The Epoxygenases CYP2J2 Activates the Nuclear Receptor PPARα In Vitro and In Vivo

Background Peroxisome proliferator-activated receptors (PPARs) are a family of three (PPARα, -β/δ, and -γ) nuclear receptors. In particular, PPARα is involved in regulation of fatty acid metabolism, cell growth and inflammation. PPARα mediates the cardiac fasting response, increasing fatty acid metabolism, decreasing glucose utilisation, and is the target for the fibrate lipid-lowering class of drugs. However, little is known regarding the endogenous generation of PPAR ligands. CYP2J2 is a lipid metabolising cytochrome P450, which produces anti-inflammatory mediators, and is considered the major epoxygenase in the human heart. Methodology/Principal Findings Expression of CYP2J2 in vitro results in an activation of PPAR responses with a particular preference for PPARα. The CYP2J2 products 8,9- and 11-12-EET also activate PPARα. In vitro, PPARα activation by its selective ligand induces the PPARα target gene pyruvate dehydrogenase kinase (PDK)4 in cardiac tissue. In vivo, in cardiac-specific CYP2J2 transgenic mice, fasting selectively augments the expression of PDK4. Conclusions/Significance Our results establish that CYP2J2 produces PPARα ligands in vitro and in vivo, and suggests that lipid metabolising CYPs are prime candidates for the integration of global lipid changes to transcriptional signalling events.


Introduction
Exogenous PPAR activators include a number of fatty acids as well as a variety of eicosanoid, HETEs, HODEs, prostaglandins, and leukotrienes. A number of lipid-metabolising pathways have therefore been suggested as sources of PPAR ligands, however none really fully satisfy the criteria required for them to be regarded as ubiquitous endogenous PPAR ligand generators [1,2]. The cyclooxygenase, and 5-, 12/15-lipoxygenase pathways are good examples: prostanoid synthase enzymes and lipoxygenase isoforms have a highly tissuespecific expression pattern that do not fully match those of the PPARs and the effects of prostanoid/lipoxygenase enzyme inhibitors or the phenotypes of the corresponding knockout animal do not match those of the PPARs [1]. Phospholipases [3] or lipoprotein lipase [4] can produce PPARa ligands from circulating lipoproteins. However, it is unclear whether these enzymes produce PPAR ligands universally. A very attractive hypothesis is that cytochrome P450 enzymes (CYPs) could provide the link. Similar to related eicosanoids, 8,9-, 11,12-, and 14,15-EET and their CYP4A hydroxylase metabolites can bind and activate a PPARa reporter gene [5], and 8,9-, 11,12-and 14,15-EETs can functionally activate both PPARa [6] and PPARc [7,8] in vitro. It is not known however, which CYPs act as potential sources of the EETs, or whether CYPs or EETs mediate any functional effects on PPARs in vivo.
There are more than 500 CYP genes primarily associated with the metabolism and detoxification of foreign chemicals. A number of CYPs also catalyze the metabolism of lipids by epoxygenases lipoxygenase-like, and vand v-1-hydroxylase activities [9]. The CYP2 gene family of epoxygenases has approximately 25 members. CYP2J2 is the only CYP2J family member expressed in man, and it is localised in the heart and vasculature, throughout the gastro-intestinal and respiratory tracts and in the kidney [9,10], where it catalyses the conversion of arachidonic acid via the epoxygenase pathway to anti-inflammatory and vascularprotective EETs [10]. Here we show CYP2J2 activates PPARa in vitro and in vivo.

CYP2J2 activates PPARs in vitro in an autocrine manner
Transient transfection of the CYP2J2 cDNA in HEK293 cells produced significant expression of CYP2J2 protein ( Figure 1A). The combination of CYP2J2 with PPARa ( Figure 1A), PPARd or PPARc ( Figure 1B) induced a synergistic activation of PPAR reporter genes, with a marked preference in terms of absolute activity for PPARa activation ( Figure 1A). pDR-1 was used as a reporter gene for PPARd activation due to the reported lack of efficacy for pACO on PPARd responses [11]. A functional PPAR was required for this activation, as no significant reporter gene activation was seen in cells co-transfected with vector reporter gene lacking the PPRE (data not shown), or when cells were co-transfected with dominant-negative (DN)-PPARa [12; Figure 1A). Similarly, the activation of PPAR reporter gene by co-transfection of PPARa and CYP2J2 required active CYP2J2, as the epoxygenase inhibitor SKF525A caused a concentrationdependent inhibition of PPARa-CYP2J2 induced PPAR reporter gene activation ( Figure 2). These endogenous products of CYP2J2 act in an intracellular manner, as only when cells are cotransfected so that PPARa and CYP2J2 are co-expressed together in the same cell is a significant synergistic activation of the PPAR reporter gene detected (data not shown).

CYP2J2 activates PPARa and inhibits NFkB activation
PPARa activation inhibits the activation of the pro-inflammatory and survival transcription factor NFkB [1,2]. IL-1b (10 ng/ml) induced NFkB reporter gene activation in HEK293 cells transfected with control plasmid cDNA. In cells transfected with the combination of CYP2J2 and PPARa, IL-1b induced NFkB activation was completely abolished (Figure 3). Inhibiting CYP2J2 with SKF525A (10 mM) restored the ability of IL-1b to activate NFkB in PPARa and CYP2J2 transfected cells  Figure 5A) had no effect on PPARa reporter gene activation. Although, 14,15-EET had no effect in our hands, consistent with the previous report [5], the CYP4A hydroxylase 14,15-EET metabolite was a potent PPARa activator ( Figure 5B). The activation of PPARa responses by 8,9-EET or 11,12-EET was completely reversed when cells were co-transfected with dominant negative DN-PPARa ( Figure 4B). PPARa activation induces PDK4 in cardiac tissue in vitro PDK4 is a tissue specific PPARa target gene that facilitates fatty acid oxidation by ''sparing'' pyruvate for oxaloacetate formation [12,13]. The highly selective PPARa ligand GW7647 induced PDK4 mRNA in mouse cardiac tissue in culture in vitro ( Figure 6); an effect which was abolished by co-incubation with the selective PPARa antagonist GW6471 ( Figure 6), or if tissue was used from PPARa knockout mice (data not shown).

Cardiac-specific CYP2J2 transgenic mice have an elevated PPARa response during fasting
The fasting response is a model of PPARa activation in vivo as a decline in insulin levels and/or a rise in lipid fuel availability facilitates PPARa activation and the up-regulation of PDK4. Moreover, this marked up-regulation of PDK4 expression in response to fasting is absent in PPARa knockout mice [14]. Therefore, PDK4 is a robust index of PPARa activation in vivo. Expression of the related proteins PDK1, PDK2 and E1a are not regulated by PPARa, and were used as controls.
Cardiomyocyte-specific CYP2J2 transgenic (Tr) mice have been generated and have a normal heart anatomy and contractile function [15]. Fed CYP2J2 Tr mice had no altered expression of PDK1, -2, -4 or E1a expression in the heart, kidney, or liver compared to wild type controls ( Figure 6; and data not shown). In fasted mice, PDK4 protein expression was selectively up-regulated in the heart ( Figure 7A and B), kidney and liver ( Figure 7C) of wild type mice. In response to fasting, wild type male mice had an approximate 2-3 fold higher induction of cardiac PDK4 expression than female mice (9.362.4 male compared to 3.760.9 female; relative fold induction; n = 426). The basal PDK4 levels between male and female mice were equivalent, so this gender difference in PPARa activity/PDK4 expression upon fasting is gender specific.
Upon fasting, male wild type and CYP2J2 Tr mice, had a comparable induction of cardiac PDK4 protein (9.362.4 wild type; 7.461.3 CYP2J2 Tr; fold expression; n = 4). In contrast, female CYP2J2 Tr exhibited a much greater induction of cardiac PDK4 protein upon fasting compared to wild-type controls ( Figure 7). PDK1, PDK2 and E1a protein expression were unchanged by 24 h of fasting in any tissue tested ( Figure 6, and data not shown).
Up-regulation of PDK4 expression is linked to a decline in circulating insulin concentrations [13,14,16]. Upon fasting, both  plasma insulin and blood glucose levels fell to equivalent levels in wild type and CYP2J2 Tr mice (Table 1). Fibrate administration is associated with suppression of circulating triglyceride levels [2], however, neither triglyceride nor non-esterified fatty acid concentrations were affected in wild type or CYP2J2 Tr mice (Table 1).
Since no systemic metabolic differences were observed, any changes in PPAR response we conclude are due to the local cardiac specific activity of CYP2J2 in the transgenic mouse.

Endogenous CYPs and the cardiac fasting response
The use of pharmacological CYP inhibitors in vivo is complicated due both to lack of specificity of inhibitors and the great heterogeneity in CYP enzymes between species. We did however examine the fasting response in CYP2J5 knockout mice [17], the only murine CYP2J family member where a knockout has been generated. There was however no difference in the circulating blood glucose levels, or the heart, liver or kidney PDK4, or heart E1a expression levels (Table S1) between knockout and wild type male or female mice either under fed or fasted conditions.

Discussion
The nature of endogenous PPAR ligands are still far from clear, as is whether PPARs act as general lipid sensors or whether high affinity ligands exist in the body. Here we show CYP2J2 can act as an endogenous epoxygenase source of high affinity PPAR ligands. When co-transfected together in vitro, CYP2J2 induces PPAR, in particular PPARa, activity. In cardiac-specific-CYP2J2 Tr mice, fasting greatly elevates the PPARa target gene PDK4. These results do not exclude a role for CYP2J2 or other CYPs as regulators of PPARb/d or -c. Indeed we found CYP2J2 can activate PPARd and PPARc, (albeit it to lower absolute levels then  PPARa in our transfection system) and it is known that lipid CYP products (though not the CYP responsible) are endogenous PPARc activators, induced by laminar shear of human endothelial cells in vitro [7,8].
Unlike other proposed PPAR ligand-generating enzymes (e.g. 12/15-lipoxygenase; [18]), CYP2J2 did not require additional arachidonic/linoleic acid substrate(s), suggesting a high level of functional coupling between the epoxygenases and PPARs. We also show for the first time a functional in vivo response for a PPAR ligand generating system. Our results do not rule out the role of other enzymes, such as phospholipases [3] or lipoprotein lipases [4] implicated in PPARa ligand generation. These enzymes are likely to produce PPAR ligands in parallel to CYPs, and/or supply free fatty acid substrates for CYPs to utilise. 8,9-EET, and 11,12-EET, but not 14,15-EET activated PPARa. 11,12-EET in contrast to 14-15-EET is highly anti-  Figure (A) shows representative western blots for 2 of the 6 animals tested for PDK4, PDK1 (antibody has cross reactivity with PDK4 indicated by changes in the lower band) and E1a in the hearts of wild type (WT) or cardiac-specific CYP2J2 Tr mice (2J2); specific bands are identified by the arrows. Figures show the relative protein expression of PDK4 in the heart (B), kidney, and liver (C as indicated) and PDK1 and E1a in the heart (as indicated) in wild type (WT) and cardiac-specific CYP2J2 Tr (2J2), fed and fasted female mice. Data represents relative densitometry of protein compared to wild type fed controls for n = 426 separate animals in each group. Only the PPARa target gene PDK4 was induced on fasting both in the heart and kidney. Upon fasting there was an approximate doubling of PDK4 in the hearts (b), but not the kidney (e), or liver (f) of female cardiac specific CYP2J2 transgenic mice. * denotes p,0.05 by unpaired t-test between the fasting response in wild-type and CYP2J2 transgenic mice. doi:10.1371/journal.pone.0007421.g007 inflammatory and vascular protective [10,19]. Therefore, we propose that PPARa is a likely anti-inflammatory target for 11,12-EET and CYP2J2. Indeed we found the combination of PPARa and CYP2J2 abolished IL-1b induced NFkB activation; a central pro-inflammatory transcription factor and PPARa target [1,2].
Many EETs, including 14,15-EET, can also act as cellular hyperpolarising agents [9,19], however, since 14,15-EET was inactive in our system, hyperpolarization mechanisms are highly unlikely to be involved. Our results are consistent with previous findings that EETs and some of their metabolites can directly bind and activate PPARa [5][6][7]. Although 14,15-EET did not activate PPARa in our hands, its CYP4A hydroxylase 14,15-EET metabolite 20,14,15-HEET, was the post potent EET product we tested. EETs can be rapidly metabolised by at least 10 different intracellular pathways, and it is estimated that when given exogenously ,10% is available free within the cell [9]. Our results therefore do indicate that alternative CYP2J2 products exist or further unknown EET metabolites [5,7,8] are potential endogenous PPARa activators.
There is considerable species difference between CYPs in man and in the mouse. CYP2J2 is the human isoform, in the mouse the situation is far more complex with up to 8 putative homologues (CYP2J5 -CYP2J13; [20]). Since epoxygenases are ubiquitous and potentially have many roles, examining the role of endogenous epoxygenases especially in the mouse is extremely difficult. We therefore chose as our main model the established cardiac specific CYP2J2-Tr mouse. We did however test the recently described CYP2J5 knockout mouse [17], the only CYP2J knockout available. However, we did not detect a change in the fasting response or in PDK4 expression, suggesting a lack of involvement of CYP2J5 in PPARa ligand generation or the more likely compensation from other mouse CYP2J or CYP2C EETproducing epoxygenases that are present.
The selective augmentation of PDK4 in cardiac-specific CYP2J2-Tr mice occurred only in female mice. The fasting PPARa response was much stronger in males, and we believe maximally activated. Interestingly, our results are consistent with known gender differences in cardiac PPARa responses in the mouse. Pharmacological stress of the hearts of PPARa knockout mice with Etomoxir to prevent mitochondrial fatty acid import, results in cardiac lipid accumulation and a 100% mortality of male mice but only 25% mortality of female mice [21].
In conclusion, in vitro CYP2J2 activates PPARa without exogenous stimuli. In vivo CYP2J2 does not appear to be rate-limiting as PPARa target gene (PDK4) expression is only augmented in cardiac-specific CYP2J2 transgenic mouse upon fasting. Therefore, CYP2J2 in vivo is an enzyme apparently quiescent, but capable or responding to changes in lipid availability to generate endogenous PPARa agonists and thereby integrate transcriptional fasting events. CYP2J2 products activate PPARs, in particular PPARa in vitro and in vivo. As lipidmetabolising CYP enzymes have a widespread expression, utilise a variety of lipid substrates and produce a large family of oxidised biologically active lipid mediators, we suggest that lipid metabolising CYPs may represent an important source of PPAR ligands throughout the body.
PPARa is known as a controller of lipid metabolism and inflammation. Linking CYP2J2 and epoxygenases to PPARa has many potential clinical implications. Variants of CYP2J2 with lower activity are known in some populations to be linked to an increased risk of coronary artery disease [22,23]. Epoxygenases such as CYP2J2 in addition to metabolising arachidonic acid may also regulate xenobiotic drug metabolism. Understanding how epoxygenases are regulated, the mediators they produce, and where they work, will give us novel information on biomarkers for dyslipidaemia and inflammation, allow us to understand sideeffects of drugs metabolised by epoxygenases, and help us to design novel PPARa ligands based on the structure of high affinity EETs and their metabolites.

Cell culture and transfections
HEK293 were maintained in DMEM containing, supplemented with Antibiotic/Antimycotic mix, and 10% FCS; 37uC; 5% CO 2 ; 95% air. Cells were transfected with Novafector and Luciferase assays performed, essentially as previously described [26] but modified for a 96 well format [27]. Luciferase activity was normalised to cell protein (BCA assay). Global cellular changes, cell morphology, and GFP expression were recorded on a Nikon TE2000 inverted florescent microscope, with a SPOT RT digital camera. In some experiments organ culture of mouse cardiac tissue was performed, essentially as previously described [28].

Ethics Statement
All animal studies were approved by the National Institute of Environmental Health Sciences Animal Care and Use Committee.

Animal experiments
Cardiac-specific CYP2J2 transgenic mice (a2MHC promoter driven) and littermate wild type C57BL6/J controls [15] along with CYP2J5 knockout mice [27] have been described previously. Animals were allowed food and water ad libitum or fasted for 24 h. In some experiments mice were given SKF525A (30 mg/kg; i.p) or vehicle (sterile PBS) immediately prior to initiation of the 24 h fasting/non-fasting period.
Immunoblotting and assays PDK1, -2, -4 and E1a and CYP2J2 protein levels were determined as previously described [13,24,25]. For animal experiments each representative immunoblot presented are results from a single gel exposed for a uniform duration, and each lane represents a preparation from a different mouse. Plasma immunoreactive insulin concentrations were measured by ELISA, using rat insulin as a standard. Plasma glucose concentrations were determined by a glucose oxidase method. Plasma NEFA and TAG levels were determined spectrophotometrically using commercial kits.

Supporting Information
Table S1 Blood glucose and PDK4 expression in female fed and fasted wild type and CYP2J5 knockout (2/2) mice. Wild type and CYP2J5 2/2 mice have similar basal levels of plasma glucose. Following 24 h of fasting, blood glucose dropped in both wild type and CYP2J2 mice to equivalent levels, while the PPARalpha target gene PDK4 was induced to similar levels in the heart, liver and kidney. The non-PPARalpha target genes E1a and PDK2 (data not shown), were unaffected by fasting. Similar results were found in male mice. This data represents the mean6s.e.m. for n = 4 animals per group. * denotes p,0.05 by paired t-test between fed and fasted levels. Found at: doi:10.1371/journal.pone.0007421.s001 (0.03 MB DOC)