Gene Profiling in the Livers of Wild-type and PPARα-Null Mice Exposed to Perfluorooctanoic Acid

Abstract

Health concerns have been raised because perfluorooctanoic acid (PFOA) is commonly found in the environment and can be detected in humans. In rodents, PFOA is a carcinogen and a developmental toxicant. PFOA is a peroxisome proliferator-activated receptor α (PPARα) activator; however, PFOA is capable of inducing heptomegaly in the PPARα-null mouse. To study the mechanism associated with PFOA toxicity, wild-type and PPARα-null mice were orally dosed for 7 days with PFOA (1 or 3 mg/kg) or the PPARα agonist Wy14,643 (50 mg/kg). Gene expression was evaluated using commercial microarrays. In wild-type mice, PFOA and Wy14,643 induced changes consistent with activation of PPARα. PFOA-treated wild-type mice deviated from Wy14,643-exposed mice with respect to genes involved in xenobiotic metabolism. In PFOA-treated null mice, changes were observed in transcripts related to fatty acid metabolism, inflammation, xenobiotic metabolism, and cell cycle regulation. Hence, a component of the PFOA response was found to be independent of PPARα. Although the signaling pathways responsible for these effects are not readily apparent, overlapping gene regulation by additional PPAR isoforms could account for changes related to fatty acid metabolism and inflammation, whereas regulation of xenobiotic metabolizing genes is suggestive of constitutive androstane receptor activation.

Introduction

Perfluorooctanoic acid (PFOA) is a perfluoroalkyl acid (PFAA) with widespread industrial applications. PFOA is primarily used to produce fluoropolymers and fluoroelastomers for the aerospace, automobile, and semiconductor industries but can also be found in a variety of commercial products including food paper coatings, textile surface treatments, cosmetics, lubricants, fire retardants, and nonstick coatings for cookware. Health concerns have been raised because PFAAs, including PFOA, are persistent and widely distributed in the environment (Simcik and Dorweiler, 2005; Yamashita et al., 2005) and can readily be detected in human blood samples (Olsen et al., 2003; Olsen et al., 2005; Calafat et al., 2006; Calafat et al., 2007). PFOA exhibits divergent pharmacokinetics across species and between males and females in some species. In humans, estimates of biological half-life are on the order of years (Burris et al., 2002; Olsen et al., 2007), whereas, in the female rabbit or rat, half-life has been measured in hours (Vanden Heuvel et al., 1991; Hundley et al., 2006).

The toxicity of PFOA has been reviewed previously (Kennedy et al., 2004; Lau et al., 2004; Lau et al., 2007). Acute exposure to PFOA is generally considered to be of low or moderate toxicity, whereas subchronic exposure is associated with hepatomegaly in rodents. Chronic exposure to PFOA has been linked with liver, testicular, and pancreatic tumors in rats (Biegel et al., 2001). Negative teratology studies were reported in the rat and rabbit (Gortner, 1982; Staples et al., 1984), which may reflect the short biological half-life of PFOA in these species. In the mouse, PFOA has been shown to cause deficits in neonatal growth and viability (Lau et al., 2006).

The mechanism associated with PFOA-induced toxicity is not established. PFOA is an activator of peroxisome proliferator-activated receptor alpha (PPARα) (Maloney and Waxman, 1999), and gene expression profiling conducted in either the adult rat or fetal mouse has demonstrated that PFOA induces changes consistent with activation of this ligand-mediated nuclear receptor (Guruge et al., 2006; Martin et al., 2007; Rosen et al., 2007). However, activation of PPARα may not be the only mode of action for PFOA. PFOA has also been shown to modestly activate PPARγ (Vanden Heuvel et al., 2006) and PPARβ/δ (Takacs and Abbott, 2007), two additional PPAR isoforms. Furthermore, the binding of PFOA to yet other members of the nuclear receptor superfamily of transcription factors has not been fully addressed. The hypothesis that PFOA may have PPARα-independent effects is supported by the finding that liver enlargement can be induced in the PPARα-null (knockout) mouse by PFOA exposure (Yang et al., 2002; Abbott et al., 2007) as well as by the observation that wild-type mice fed PFOA have fatty livers, an effect not observed for other PPARα agonists (Kudo and Kawashima, 1997).

In recent years, the PPARα-knockout mouse (Lee et al., 1995) has been used to investigate the molecular mechanisms of compounds that function as peroxisome proliferators and to examine the role of PPARα in both nutrition and disease. Knockout mice also provide a useful model to characterize transcriptional changes that are independent of PPARα. A better understanding of the role of both PPARα and non-PPARα-related events as they relate to PFOA-induced toxicity is an important aspect of human health risk assessment because the relevance of PPARα to human toxicity, at least in terms of liver tumor formation, has been questioned (Cattley et al., 1998; Klaunig et al., 2003).

In the current study, wild-type and PPARα-knockout adult male mice were exposed to either PFOA or Wy14,643, an established PPARα agonist. Hepatic gene expression was then evaluated using full-genome expression microarrays. Additional liver tissue was examined for morphological effects by both light and electron microscopy as well as for changes in cell proliferation (see Wolf et al., 2008, this issue). Our working hypothesis was that non-PPARα-dependent changes would be apparent in knockout mice exposed to PFOA. These data could then be used to characterize the molecular mechanisms related to PFOA-induced toxicity.

Materials and Methods

Animals and Dosing

Studies were approved by the U.S. EPA ORD/NHEERL Institutional Animal Care and Use Committee. The procedures and facilities used followed the recommendations of the 1996 NRC “Guide for the Care and Use of Laboratory Animals,” the Animal Welfare Act, and the Public Health Service Policy on the Humane Care and Use of Laboratory Animals. PPARα-null mice (129S4/SvJae-Ppara ^tm1Gonz/J, stock #003580) and wild-type mice (129S1/SvlmJ, stock #002448) were originally purchased from The Jackson Laboratory (Bar Harbor, ME) and maintained as an inbred colony on the 129/Sv background at the U.S. EPA, Research Triangle Park, NC. Knockout and wild-type male mice used in the study were housed 4 per cage and were allowed to acclimate for a period of 1 week prior to the start of the study. Food (LabDiet 5001, PMI Nutrition International, St. Louis, MO) and municipal tap water were provided ad libitum. Animal facilities were controlled for temperature (20°C–24°C) and relative humidity (40%–60%) and kept under a 12-h light-dark cycle. Mice at 10–12 weeks of age were dosed by gavage for 7 consecutive days with either 0, 1, or 3 mg/kg PFOA (ammonium salt, catalog #77262, Fluka Chemical, Steinheim, Switzerland) in deionized water, or 0 or 50 mg/kg Wy14,643 (catalog #C7081, Sigma-Aldrich Co., St. Louis, MO) in 0.5% methylcellulose. This dosing paradigm was based on published data and represented conditions in which steady-state serum levels would be achieved for PFOA (Lau et al., 2006). The dose level for Wy14,643 was also based on previously published data (Anderson et al., 2004). With the exception of hepatomegaly, these doses were below those required to induce acute toxicity. All dosing solutions were freshly prepared each day. At the end of the dosing period, animals were euthanized by CO₂ asphyxiation and tissue was collected from the left lobe of the liver for preparation of total RNA and histopathology (see Wolf et al., 2008, this issue).

RNA Preparation

Collected tissue was immediately homogenized in TRI reagent (Sigma Chemical, St. Louis, MO) and processed on the same day through alcohol precipitation according to the manufacturer’s directions. RNA pellets were then washed in cold 80% ethanol and stored at –80°C until further use. Following resuspension in nuclease-free water (Ambion, Austin, TX), the RNA was quantified and evaluated for purity (260 nm/280 nm ratio) using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE). 100 μg of each sample was then further purified using RNeasy spin columns (Qiagen, Valencia, CA) according to the manufacturer’s directions. Approximately 250 ng of each sample was evaluated for quality using a 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). RNA samples were used only if they were found to have an RNA Integrity Number (RIN, 2100 Expert software, version B.01.03) of at least 8.5 (Imbeaud et al., 2005).

Gene Expression Profiling

Transcript analysis was conducted using Applied Biosystems Mouse Genome Survey Microarrays (Applied Biosystems, Foster City, CA). These broad-expression single-channel microarrays use spotted 60-mer oligonucleotide probes directed at approximately 32,000 curated genes from the mouse genome. Detection of digoxigenin-labeled target cRNA is based on chemiluminescence. In addition to transcript-specific probes, each array feature is co-spotted with a 24-mer oligonucleotide probe that is complementary to a fluorescent-labeled hybridization control target. The fluorescent signal, which is independent of the gene expression signal, then provides information for grid alignment, per spot normalization, and spatial normalization as conducted by the array imaging software (Expression Array System Software version 1.1.1, Applied Biosystems).

For each sample, labeling and hybridization reagents were prepared using master mixes at each step to minimize technical error. Digoxigenin-labeled cRNA was synthesized from 500 ng total RNA using a protocol and labeling kit obtained from the microarray manufacturer, which included one round of T7 amplification (Nanoamp RT-IVT labeling kit, Part #4365715, Applied Biosystems). Hybridization of target cRNA, labeling controls, and hybridization controls was then conducted at 55°C for 16 h in a shaking incubator (Model C24, New Brunswick Scientific, Edison, NJ) using a vendor-provided protocol and reagents (Chemiluminescence Detection Kit, Part #4342142, Applied Biosystems). Microarray images were obtained using an Applied Biosystems 1700 Chemiluminescent Microarray Analyzer (Part #4338036). Output from AB1700 Analyzer and Expression Array System Software included nonnormalized probe signal and microarray QA metrics (S/N ratio, average probe signal, median background, number of genes detected, and data for spike-in and chemiluminescent controls). Data are available through the Gene Expression Omnibus at the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/geo) as accession number GSE9796.

Experimental Design and Analysis

Four biological replicates consisting of individual RNA samples were included in each dose group with the exception of the Wy14,643 control group where only 3 animals were available for analysis. All samples were labeled and hybridized to microarrays using a balanced block design with one biological replicate per group included in each block. Data were quantile-normalized using package ab1700 in Bioconductor (http://www.bioconductor.org/) and filtered to remove nonexpressed genes using a S/N threshold ≤3 (ratio of signal to signal standard deviation) as recommended by the array manufacturer (Applied Biosystems 1700 Chemiluminescent Microarray Analyzer User Bulletin, Part #4367370 Rev. A). Expression data were then analyzed using a multivariate ANOVA model in SAS version 9.1 (Cary, NC). When the main effect for treatment was found to be significant (p ≤0.05), t test of the least square mean was used as a post hoc test to evaluate an individual treatment mean versus its concurrent control (p ≤0.0025). Significant transcripts were evaluated for relevance to canonical pathway and biological function using Ingenuity Pathways Analysis (Ingenuity Systems, www.ingenuity.com) in combination with the Panther Classification System (http://www.pantherdb.org). Data reporting stringency for an individual observation was potentially relaxed (p ≤0.03) to include additional genes within an enriched canonical pathway or functional group if consistent trends were observed across treatment groups. When multiple probes existed for a given transcript, the data point selected for analysis was based on minimum p value. Principle component analysis was conducted in Genespring GX (Agilent Technologies, Palo Alto, CA), which required an additional per gene normalization to the median in order to scale the data for graphing. Hierarchical clustering and heat maps were generated using Eisen Lab Cluster and Treeview software (http://rana.lbl.gov/EisenSoftware.htm).

Evaluation of Selected Genes by Real-Time RT-PCR

Based on the results from the microarray experiment, real-time RT-PCR (reverse transcription polymerase chain reaction) analysis of selected genes was conducted using the same RNA samples utilized during the microarray analysis. 2 μg total RNA was initially digested using 2 units DNaseI (#M6101, Promega Corporation, Madison, WI) for 30 min at 37°C followed by 10 min at 65°C in a buffer containing 40 mM Tris (pH 8.0), 10 mM MgSO₄, and 1 mM CaCl₂. The RNA was then quantified using a Quant-iT RiboGreen RNA assay kit according to the manufacturer’s protocol (#R11490, Invitrogen Corporation, Carlsbad, CA) and approximately 1.5 μg RNA reverse transcribed using a High Capacity cDNA Archive Kit according to the provided protocol (#4322171, Applied Biosystems). Amplification was performed on an Applied Biosystems model 7900HT Fast Real-Time PCR System in duplicate using 25 ng cDNA and TaqMan Universal PCR Master Mix (#4304437, Applied Biosystems) in a total volume of 12 μl. Beta-2-microglobulin (B2m, Entrez #12010), which was uniformly expressed among all samples (cycle threshold standard deviation less than 0.5), was used as an endogenous reference gene. The following TaqMan assays (Applied Biosystems) were included in the study: B2m (#Mm00437762_m1), Ccnd1 (#Mm00432359 _m1), Gadd45b (#Mm00435123_m1), Peci (#Mm00478725_m1), Ehhadh (#Mm00470091_s1), Cyp3a41 (#Mm00776855_mH), Pdk4 (#Mm00443325_m1), Slc01c1 (#Mm0045185), Cyp3a11 (#Mm00435123_m1), Cyp1a1 (#Mm00487218_m1), Cyp2a5 (#Mm00487248_g), Por (#Mm00435876_m1), Sult1a1 (#Mm 00467072), Pxmp4 (#Mm00480657), Ech1 (#Mm00469322), and Cyp2b10 (#Mm00456591_m1). Fold change was calculated using the 2^−ΔΔCT method of Livak and Schmittgen (2001).

Results

Hepatomegaly was observed in wild-type mice exposed to either Wy14,643 or PFOA, as well as in PFOA-treated knockout mice. This was accompanied by peroxisome proliferation and increased cell proliferation in wild-type mice but not in knockout mice at those doses used to evaluate gene expression. No relevant histopathological changes were observed in Wy14,643-exposed knockout mice. A thorough analysis of the histopathology data is available in this issue (see Wolf et al., 2008).

All microarrays were found to be of high quality based on visual evaluation of each array image in addition to quality control metrics established by the microarray manufacturer (Applied Biosystems 1700 Chemiluminescent Microarray Analyzer User Guide, Part #4338852 rev. B). These included average signal, average S/N, median background, number of genes detected (41%–52% across all arrays), and number of failed genes, in addition to output from spike-in-controls for labeling, hybridization, and chemiluminescence.

The number of significant (p ≤ 0.0025) and fully annotated genes used to evaluate the data for relevance to canonical pathway or biological function is shown in Table 1. In terms of the total number of genes altered, the effect of 1 mg/kg PFOA in wild-type mice was less robust than that observed at 3 mg/kg but was nevertheless similar in response since 85% of the genes altered in the lower dose group were also significantly changed in the higher dose group (p ≤ 0.0025). In addition, the number of differentially expressed genes was similar in wild-type mice exposed to either Wy14,643, the PPARα-positive control, or 3 mg/kg PFOA; therefore, the highest PFOA dose group was considered the most useful for comparison of data across treatment groups.

As expected, putative PPARα target genes such as Acox1, Me1, Ehhadh (Bien), Slc27a1 (Fatp1), Hsd17b4, Hadha, Hadhb, and Pdk4 (Motojima et al., 1998; Akiyama et al., 2001; Mandard et al., 2004; Tamura et al., 2006; Rakhshandehroo et al., 2007) were found to be up-regulated in wild-type mice exposed to either Wy14,643 or PFOA but were unchanged in treated knockout mice (Figure 1). When significant genes (p ≤ 0.0025) were clustered across treatment, changes observed in wild-type mice were generally consistent across the treatment groups (Figure 2). It was also apparent that while the majority of PFOA-related effects were mediated through PPARα, non-PPARα-related effects were also evident in PFOA-exposed knockout animals. This was in contrast to Wy14,643 where minimal effects on knockout mice were observed compared to the knockout controls. Principle components analysis performed across treatment groups using only those genes that were significant for at least one treatment contrast (p ≤ 0.0025) further suggested that the response of the knockout animals exposed to the highest dose of PFOA shared some similarity with treated wild-type animals (Figure 3).

Gene Expression Changes in Wild-Type Mice

Although significant overlap in the data was observed between wild-type mice exposed to either 50 mg/kg Wy14,643 or 3 mg/kg PFOA, only approximately 50% of the altered genes were changed in both groups (p ≤ 0.0025). The primary effect of either Wy14,643 or PFOA on wild-type mice was on genes associated with fatty acid metabolism (Figure 4), an effect consistent with activation of PPARα (Mandard et al., 2004; Lefebvre et al., 2006). Additional outstanding alterations included genes related to inflammation (complement/coagulation cascades) (Figure 5) and xenobiotic metabolism (Figure 6). Altered expression of genes associated with cell cycle control (Figure 7), peroxisome biogenesis (Figure 8), and proteasome structure/organization (Figure 9) was evident in both treatment groups as well, although changes observed in the latter two functional categories were especially robust in PFOA-treated mice. In addition, gene expression changes related to cholesterol biosynthesis were also observed, but this was largely limited to wild-type mice exposed to Wy14,643 (Figure 10). Even though both treatments altered the expression of xenobiotic metabolizing genes, this response differed between the Wy14,643 and PFOA-exposed wild-type mice. In particular, there was notable up-regulation of Cyp2b and Cyp2c genes as a result of PFOA exposure (Figure 6).

Both Wy14,643 and PFOA increased hepatic cell proliferation in wild-type mice (see Wolf et al., 2008), and among the genes up-regulated by either compound were Ccnd1 (Cyclin D1) and cJun (Figure 7). Ccnd1 is a cyclin gene involved in progression of cells through the G₁ phase of the cell cycle, whereas cJun, a component of AP-1, has been shown to positively influence hepatocyte proliferation during liver regeneration (Stepniak et al., 2006) and is a potential upstream regulator of Ccnd1 (Wisdom et al., 1999). Additional up-regulated genes in these mice included Cdkn1a (p21) and Gadd45b. While such genes have been associated with cell cycle inhibition (Hengst and Reed, 1998; Vairapandi et al., 2002), increased expression of Cdkn1a has previously been observed in proliferating fibroblasts (Li et al., 1994) while Gadd45b was found to be up-regulated in the regenerating liver (Xu et al., 2007) and has been shown, under certain conditions, to be an inhibitor of apoptosis (De Smaele et al., 2001).

Gene Expression Changes in PPARα-Null Mice

While few changes were observed in PPARα-null mice exposed to Wy14,643, differential expression of a moderate number of transcripts was observed in knockout mice as the result of exposure to PFOA (Table 1). Some of these transcripts were similarly altered in both wild-type and knockout animals, including genes associated with fatty acid metabolism (Figure 4), inflammation (Figure 5), xenobiotic metabolism (Figure 6), and cell cycle control (Figure 7). As in wild-type mice, up-regulation of certain Cyp2b and Cyp2c genes such as Cyp2b10, a marker of the constitutive androstane receptor (CAR) (Honkakoski et al., 1998; Kawamoto et al., 1999; Wei et al., 2000), was observed (Figure 6) as were changes related to cell cycle regulation including Ccnd1, cJun, and Gadd445b (Figure 7), although, in contrast to wild-type mice, there was no effect on cell proliferation in knockout mice (see Wolf et al., 2008). Baat (entrez# 12012), a free bile acid conjugating enzyme (Falany et al., 1997) and a transcript shown to be positively regulated by the farnesoid X receptor (FXR) (Pircher et al., 2003) but negatively regulated by PPARα (Solaas et al., 2004), was down-regulated by PFOA (p ≤ 0.0025) in both wild-type and knockout animals (data not shown). Additional robust changes (greater than twofold, p ≤ 0.0025) observed in either wild-type or knockout animals as a result of PFOA exposure included increased expression of both Igf1 (entrez# 16000) and Angpt14 (entrez# 57875) in addition to decreased expression of Prlr (entrez# 19116) (data not shown).

Real-Time RT-PCR of Selected Genes

There was good agreement between the microarray and RT-PCR results among those genes selected for real-time PCR analysis (Figure 11). Differences were noted, however, with respect to both Ehhadh and C4a. Ehhadh, a PPARα regulated gene (Akiyama et al., 2001), which, based on the microarray data, was markedly up-regulated exclusively in treated wild-type mice (greater than 20-fold following either Wy14,643 or 3 mg/kg PFOA), also demonstrated increased expression in PFOA-exposed knockout mice as indicated by RT-PCR. In addition, down-regulation of C4a could not be readily demonstrated by RT-PCR since the deviation from control in either treatment group was less than one cycle threshold, although the trend in C4a expression was similar in both assays. Cyp3a11, an established marker gene of the pregnane X receptor (PXR) (Kliewer et al., 1998; Xie et al., 2000) and a gene that did not appear on the Applied Biosystems microarray, was up-regulated in PFOA-treated knockout mice but not in wild-type mice exposed to either PFOA or Wy14,643. Based on PCR cycle threshold data, Cyp2b10 was expressed in the livers of control and Wy14,643-treated mice as a low copy number gene; therefore, fold change data were reported only for PFOA-treated mice where clear induction was observed. The expression of both Sult1a1 and Slc01c1, also known as Oatp2, was unchanged across all treatment groups.

Discussion

It is clear that many of the transcriptional changes induced by PFOA are mediated through PPARα activation. It was also apparent from the current data that PFOA can alter the expression of genes related to fatty acid metabolism, inflammation, xenobiotic metabolism, and cell cycle progression independently of PPARα. In contrast, the PPARα agonist Wy14,643 was found to have only minimal effects on gene expression in the PPARα-null mouse. Thus, PFOA has multiple modes of action and is capable of functioning as a biologically active xenobiotic in the absence of PPARα signaling.

Regulation of lipid metabolism by PPARα is well described (Desvergne and Wahli, 1999; Hihi et al., 2002; Mandard et al., 2004; Lefebvre et al., 2006), as is the role of PPARα in modifying the inflammatory response (Corton et al., 1998; Delerive et al., 2001; Cuzzocrea et al., 2006). Therefore, the finding that PFOA can regulate genes within these functional groups independently of PPARα was not anticipated. The explanation is not readily apparent; however, it is possible that other members of the nuclear receptor superfamily of transcription factors were involved. For example, PPARγ and PPARβ/δ have been shown to maintain functions that are distinct as well as overlapping with PPARα. While both PPARγ and PPARβ/δ can be found in the liver, PPARγ is predominately expressed in adipose tissue where it plays a role in adipogenesis and lipid storage (Chawla et al., 1994; Tontonoz et al., 1994). On the other hand, PPARβ/δ, the predominate PPAR isoform in skeletal muscle (Muoio et al., 2002), is more broadly expressed and has been shown to regulate fatty acid catabolism in skeletal muscle and adipose tissue (Muoio et al., 2002; Dressel et al., 2003; Holst et al., 2003; Tanaka et al., 2003; Wang et al., 2003) as well as having a role in modifying the inflammatory response in macrophages (Lee et al., 2003), cardiomyocytes (Ding et al., 2006), and endothelial cells (Rival et al., 2002). The function of PPARβ/δ in the liver, however, has not been determined, and evidence for compensation by PPARγ has been reported in the PPARα-null mouse fed a high-fat diet (Patsouris et al., 2006). Interestingly, hepatic peroxisome proliferation, a cellular response normally linked to activation of PPARα, has been observed in PPARα knockout mice treated with either a PPARγ agonist or a PPARγ/β/δ mixed agonist but not Wy14,643, thus supporting the concept that a functional overlap may exist between the various PPAR isoforms (DeLuca et al., 2000). More recently, peroxisome proliferation was reported in PPARα-null mice treated with fenofibrate, a drug thought to function primarily through transactivation of PPARα, further suggesting that a non-PPARα mechanism exists for peroxisome regulation in the murine liver (Zhang et al., 2006). In the current study, up-regulation of genes in knockout mice related to peroxisome biogenesis or function such as Pxmp4, Ech1, and Peci suggested that peroxisome function may have been altered by PFOA, although morphological changes associated with peroxisome proliferation were not observed in these mice (see Wolf et al., 2008).

Transient transfection assays have further supported the notion that PFOA has the potential to modify gene expression via activation of either PPARγ or PPARβ/δ, although, at reduced effectiveness compared to PPARα. Vanden Heuvel et al. (2006) demonstrated modest activation of murine PPARγ but not PPARβ/δ by PFOA, whereas Takacs and Abbott (2007) more recently reported that PFOA weakly activates murine PPARβ/δ but not PPARγ. Neither study, however, demonstrated activation of either human PPARγ or PPARβ/δ by PFOA.

Although activation of PPARα has been shown to influence the expression of certain hepatic xenobiotic metabolizing genes (Fan et al., 2004), the influence of PFOA on genes involved in phase I metabolism, including Cyp2b10, Cyp2c55, Cyp2b13, and Por (Cytochrome P-P450 reductase), may be unrelated to PPARα transcriptional control. In the current study, these changes were found in both wild-type and knockout animals and shared little similarity with patterns observed in mice treated with Wy14,643. In recent years, the mechanism associated with the control of xenobiotic metabolism has become increasingly clear. Nuclear receptors such as the aryl hydrocarbon receptor (AhR), CAR, PXR, and the nuclear factor-erythoroid derived 2-like 2 (Nrf2) have been shown to play an important role in regulating phase I and II metabolizing enzymes as well as phase III transporters (Handschin and Meyer, 2003; Xu et al., 2005). Hence, it is possible that one or more of these transcription factors may be involved in the xenobiotic response to PFOA. As a result, the involvement of CAR/PXR is actively being investigated by our group as well as others (Elcombe et al., 2007). Indeed, hepatomegaly, which is observed in the PFOA-treated PPARα knockout mouse (Yang et al., 2002; Abbott et al., 2007), is also a feature of CAR transactivation (Huang et al., 2005), although in the case of PFOA, such effects may not be related to enhanced cell proliferation (see Wolf et al., 2008). Furthermore, while it is established that hepatic inflammation can reduce nuclear receptor–mediated cytochrome P450 gene expression (Morgan, 2001), the reciprocal may also be true since activation of PXR in the mouse or the steroid and xenobiotic receptor (SXR), the human ortholog of PXR, has been shown to negatively regulate inflammation in various tissues via a pathway linked to NF-kB (Zhou et al., 2006). Thus, a connection may exist between the observed changes in xenobiotic metabolizing genes and those genes related to inflammation.

In the current study, exposure to PFOA resulted in up-regulation of Cyp2b10, a typical CAR marker gene (Honkakoski et al., 1998; Kawamoto et al., 1999; Wei et al., 2000), as well as various other CAR-responsive genes including Por, Gadd45b, and Ccnd1 (Ledda-Columbano et al., 2000; Ueda et al., 2002; Columbano et al., 2005) indicating the involvement of CAR. On the other hand, additional CAR-regulated genes such as Cyp1a1 (Maglich et al., 2002), Sult1a1 (Lee et al., 2007), and Cyp3a11 (Maglich et al., 2002; Ueda et al., 2002) remained generally unchanged, although altered Cyp3a11 expression was observed in knockout animals. Moreover, the lack of consistent increases in Cyp1a1 and Cyp3a11 expression argued against activation of AhR and PXR, respectively, since Cyp1a1 is also recognized as an AhR marker gene (Whitlock, 1999), while Cyp3a11, along with other Cyp3a genes, has been shown to be regulated by PXR (Kliewer et al., 1998; Xie et al., 2000). Transcriptional control of xenobiotic metabolism, however, is complex and involves crosstalk between multiple nuclear receptors (Pascussi et al., 2007). Hence, the role of CAR and other nuclear receptors in PFOA-mediated toxicity remains unclear.

Given that PFOA is capable of inducing hepatomegaly in PPARα-null mice, it was not surprising that changes in hepatic gene expression were observed in treated knockout animals. Whether or not certain non-PPARα-related modes of action contribute to the toxicity of PFOA is a critical question. Although PFOA is capable of inducing hepatomegaly in knockout mice, increased liver weight is not necessarily related to either cell proliferation or peroxisome proliferation (see Wolf et al., 2008). In addition, activation of PPARα has been shown to be a requirement for PFOA-induced developmental toxicity in the mouse (Abbott et al., 2007), although this does not rule out the possibility that PPARα-independent changes may also be involved. A better understanding of the mode of action of PFOA, as well as that of other PFAAs, would be an important contribution to human risk assessment since chronic hepatic activation of PPARα is arguably less relevant in humans compared to rodents (Cattley et al., 1998; Klaunig et al., 2003; Morimura et al., 2006).

In conclusion, unlike the PPARα agonist Wy14,643, PFOA is capable of inducing effects independently of PPARα. Genes altered in the PPARα-null mouse following exposure to PFOA included those associated with fatty acid metabolism, inflammation, xenobiotic metabolism, and cell cycle progression. The specific signaling pathway(s) responsible for these effects is not readily apparent, but it is conceivable that other members of the nuclear receptor superfamily such as PPARβ/δ and CAR may be involved.