netherlands). RNA concentration was measured with Nanodrop 1000 spectrophotometer and for subsequent quantitative PCRs, 500 ng RNA was used as input to synthesize cDNA by using iScript cDNA synthesis kit (Bio-rad Laboratories, Veenendaal, The Netherlands). Gene expression was measured by using Sensimix (Bioline, GC Biotech, Alphen aan den Rijn, The Netherlands) on a CFX384 real-time PCR detection system (Bio-Rad Laboratories, Veenendaal, the Netherlands). Gene expression data were normalized to Cyclophilin for mouse tissues and to RPL27 for the HepaRG cells. A list of primer sequences is presented in Table 1.

RNA sequencing
For RNA sequencing on mouse liver, 4 mice per group were used. Total RNA from liver was isolated as stated above. RNA integrity was determined using an Agilent 2100 Bioanalyzer with RNA 6000 microchips (Agilent Technologies, Santa Clara, CA). Library construction and RNA sequencing runs on the BGISEQ-500 platform [38] were conducted at Beijing Genomics Institute (BGI, Hong Kong). At BGI, Genomic DNA was removed with two digestions using Amplification grade DNAse I (Invitrogen, USA). The RNA was sheared and reverse transcribed using random primers to obtain cDNA, which was used for library construction. The library quality was determined using a Bioanalyzer 2100. Thereafter, the library was used for 100bp paired-end sequencing on the sequencing platform BGISEQ-500 (BGI). All the generated raw sequencing reads were filtered by removing reads with adaptors, reads with more than 10% of unknown bases, and low-quality reads. Clean reads were then obtained and stored in FASTQ format.

Processing of RNA sequencing reads
The RNA-seq reads were used to quantify transcript abundances. The tool Salmon [39] (version 1.5.1) was used to map the reads to the GRCm39 mouse genome assembly-based transcriptome sequences as annotated by the GENCODE consortium [40] (release M27). The obtained transcript abundance estimates and lengths were imported in R using the package tximport [41] (version 1.22.0), scaled by average transcript length and library size, and summarized at the gene-level. Differential gene expression was determined using the package limma [42] (version 3.50.0) utilizing the obtained scaled gene-level counts. Briefly, before statistical analyses, nonspecific filtering of the count table was performed to increase detection power [43], based on Table 1 e List with primers for qPCR.

Name
Forward Reverse ATGGCATCCGTTGCAGTTGAT GATGGTCTTCACACCGTTCTC the requirement that a gene should have an expression level greater than 10 counts, i.e. w0.50 count per million reads (cpm) mapped, for at least 4 libraries across all 32 samples. Differences in library size were adjusted by the trimmed mean of M-values normalization method Graphs are presented as mean AE SEM (n ¼ 10e12 mice per group). Asterisks indicate significant differences between treatment vs control *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Hashtags indicate significant differences between wildtype vs PPARa À/À mice within one treatment group ( ## p < 0.01).
Original Article 4 [44], implemented in the package edgeR [45] (version3.36.0). Counts were transformed to log2 (cpm) values and associated precision weights and entered into the limma analysis pipeline [46]. Differentially expressed genes were identified by using generalized linear models that incorporate empirical Bayesian methods [42,47]. Genes were defined as significantly changed when P 0.001 and fold change >1.5. RNA-seq data have been deposited to Gene Expression Omnibus (GEO) under accession number GSE212294.

Biological interpretation of transcriptome data
Changes in gene expression were related to biologically meaningful changes using gene set enrichment analysis (GSEA) [48]. GSEA evaluates gene expression at the level of gene sets that are based on prior biological knowledge, e.g. published information about biochemical pathways or signal transduction routes, allowing more reproducible and interpretable analysis of gene expression data. As no gene selection step (fold change and/or p-value cut-off) is used, GSEA is an unbiased approach. Gene sets were retrieved from the expert-curated KEGG pathway database [49]. Only gene sets comprising more than 15 and fewer than 500 genes were taken into account. The statistical significance of GSEA was determined using 10,000 permutations.

Statistics
Data are presented as mean AE SEM. Statistical significance of treatment versus control was determined by two-way ANOVA with Dunnett multiple comparisons test. For statistical testing of wildtype versus PPARa À/À mice, two-way ANOVA with Sídák's multiple comparisons test was used. A value of p < 0.05 was considered as statistically significant. Data were visualized and analyzed using Prism version 9.0 (GraphPad Software, San Diego, CA, USA).

RESULTS
In the current study, we set out to better understand the potential metabolism-disrupting effects of PFOA and GenX in a mouse model of obesity, glucose intolerance, and NAFLD. To this end, C57BL/6J mice fed a high-fat diet were exposed to PFOA or GenX in the drinking water for 20 weeks ( Figure 1A). Mice were exposed to lower doses of PFOA or GenX than in previous rodent studies in order to better relate the findings to human exposure levels (0.05 and 0.3 mg/kg body weight/ day for PFOA, 0.3 mg/kg body weight/day for GenX [22,24,28,31,50]. In addition, to further explore the role of PPARa in mediating the potential metabolism-disrupting effects of PFOA and GenX, the study was conducted in wildtype and PPARa À/À mice. 3.1. PFOA decreases body weight in wildtype and PPARa À/À mice During the 20 weeks of high-fat feeding, high-dose PFOA treatment significantly reduced weight gain in wildtype and PPARa À/À mice. (Figure 1B, C). By contrast, treatment with GenX only reduced weight gain in the wildtype but not in the PPARa À/À mice. Exposure to PFOA or GenX resulted in a significant reduction in food intake in the wildtype mice, consistent with the anorexic effects of known PPARa agonists [51e53] ( Figure 1D). The changes in body weight in the mice exposed to GenX and high-dose PFOA were accompanied by a significant reduction in fat mass, but not lean mass ( Figure 1E). Liver weights . Asterisks indicate significant differences between treatment vs control (*p < 0.05, **p < 0.01, ***p < 0.001 ****p < 0.0001). Hashtags indicate significant differences between wildtype vs PPARa À/À mice within one treatment group ( # p < 0.05, ## p < 0.01, ### p < 0.001, #### p < 0.0001). were significantly increased in the mice exposed to high-dose PFOA as compared to the control mice, irrespective of genotype (Fig. 1F). To assess if the PFOA and GenX treatment might affect glucose tolerance and insulin sensitivity, glucose and insulin tolerance tests were performed. Overall, PPARa À/À mice displayed increased glucose tolerance as compared to wildtype mice (Figure 2A, B), as well as increased insulin tolerance (Figure 2C, D). In the wildtype and PPARa À/ À mice, high-dose PFOA significantly improved glucose and insulin tolerance. By contrast, the low-dose PFOA and GenX treatments did not significantly impact glucose and insulin tolerance.

High-dose PFOA reduces plasma triglycerides and cholesterol
Because several studies have linked PFAS exposure to changes in levels of plasma triglycerides and cholesterol [9,23,27,50,54], we assessed plasma triglycerides, cholesterol, and other metabolites in the mice exposed to PFOA and GenX. As previously shown [55], ablation of PPARa was in general associated with significantly higher plasma triglycerides, NEFA, and glycerol levels, and significantly lower plasma glucose levels ( Figure 3AeD). Treatment with high-dose PFOA significantly reduced plasma triglyceride levels in the wildtype mice but not the PPARa À/À mice ( Figure 3A). High-dose PFOA also significantly decreased plasma cholesterol levels, which was more pronounced in the PPARa À/À mice than in the wildtype mice ( Figure 3E). Furthermore, high-dose PFOA significantly decreased plasma NEFA and glycerol levels, which was observed in both wildtype and PPARa À/À mice ( Figure 3B, C). By contrast, treatment with lowdose PFOA or GenX did not significantly alter plasma triglycerides, cholesterol, NEFA, and glycerol levels in either wildtype or PPARa À/À mice. None of the treatments significantly changed plasma glucose or b-hydroxybutyrate levels in either wildtype or PPARa À/À mice ( Figure 3D, F). Taken together, these data indicate that treatment with high-dose PFOA but not GenX or low-dose PFOA significantly reduced plasma triglycerides, cholesterol, NEFA, and glycerol levels, which was independent of PPARa.
3.3. Hepatic lipid metabolism is affected in PFOA and GenX treated mice The liver is likely the main target organ of many types of PFAS [22,25,28,32]. As we found increased liver weight in the mice exposed to high-dose PFOA, we next set out to further assess the effects of PFOA and GenX on relevant metabolic parameters in the liver. Liver glycogen levels were elevated in PPARa À/À mice compared to the wildtype mice ( Figure 4A). Next to this, treatment with low-and high-dose PFOA significantly reduced liver glycogen levels, which was seen in both wildtype and PPARa À/À mice. In agreement with previous studies [14,56], liver triglyceride content was significantly higher in untreated PPARa À/À mice than in untreated wildtype mice ( Figure 4B). Exposure to PFOA or GenX significantly increased liver triglyceride content in wildtype mice, which was abolished in PPARa À/À mice ( Figure 4B), suggesting that PPARa mediates the induction in liver triglycerides by PFOA and GenX. H&E staining was performed to examine the histology of the liver in the various treatment groups. In line with the quantitative analysis of triglyceride content, lipid droplet accumulation was higher in untreated PPARa À/À mice than in untreated wildtype mice ( Figure 4C). Treatment with PFOA or GenX noticeably increased lipid accumulation, which was abolished in the PPARa À/À mice. Taken together, these data suggest that exposure to PFOA or GenX increased hepatic lipid accumulation in a PPARa-dependent manner. . Asterisks indicate significant differences between treatment vs control (*p < 0.05, **p < 0.01, ***p < 0.001 ****p < 0.0001). Hashtags indicate significant differences between wildtype vs PPARa À/À mice within one treatment group ( # p < 0.05, ## p < 0.01, ### p < 0.001, #### p < 0.0001). Data are normalized to cyclophilin and expressed relative to wildtype control. Graphs are presented as mean AE SEM (n ¼ 9e12 mice per group). Asterisks indicate significant differences between treatment vs control (*p < 0.05, **p < 0.01, ***p < 0.001 ****p < 0.0001). Hashtags indicate significant differences between wildtype vs PPARa À/À mice within one treatment group ( # p < 0.05, ## p < 0.01, ### p < 0.001, #### p < 0.0001). PFOA is known to activate PPARa [18,20]. Consistent with this notion, high-dose PFOA significantly increased the hepatic expression of the classical PPARa target genes Cd36, Cyp4a14, Ehhadh, and Lpl ( Figure 4D), while low-dose PFOA only increased the expression of Cyp4a14. Similarly, treatment with GenX significantly induced hepatic Cd36, Cyp4a14, Ehhadh, and Lpl mRNA. Remarkably, hepatic mRNA expression of endocrine factor Fgf21 was increased only after treatment with GenX in the wildtype mice. The stimulation of PPARa target gene expression by PFOA and GenX was abolished in the PPARa À/À mice, suggesting that the effects of PFOA and GenX on the above genes are mediated by PPARa.
3.4. PFOA and GenX induce distinct effects on the hepatic transcriptome in PPARa À/À and wildtype mice To obtain a more global view of the effects of PFOA and GenX on hepatic gene expression and to further examine the role of PPARa, we performed RNA sequencing on all groups, using 4 biological replicates per group. First, hierarchical clustering and principal component analysis (PCA) were performed to assess the global transcriptomic changes in wildtype and PPARa À/À mice after exposure to PFOA or GenX ( Figure 5A). As expected, hierarchical clustering and PCA plots revealed clear clustering based on genotype. In addition, whereas the mice that received GenX or low-dose PFOA did not form distinct clusters in either wildtype or PPARa À/À mice, the mice that received high-dose PFOA clustered separately from the other groups, indicating the marked effect of high-dose PFOA on hepatic gene expression. Next, volcano plot analysis was performed to assess the magnitude of the effect of PFOA or GenX treatment on gene expression. In the wildtype mice, the largest effects were observed in the high-dose PFOA group ( Figure 5B), with in total 788 genes significantly changed when applying a threshold of p 0.001 and fold-change >1.5 (489 up, 299 down; Figure 5D). Interestingly, while the overall effect of high-dose PFOA was substantially reduced in PPARa À/À mice ( Figure 5C), 294 genes were still significantly altered by high-dose PFOA in the absence of PPARa (207 up, 87 down; Figure 5E). Of the genes induced by high-dose PFOA in wildtype mice, 88% was dependent on PPARa. Compared to high-dose PFOA treatment, lowdose PFOA only induced limited changes in the hepatic transcriptome in both genotypes (Figure 5BeC). The overall effect of GenX on the hepatic transcriptome in the wildtype mice was intermediate between the high-and low-dose PFOA groups (79 up, 31 down; Figure 5D). Remarkably, no significant gene regulation by GenX was observed in livers of PPARa À/À mice (1 up, 3 down; Figure 5E). Specifically, 99% of the genes induced by GenX in wildtype mice was regulated in a PPARa-dependent manner. Next, we made a heatmap based on significantly regulated genes in the high-dose PFOA group compared to control-treated mice ( Figure 5F). The figure illustrates that high-dose PFOA markedly impacted hepatic gene expression, which was attenuateddbut still clearly visibledin the GenX and low-dose PFOA groups. Consistent with the other analyses, the effects of GenX on gene expression were completely abolished in the PPARa À/À mice ( Figure 5F), whereas the effects of low-and high-dose PFOA were strongly attenuated in the PPARa À/À mice. Similar results were obtained when zooming in on the top 50 upregulated genes by high-dose PFOA ( Figure 5G), many of which are well-established PPARa target genes [57]. Taken together, the gene expression data suggest that, (1) high-dose PFOA markedly influences hepatic gene expression, followed by GenX and low-dose PFOA, (2) the effects of PFOA on hepatic gene expression are predominantly mediated by PPARa, while the effects of GenX are entirely mediated by PPARa.
3.5. High-dose PFOA affects PXR and CAR signaling in the liver in the absence of PPARa Next, we aimed to get more insight into the functional impact of PFOA and GenX on biological pathways by performing gene set enrichment analysis (GSEA) (Figure 6AeC). As expected, pathways positively enriched by high-dose PFOA in wildtype mice were related to PPARa signaling, fatty acid metabolism, and oxidative phosphorylation ( Figure 6A). Similar results were obtained for GenX ( Figure 6B). Importantly, the enrichment scores for the PFOA-induced pathways were much lower in the PPARa À/À mice, indicating strong PPARa dependency ( Figure 6C). A heatmap with the top 50 positively enriched genes in the gene set mPPARa Target Genes is shown in Figure 6D, showing unequivocal PPARa-dependent gene regulation (total core enrichment of 117 out of 150 genes). GSEA was also performed for high-dose PFOA in the PPARa À/À mice. Significant positive enrichment was observed for pathways related to xenobiotic metabolism, steroid hormone biosynthesis, and omega-6 fatty acid metabolism ( Figure 6C). Intriguingly, the enrichment scores for these pathways were much lower in the wildtype than PPARa À/À mice, suggesting that the stimulation of xenobiotic metabolism and steroid synthesis by PFOA is attenuated by the presence of PPARa. A heatmap of the positively enriched genes in the gene set Chemical Carcinogenesis is shown in Figure 6E, showing stronger regulation by PFOA in the PPARa À/À mice than in the wildtype mice. Many of these genes are significantly upregulated by the rodentspecific Pregnane X Receptor (PXR) agonist pregnenolone 16a-carbonitrile (PCN) (GSE136667) [58], as well as by the Constitutive Androstane Receptor (CAR) agonist TCPOBOP (GSE186654) [59], suggesting that they are PXR and CAR target genes. Next to PPARa, Sterol Regulatory Element Binding Proteins have been linked to lipogenic (mainly SREBP-1) or cholesterogenic effects (mainly SREBP-2) of PFAS exposure [34,60,61]. Therefore, we assessed whether PFOA and GenX influenced SREBP-dependent gene regulation ( Figure 6F). PFOA and GenX modestly induced the hepatic expression of SREBP-1 target genes involved in fatty acid synthesis, including Acaca, Acacb, Fasn, Scd1, Elovl5 and Elovl6 [62], whereas PFOA repressed the expression of these genes in PPARa À/À mice. These data suggest PPARa-dependent upregulation and PPARa-independent downregulation of lipogenic genes by PFOA. Consistent with the PPARa specificity of GenX, no effects of GenX were observed in PPARa À/À mice. PFOA slightly downregulated the expression of genes involved in cholesterol synthesis in wildtype and PPARa À/À mice, suggesting PPARa-independent regulation [63]. Our qPCR and RNAseq data indicate that the effects of GenX on hepatic gene expression in mice are entirely mediated by PPARa. To examine if GenX also activates PPARa in human hepatocytes, we treated human HepaRG cells with GenX and studied the expression of the established PPARa target genes FABP4 and PLIN2, which were selected based on their strong sensitivity to PPARa activation in HepaRG cells ( Figure 6G) [64]. GenX dose-dependently stimulated the expression of FABP4 and PLIN2 in HepaRG cells yet was less potent than PFOA at the same concentration. These data suggest that GenX activates human PPARa, yet is a weaker agonist than PFOA.

DISCUSSION
In this paper, we set out to study the impact of PFOA and GenX in a mouse model of diet-induced obesity, glucose intolerance, and NAFLD, and investigate the role of PPARa in mediating the metabolic effects of PFOA and GenX. Previous studies either examined the effect of PFOA in chow-fed mice or did not include PPARa À/À mice. Our research  . Asterisks indicate significant differences between treatment vs DMSO control (*p < 0.05, **p < 0.01, ****p < 0.0001). Hashtags indicate significant differences between PFOA vs GenX ( ## p < 0.01, ### p < 0.001). reveals the major disruptive effects of PFOA and GenX on hepatic and systemic metabolism. Our main findings are: (1) high-dose PFOA improved glucose and insulin tolerance, which was independent of PPARa and likely related to reduced fat mass and body weight. (2) High-dose PFOA significantly reduced plasma triglycerides, cholesterol, NEFA, and glycerol, which except for triglycerides was independent of PPARa. (3) GenX and PFOA increased liver triglyceride levels in a PPARa-dependent manner. (4) The overall magnitude of transcriptome changes in the liver followed the order high-dose PFOA > GenX > lowdose PFOA. (5) 88% of the genes significantly induced by high-dose PFOA were regulated in a PPARa-dependent manner. For GenX, this was 99%, indicating that GenX is a more specific PPARa agonist than PFOA. (6) The PPARa independent effects of PFOA on hepatic gene expression are likely partially mediated by PXR and CAR. Previous studies have shown that in vivo exposure to GenX leads to upregulation of PPARa target genes in rat and mouse livers, strongly suggesting that GenX is a PPARa agonist [36,37,65]. Importantly, we found that the effects of GenX on hepatic gene expression were completely abolished in PPARa À/À mice, indicating that GenX acts exclusively via PPARa. Specifically, 99% of the upregulation of gene expression by GenX in mouse liver was dependent on PPARa. Equally high percentages of PPARa-dependent gene regulation were previously obtained for the highly specific PPARa agonists Wy-14,643 and fenofibrate [66,67]. This implies that the chemical contaminant GenX acts in the same way as certain hypolipidemic drugs used for lowering the risk of cardiovascular disease, although with a weaker agonistic effect. Importantly, GenX also upregulated the PPARa target genes PLIN2 and FABP4 in HepaRG cells, indicating that GenX is also an agonist of human PPARa. Despite activating PPARa, GenX did not influence plasma glucose, cholesterol, triglycerides, glycerol, and NEFA. Previously, we and others observed a decrease in plasma triglycerides and NEFA and an increase in plasma cholesterol upon fenofibrate treatment [68e72]. The reason why these parameters were not changed upon GenX treatment is likely because of the relatively low dose used and because GenX is a comparatively weak PPARa agonist. Consistent with this notion, higher exposure levels of GenX were reported to lower plasma triglyceride levels in rat and mouse dams [65]. Numerous studies have demonstrated that PFOA is a potent PPARa agonist in mice and humans [18,20,73]. In our study, most of the effects of high-dose PFOA on the liver transcriptome were abolished in PPARa À/À mice. Specifically, 88% of the regulation of gene expression by high-dose PFOA in mouse liver was dependent on PPARa. This number is in line with data from Rosen et al., which showed w86% of PPARa-dependent gene regulation in 129S1/SvlmJ mice after 7 days of exposure to 3 mg/kg body weight/day of PFOA [67,74]. Interestingly, similar percentages of PPARa-dependent gene regulation in mouse liver were obtained for in vivo treatment with unsaturated fatty acids Figure 7: Effect of PFOA on metabolic processes in mouse liver. Manually constructed biochemical map of gene expression changes of PPARa-regulated genes (p < 0.001, fold change >1.5) after exposure to high dose PFOA for 20 weeks compared to control [57]. Genes in red indicate upregulation of expression. [66]. A transcriptome map was created of the PFOA-induced changes in expression of PPARa-regulated genes (Figure 7). The map illustrates the profound impact of PFOA on numerous PPARa-dependent pathways, including fatty acid uptake, binding, and activatio;, microsomal, peroxisomal, and mitochondrial fatty acid oxidation; ketogenesis, and triglyceride turnover. Based on comparative analyses of hepatic gene regulation by the PXR and CAR agonists PCN and TCPOBOP, respectively, we deduced that the PPARa-independent gene regulation by PFOA is partially mediated by PXR and CAR. A role of PXR in mediating PPARa-independent gene regulation by PFOA has been previously recognized [21,23,67,75]. In addition, PFOA has been shown to activate CAR in different model systems [21e23, 76,77]. This is also evident in our model, in which the expression of the CAR target gene Cyp2b10 was strongly upregulated by PFOA in both wildtype and PPARa À/À mice. Since PXR and CAR are known to activate partly overlapping sets of genes, it is difficult to disentangle the separate roles of PXR and CAR in mediating the transcriptional effects of PFOA. Intriguingly, similar to observations by Rosen et al. [22], the induction of PXR/CAR targets by PFOA was more pronounced in PPARa À/À mice than in wildtype mice. It can be speculated that this increased sensitivity to PXR/CAR activation in the PPARa À/À mice might be due to a suppressive effect of PPARa on PXR and CAR, as crosstalk between nuclear receptors is known to exist [78]. However, so far there is no firm evidence supporting a suppressive effect of PPARa on PXR or CAR functioning. Alternatively, it is possible that in the absence of PPARa, more PFOA is available to bind and activate PXR/CAR. In contrast to GenX, high-dose PFOA significantly decreased plasma triglycerides, cholesterol, and NEFA levels. Intriguingly, the suppressive effect of PFOA on plasma cholesterol was magnified in the PPARa À/À mice. Taking into account the increased sensitivity to PXR activation in the PPARa À/À mice and considering that PXR activation lowers plasma cholesterol, it can be speculated that PFOA reduces plasma cholesterol via PXR [79]. Concerning NEFA, plasma levels were decreased by high-dose PFOA in the wildtype and PPARa À/À mice, suggesting that the effect is independent of PPARa. Since plasma NEFA levels are reportedly not affected by PXR activation [79], the reduction in plasma NEFA by PFOA is thus likely independent of PPARa and PXR. In contrast, the reduction in plasma triglyceride levels by high-dose PFOA was abolished in the PPARa À/À mice, suggesting that this effect is dependent on PPARa. As indicated above, lowering of plasma triglycerides is a well-known therapeutic effect of synthetic PPARa agonists. Unlike PPARa activation, PXR activation does not seem to influence plasma triglycerides [79]. In contrast to rodent data, which generally show a reduction in plasma triglycerides and cholesterol upon PFAS treatment, epidemiological data mainly show positive associations between exposure to PFAS and total and LDL cholesterol and in some instances triglycerides [9,25,54]. A possible explanation for this apparent discrepancy is that PFAS may activate different receptors in mice and humans. Indeed, while PFOA potently activates both human and mouse PPARa, it only activates mouse PXR/CAR [20,21,34,73,76]. Accordingly, PXR/CAR are unlikely to be involved in mediating the biological effects of PFAS in humans. A conundrum remains, however, because rather than raising plasma triglycerides, PPARa activation in humans lowers plasma triglyceride levels. Accordingly, whether the association between PFAS exposure and serum lipid levels is causal remains unclear. PPARa deficiency is known to be associated with elevated liver triglycerides [14,24,56,66,80,81], which we confirmed in our study. Intriguingly, liver triglycerides were also increased by PFOA or GenX treatment in a PPARa-dependent manner. The results are consistent with other studies showing that different PFAS increase hepatic triglyceride content in wildtype or humanized PPARa mice but not in PPARa À/À mice [28,31,82]. It should be noted, though, that PFOA has also been reported to decrease hepatic triglyceride content in wildtype mice [32]. The latter study differed from ours in that PFOA was given to mice that had already received a high-fat diet for 16 weeks, suggesting that the effect of PFOA on hepatic triglyceride levels may depend on the nutritional/metabolic context. Since PPARa regulates hundreds of genes involved in hepatic lipid metabolism, including many genes involved in fatty acid oxidation, triglyceride storage, and lipolysis, it is difficult to pinpoint the exact mechanism underlying the effect of PFOA on hepatic triglycerides. Surprisingly, low-dose PFOA treatment significantly increased liver triglyceride content but not liver weight or plasma parameters. Although the changes in hepatic gene expression triggered by low-dose PFOA were modest, heatmaps revealed similar trends in expression patterns between low-and high-dose PFOA treatment. This suggests that low-dose PFOA induced relatively subtle expression changes, which upon chronic exposure could lead to more drastic effects, such as increased liver triglycerides. Such a scenario is not unrealistic, as PFAS are highly persistent and bio-accumulate in the body, leading to chronic exposure. SREBPs have been implicated in the effects of PFAS [21,60,61]. In the current study, genes involved in de novo lipogenesis, which are under transcriptional control of SREBP-1, were induced in response to PFOA and GenX. Hence, the triglyceride accumulation in livers of PFOA and GenX-treated mice may be partly explained by increased lipogenesis. Remarkably, we found a downregulation of lipogenic genes by PFOA in the PPARa À/À mice, suggesting crosstalk between SREBP-1 and PPARa, likely involving an upregulation of SREBP-1 by PPARa [62,83]. Accordingly, the stimulation of lipogenic genes by PFOA and GenX might be due to an indirect activation of SREBP-1 via PPARa. Next to SREBP-1, studies also reported the effects of PFOA on SREBP-2 signaling and cholesterol biosynthesis [24,34]. In our study, we only observed a marginal downregulation of SREBP-2 target genes in response to PFOA. An interesting and seemingly paradoxical observation is that both PPARa activation by PFOA and GenX as well as PPARa deficiency led to higher hepatic triglyceride levels. It should be noted, though, that the effects of PPARa activation and deficiency on liver phenotype do not necessarily have to be the opposite. Whereas some genes are strongly induced by PPARa activation but minimally affected by PPARa deficiency, other genes are minimally induced by PPARa activation but strongly suppressed by PPARa deficiency [13]. Depending on the specific role of the altered genes in fatty acid catabolism and triglyceride synthesis/storage, PPARa activation and deficiency might both increase hepatic triglyceride levels, albeit via different mechanisms. Limited research has been conducted on the effect of PFAS on triglyceride levels in the human liver. Recently, Sen et al. found a positive association between PFAS concentrations in serum and NALFD-associated lipid changes in the livers of humans [84]. In human HepaRG cells, high concentrations of PFOA, PFOS, and PFNA increased triglyceride content [34]. By contrast, in primary human hepatocytes, PFOA and PFOS did not significantly alter triglyceride content [85]. Further study on the impact of PFAS on triglyceride content in the human liver is warranted, for example, in hepatocyte humanized mice [86,87]. One of the strengths of the current study is that we exposed mice to relatively low doses of PFOA and GenX. Specifically, the exposure levels of the high-and low-dose PFOA treatment were 10-and 60-fold lower, respectively, than used in most studies. Nevertheless, they still exceed expected human exposure levels based on current intake via food, drinking water, or other sources [88]. For instance, the maximum upper boundary of intake for the group with the highest risk of PFAS exposuredbased on the sum of the most common types of PFAS (PFOA, PFOS, PFNA, PFHxS)dis currently estimated at 96 ng/kg body weight per week [7]. The lowest PFOA concentration applied in the current study would lead to an exposure of 350 mg/kg body weight/week and is thus several orders of magnitude higher. Nevertheless, high variability in PFAS levels in foods exists, which creates a large uncertainty in determining true exposure levels. Indeed, people living in high-risk areas are likely exposed to much higher concentrations of PFAS, with concentrations in drinking water reaching up to 1475 ng/L for PFOA alone [8,89,90]. Moreover, humans are usually exposed to a mixture of different types of PFAS, which could result in higher total PFAS levels.
Our study also has limitations. First, we found a significant decrease in food intake after treatment with PFOA and GenX, which might impact certain metabolic parameters. The reduced food intake is in line with findings from other PPARa agonists, such as oleoylethanolamide, GW-7647, and fenofibrate [51e53]. It should be noted, though, that except for the high-dose PFOA group, the reduction in food intake was not accompanied by a decrease in body weight. Second, we found a significant improvement in glucose and insulin tolerance in wildtype and PPARa À/À mice exposed to high-dose PFOA. However, it is conceivable that the improved glucose homeostasis is a mere reflection of the lower fat mass and thereby body weights in these mice rather than a direct effect of PFOA on glucose homeostasis. Third, we used mice that were deficient in PPARa in all tissues rather than just in the hepatocytes. As PPARa is well expressed in the heart, muscle, intestine, and brown adipose tissue [91,92], it is possible that part of the effects of PFOA and GenX may be conveyed by PPARa in extra-hepatic tissues.
In conclusion, our results show dose-dependent disturbances in hepatic lipid metabolism by PFOA in mice. At a relatively low dose, PFOA increased hepatic triglyceride levels. The metabolic and transcriptomic effects of PFOA were mainly mediated by PPARa, although the involvement of PXR and CAR was also evident. Compared to PFOA, GenX was found to be a less potent but more specific PPARa activator that also raised hepatic triglyceride levels. Our data thus justify the concern about the disruptive effect of PFAS on hepatic and systemic metabolism and stress the need for regulation of these chemicals.

DATA AVAILABILITY
Data will be made available on request.

ACKNOWLEDGMENTS
We would like to thank Shohreh Keshtkar and Karin Mudde (Wageningen University, Wageningen, The Netherlands) for their technical assistance. The project EDCMET has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No. 825762.