The Pollutant Diethylhexyl Phthalate Regulates Hepatic Energy Metabolism via Species-Specific PPARα-Dependent Mechanisms

Background The modulation of energetic homeostasis by pollutants has recently emerged as a potential contributor to the onset of metabolic disorders. Diethylhexyl phthalate (DEHP) is a widely used industrial plasticizer to which humans are widely exposed. Phthalates can activate the three peroxisome proliferator–activated receptor (PPAR) isotypes on cellular models and induce peroxisome proliferation in rodents. Objectives In this study, we aimed to evaluate the systemic and metabolic consequences of DEHP exposure that have remained so far unexplored and to characterize the underlying molecular mechanisms of action. Methods As a proof of concept and mechanism, genetically engineered mouse models of PPARs were exposed to high doses of DEHP, followed by metabolic and molecular analyses. Results DEHP-treated mice were protected from diet-induced obesity via PPARα-dependent activation of hepatic fatty acid catabolism, whereas the activity of neither PPARβ nor PPARγ was affected. However, the lean phenotype observed in response to DEHP in wild-type mice was surprisingly abolished in PPARα-humanized mice. These species differences are associated with a different pattern of coregulator recruitment. Conclusion These results demonstrate that DEHP exerts species-specific metabolic actions that rely to a large extent on PPARα signaling and highlight the metabolic importance of the species-specific activation of PPARα by xenobiotic compounds.


Research
The prevalence of obesity and its associated metabolic complications has dramatically increased during the past decades; it has been suggested that this could, to some extent, be linked to the exposure to environmental pol lutants, which coincidently increased during the same period (BaillieHamilton 2002;Grun and Blumberg 2007;Heindel 2003;Newbold et al. 2007). Endocrine disruptors constitute a wide class of chemicals that can affect human and animal populations by interfering with the synthesis, elimination, and mechanisms of action of hormones (Markey et al. 2002;Waring and Harris 2005). One of their key mechanisms of action is the modulation of gene expression programs by targeting the activity of a family of nuclear receptors that are activated by intracellular lipophilic hormones and mediators. The concept of endocrine dis ruption, which initially arose because of the interference with reproductive biology, is now gradually being broadened to other receptors implicated in different aspects of homeostasis (Tabb and Blumberg 2006).
Metabolic homeostasis requires a controlled balance between energy storage and use, and several nuclear receptors as well as their coregu lators are instrumental in regulating these pro cesses Feige and Auwerx 2007). Among these, peroxisome proliferatoractivated receptors (PPARs) play a prominent role by acting as lipid sensors that cooperate in different organs to adapt gene expression to a given metabolic status (Desvergne et al. 2004;Evans et al. 2004;Feige et al. 2006). PPARα (NR1C1) is the founding member of the family and was initially isolated as the receptor inducing the hepatic proliferation of peroxisomes in rodents in response to synthetic chemicals. However, this function represents only a small subset of the physiologic func tions regulated by this receptor. PPARα and PPARβ/δ (NR1C2, referred to here as PPARβ) share partially overlapping functions in the con trol of catabolic metabolism by promoting fatty acid oxidation in tissues with high metabolic rates such as liver or muscle (Desvergne et al. 2004;Evans et al. 2004;Feige et al. 2006). In contrast, PPARγ (NR1C3) controls fat storage in adipose tissue by promoting differentiation and survival of adipocytes and also plays major roles in the control of insulin sensitivity .
Several endocrine disruptors, including pes ticides, industrial solvents, and plasticizers, can activate PPARs in cellular models (Bility et al. 2004;Grun et al. 2006;Hurst and Waxman 2003;Kanayama et al. 2005;Lapinskas et al. 2005;Takeuchi et al. 2006). However, the metabolic consequences of PPAR activation by endocrine disruptors remain largely unknown, although some physiologic consequences are emerging, such as adipogenic action of organo tins (Grun et al. 2006) and PPARαdependent induction of hepatic peroxisome proliferation by phthalates (Lapinskas et al. 2005;Ward et al. 1998). Diethylhexyl phthalate (DEHP) is among the phthalate esters most abundantly used as industrial plasticizers and is also found in cosmetics, as well as in industrial paints and solvents. DEHP is found in flexible plastics used for manufacturing a wide variety of daily products, including medical devices and food packaging, and its propensity to leach can lead to high levels of human exposure [National Toxicology Program Center for the Evaluation of Risks to Human Reproduction (NTP CERHR) 2000]. When ingested through contaminated food, DEHP is converted by intestinal lipases to its monoester equivalent monoethylhexyl phthalate (MEHP), which is then preferentially absorbed (Huber et al. 1996). MEHP can then be metabolized into a myriad of secondary metabolites that can be excreted (Rusyn et al. 2006), among which 2ethylhexanoic acid has also been reported as a lowaffinity PPARα activator (Lapinskas et al. 2005;Maloney and Waxman 1999). The bio logical effects of exposure to DEHP are hence of major concern, but their possible adverse effects on human health remain obscure.
These observations led us to investigate the molecular aspects of phthalatemediated activation of PPARγ. In cellular models, MEHP induces adipo genesis by modulating PPARγ activity , suggesting that DEHP could promote obesity in vivo if its MEHP metabolite reaches adipose tissue (Grun and Blumberg 2007;Heindel 2003). However, MEHP also activates PPARα and PPARβ in vitro (Bility et al. 2004;Hurst and Waxman 2003;Lapinskas et al. 2005), suggesting that DEHP expo sure may also promote fatty acid oxidation. In the present study, we addressed the meta bolic consequences of DEHP exposure in mice. Intriguingly, DEHP exposure protected mice from weight gain under both a regular diet and a diet containing high fat content; this effect could be attributed solely to the hepatic oxida tive functions of PPARα. The lean phenotype of DEHPtreated mice was not observed in mice humanized for PPARα where the mouse receptor has been replaced by a human PPARα (hPPARα) transgene. At the molecu lar level, the functional differences between mouse PPARα (mPPARα) and hPPARα are associated with a different pattern of coregulator recruitment in the presence of MEHP. Altogether, our obser vations highlight key physiopathologic con sequences of chronic exposure to DEHP and highlight the metabolic importance of the spe ciesspecific metabolic consequences of PPARα activation by xenobiotics compounds.

Materials and Methods
Animal experiments. We carried out DEHP exposure in wildtype (WT) mice using C57Bl6J mice (Charles River, L'Arbresle, France), which were fed a chow diet (CD) or a highfat diet (HFD) containing 60% fat (ProvimiKliba, Kaiseraugst, Switzerland), sup plemented with 10 mL sunflower oil (vehicle) per kilogram of feed in the presence or absence of DEHP (Sigma Aldrich, Buchs, Switzerland). Genetically engineered mouse models are described in the Supplemental Material avail able online (doi:10.1289/ehp.0901217.S1 via http://dx.doi.org/). Animal experiments were conducted humanely and with regard for alle viation of suffering, and were approved by the relevant animal welfare commissions.
Metabolic phenotyping. We performed glucose tolerance tests after an intraperitoneal injection of 2 g glucose/kg body mass after 4 hr of fasting. Fat and lean body composi tions were measured by nuclear magnetic resonance (NMR) on an EchoMRI (mag netic resonance imaging) 100 apparatus (Echo Medical Systems, Houston, TX, USA). Indirect calorimetry in metabolic cages was performed with a Comprehensive Lab Animal Monitoring System (Columbus Instruments, Columbus, OH, USA) during 24 hr after a 36 to 48hr acclimation period. The exercise test was performed on a treadmill (Columbus Instruments); a 6min training period at low velocity was given 24 hr before the exercise (Feige et al. 2008;Lagouge et al. 2006). Mice were exercised until they reached fatigue [defined by resting for > 50% of the time on the shock pad; see Supplemental Material, Figure 2 (doi:10.1289/ehp.0901217.S1)]. Biochemical assays and quantitative PCR (poly merase chain reaction) techniques are described in the Supplemental Material (doi:10.1289/ ehp.0901217.S1).
Reporter gene and pull-down assays. We performed PPAR luciferasebased reporter gene assays in C2C12 mouse myoblast cells using plasmids, compounds, and procedures described previously (Feige et al. 2005. Pulldowns using glutathione Stransferase (GST)labeled coregulators and 35 Slabeled PPARα produced in reticulocyte lysates were performed as described previously (Feige et al. 2005 in the presence of vehicle, 100 µM Wy14643 (Cayman Chemicals, Ann Harbor, MI, USA), or 1,000 µM MEHP (Imperial Chemical Industries, Slough, UK). The pri mary culture of hepatocytes and adenoviral infections are described in the Supplemental Materials (doi:10.1289/ehp.0901217.S1).
Statistical analyses. All data are reported as mean ± SE. Results were considered statis tically significant when pvalues were < 0.05 by unpaired twotailed Student's ttest.

DEHP exposure protects from obesity in mice.
The ability of the DEHP metabolite MEHP to activate PPARα and PPARβ, on the one hand, and PPARγ on the other, suggests that this endocrine disruptor could potentially influence the two arms of the energetic bal ance by affecting both energy expenditure and storage. To understand the global meta bolic output of an exposure to DEHP, we fed WT C57Bl6J mice with a regular CD containing vehicle or DEHP. Animals were thereby exposed to the compound through intestinal absorption, a route that mimics one of the most frequent sources of exposure in humans and favors the conversion of DEHP to its major bioactive metabolite MEHP. DEHP was incorporated in the diet at two concentrations, leading to average exposures of 100 and 1,000 mg/kg body mass/day), dos ages that were previously reported to be less than and greater than the minimal exposure required for hepatic peroxisome proliferation, respectively (David et al. 1999). The treatment started just after weaning at 3 weeks of age. At week 13, mice treated with 1,000 mg/kg/ day DEHP gained approximately 15% less weight than did controls or mice treated with 100 mg/kg/day; this difference did not result from different feeding behaviors because all three groups consumed the same amount of food ( Figure 1A). Consistently, the lean mass meas ured by NMR was not affected by the treatment at either dose, but the weight dif ference at the high dose was solely caused by reduced total body fat ( Figure 1B). Along the same line, the mass of the epididymal white adipose tissue (epiWAT) at sacrifice was

Fat mass (g) Lean mass (g)
Week 0 W eek 13 Week 13 Week 0 Vehicle DEHP 100 mg/kg/day DEHP 1,000 mg/kg/day * * * * volume 118 | number 2 | February 2010 • Environmental Health Perspectives also reduced by 40% ( Figure 1C). Only the high dose of DEHP induced a hepatomegaly ( Figure 1C), testifying that significant per oxisome and hepatocyte proliferation occurred only at 1,000 mg/kg/day (Rusyn et al. 2006). In the blood, triglyceride and free fatty acid levels were reduced in DEHPtreated mice ( Figure 1D), suggesting that the reduced fat mass of these animals reflects enhanced fatty acid oxidation. DEHP treatment consistently increased plasma total ketone bodies, a marker of hepatic fatty acid oxidation. Plasma glu cose levels were not affected by DEHP in fed or fasted animals ( Figure 1D and data not shown, respectively). Although DEHP treat ment did not modify glucose tolerance [see Supplemental Material, Figure 1 (doi:10.1289/ ehp.0901217.S1)], insulin levels were signifi cantly lower in mice treated with 1,000 mg/ kg/day DEHP ( Figure 1D). Altogether, these results demonstrate that DEHP exposure reduces adiposity under CD.
To understand whether a diet high in fat would exacerbate or counterbalance this lean phenotype, we fed adult WT mice the HFD (containing 60% calories as fat) supple mented with vehicle or 500 mg/kg/day DEHP ( Figure 2). The HFD alone led to a doubling of body mass (Figure 2A), essentially due to the increased fat mass from 8-10% of body mass before treatment (as measured by NMR), to 30% after 13 weeks of HFD, whereas fat mass remained at 8-10% after 13 weeks of CD (data not shown). DEHP strongly protected mice from HFDinduced obesity without affecting food intake ( Figure 2A). Although hepato megaly was observed in DEHPtreated mice because of PPARαdependent peroxisome and hepatocyte proliferation (Rusyn et al. 2006), in vivo analyses of body composition revealed that the reduced body mass resulted from lower fat mass ( Figure 2B,C). Consistent with its protective effect on adiposity, DEHP inhib ited adipocyte hyper trophy and hepatic lipid droplet accumulation ( Figure 2D). Despite peroxisome proliferation in hepatocytes in response to DEHP, the global histology of the liver was normal, with no signs of inflam mation or necrosis ( Figure 2D). The normal plasma levels of alanine aminotransferase and aspartate aminotransferase ( Figure 2E), two markers of liver damage, further suggest that the lean phenotype of DEHPtreated mice is not the consequence of hepatic toxicity. In addition, the reduced adiposity did not result from a difference in lipid absorption or excre tion, because lipid concentrations in the feces were not affected by the treatment ( Figure 2F). DEHP exposure did not affect free fatty acid levels but significantly reduced triglyceride and increased ketone body concentrations, thereby suggesting a catabolic state primarily occurring in the liver ( Figure 2G). As observed under the CD, insulin levels were also reduced in the blood of DEHPtreated animals, whereas gly cemia was not affected.
Metabolic phenotyping of DEHP-treated mice. Indirect calorimetry revealed that DEHP treatment increases oxygen consump tion and carbon dioxide release both during day and night ( Figure 3A), thus providing evidence of an increased metabolic rate in DEHPtreated mice. However, the fuel pref erence was not modified by the exposure to DEHP because the respiratory exchange ratio [ratio between carbon dioxide (CO 2 ) release and oxygen (O 2 ) consumption] was not affected. Although the protective effect of DEHP on body weight gain was not a con sequence of increased spontaneous locomo tor activity ( Figure 3A), we evaluated muscle functions by an exercise test that evaluates the global oxidative capacities of skeletal muscle. The distance ran until fatigue did not differ between DEHPtreated mice and their con trols ( Figure 3B), indicating that muscle is not the primary site of MEHP action. The active DEHP metabolite MEHP can activate the three PPAR isotypes, of which PPARγ plays a crucial role in controlling insu lin sensitivity . A full PPARγ activation seems unlikely in DEHP treated mice because PPARγ activation by full agonists such as thiazolidinediones promotes adipogenesis and weight gain. However, we recently reported that MEHP acts as a selective PPARγ modulator ), a class of compounds capable of uncoupling the action of PPARγ on adipogenesis from those on insu lin sensitivity and glucose tolerance ). We therefore meas ured glucose tolerance and found that DEHP admin istration mildly enhanced glucose tolerance ( Figure 3C). However, given the detrimental action of adiposity on glucose tolerance, this effect is most likely a consequence of the lean phenotype rather than of PPARγ activation.

DEHP activates hepatic fatty acid oxidation through PPARα.
One key question to fully understand the metabolic consequences of DEHP exposure is the individual responses of organs controlling energy expenditure and storage. Because MEHP can activate the three PPAR isotypes in cellular systems, all PPAR expressing tissues are potential targets if they are exposed to MEHP upon DEHP inges tion. We thus measured the expression of well characterized PPAR target genes (Figure 4). In the liver, DEHP exposure induced the expres sion of genes implicated in various aspects of fatty acid oxidation, including mitochondrial βoxidation (medium chain acylCoA dehy drogenase; MCAD), peroxisomal βoxidation (AcylCoA oxidase 1; ACOX), intracellular fatty acid shuttling (fatty acid binding protein 1; FABP-1), and mitochondrial fatty acid import (carnitine palmitoyl transferase 1a; CPT-1a) ( Figure 4A). Strikingly, DEHP also robustly induced the hepatic expression of fibroblast growth factor 21 (FGF-21), a gene recently described to control lipid oxidation and keto genesis in the liver and lipolysis in the adipose tissue (Badman et al. 2007;Inagaki et al. 2007;Kharitonenkov et al. 2005). These gene profil ing experiments were functionally confirmed by the biochemical measurement of hepatic hydroxyacylCoA dehydrogenase activity, which demon strated that fatty acid oxidation was increased in the liver of DEHPtreated mice ( Figure 4E). Because PPARα is the most prominent hepatic PPAR isotype and all five of the genes we tested are reported to be under the direct transcriptional control of PPARα, these observations suggested that the lean phe notype of DEHPtreated mice results at least in part from hepatic fatty acid oxidation mediated by PPARα. Other energydissipating tissues expressing PPARs do not seem to contribute to this phenotype as oxidative target genes in skeletal muscle [long chain acylCoA dehydro genase (LCAD) and pyruvate dehydrogenase kinase 4 (PDK4)] and in brown adipose tissue [uncoupling protein 1 (UCP1) and PPARγ coactivator 1α (PGC-1α)] were not affected ( Figure 4B,D). In addition, the expression of adipogenic target genes under direct transcrip tional control of PPARγ in white adipose tissue (WAT) was not affected ( Figure 4C). These gene expression data therefore demonstrate that DEHP exposure does not affect adipo genesis in vivo and drives a catabolic response occurring primarily in the liver, most probably through activation of PPARα. . Gene expression was normalized to three housekeeping genes (glyceraldehyde-3-phosphate dehydrogenase, β glucuronidase, and ribosomal protein S9 for the liver; eukaryotic translation elongation factor 1α1, TATA box binding protein, and ribosomal protein S9 for the other tissues). Abbreviations: ACOX, acyl-CoA oxidase 1; ACS-1, acetyl-CoA synthetase 1; AdipoQ, adiponectin; CPT-1a, carnitine palmitoyltransferase 1a (liver); FABP-1, fatty acid binding protein 1 (liver); FABP-4/aP2, fatty acid binding protein 4 (adipocyte); FATP-1, fatty acid transport protein 1; FGF-21, fibroblast growth factor 21; LCAD, long chain acyl-CoA dehydrogenase; MCAD, medium chain acyl-CoA dehydrogenase; PDK4, pyruvate dehydrogenase kinase 4; PGC-1α, PPARγ coactivator 1α; UCP1, uncoupling protein 1. (E) 3-Hydroxyacyl-coenzyme A dehydrogenase (HAD) activity was measured in liver extracts of DEHP-treated mice given the HFD. *p ≤ 0.05 compared with vehicle treatment. Because the oxidative functions of PPARα and PPARβ partially overlap, we dissected the contribution of each individual recep tor by feeding PPARα and PPARβnull mice the HFD supplemented with DEHP. We observed the protective effects of DEHP on weight gain in WT mice as well as in PPARβnull mice ( Figure 5A,B). In contrast, DEHP did not influence the body mass of PPARαnull mice ( Figure 5A), demonstrat ing that the antiobesity action of DEHP requires PPARα only. This conclusion was further corroborated by reduced fat mass after DEHP exposure in PPARβnull mice but not PPARαnull mice ( Figure 5C-F). Similarly, the DEHPdependent hepatomegaly caused by peroxisome proliferation was PPARα but not PPARβdependent ( Figure 5E,F). PPARα was also indispensable to the hepatic induc tion of oxidative gene expression after DEHP exposure ( Figure 5G), whereas the influence of PPARβ was modest to non existent based on the genes analyzed ( Figure 5H). To fur ther demon strate the role of hepatic PPARα in the response to DEHP, we isolated primary hepatocytes from WT and PPARαnull mice and showed that only WT hepatocytes were responsive to MEHP. In addition, the lack of response to MEHP in PPARαnull hepato cytes was restored by adenoviralmediated over expression of PPARα ( Figure 5I). Altogether, these results demonstrate that DEHP protects from obesity by activating the catabolic func tions of PPARα in the liver.
DEHP protects WT mice but not humanized models from obesity. The identification of hepatic PPARαdependent activation as the cause of the lean phenotype of DEHPtreated mice raises the question of how this protective action may translate to humans exposed to DEHP. In mice, peroxisome and hepatic prolif eration is directly controlled by PPARα activa tion and coexists with metabolic actions of the receptor on fatty acid oxidation, whereas only the actions on fatty acid oxidation extend to humans. To overcome this dual role of PPARα in mice, we analyzed the action of DEHP on HFD in a humanized mouse model of PPARα that is insensitive to hepatocellular proliferation (Yang et al. 2008). In this model, the entire human PPARα gene and its regulatory regions were introduced into the mPPARα null back ground, thereby allowing the expression, reg ulation, and function of human PPARα in the mouse. Although DEHP limited weight gain in WT mice as shown above, PPARα humanized mice were not protected against dietinduced obesity ( Figure 6A), demon strating that this beneficial effect is limited to mPPARα activation. Indeed, and as expected, humanized PPARα mice did not develop the DEHPinduced hepatomegaly caused by cell proliferation (Figure 6B), but presented a slight excess of adiposity upon DEHP expo sure. This was evidenced by an increase in total body mass and in epiWAT ( Figure 6A,B). We thus evaluated gene expression in the liver and WAT to understand how DEHP may promote fat accumulation in humanized PPARα mice. Strikingly, although this PPARαhumanized mice model can respond to a classic PPARα agonist (Yang et al. 2008), we observed no activation of the genes and enzymatic activities controlling hepatic βoxidation in response to DEHP ( Figure 6C,D). The fact that DEHP treated PPARαhumanized mice gained more weight than did their untreated counterparts was furthermore not caused by enhanced adipo genesis and fat storage via PPARγ acti vation in WAT because the expression of direct adipogenic PPARγ target genes was not modified (data not shown). Altogether, these results demon strate that the protective effects of DEHP on weight gain are dependent on the species origin of the PPARα receptor and most probably do not extend to humans, in which  C57Bl6J background (B, D, F, H) were fed HFD and exposed to vehicle or 500 mg/kg DEHP as described in Figure 2 the metabolic consequences of DEHP expo sure, if any, should rather be detrimental. According to the humanized mouse model used in this study, the mechanism involved in the speciesspecific response to DEHP seems to rely solely on differences in the receptor itself. To understand the molecular basis of this differential action, we first evaluated the ability of mPPARα and hPPARα to activate the PPAR response element of a reporter con struct in the presence of the full PPAR agonist Wy14643 or of MEHP, the active metabolite of DEHP ( Figure 7A). In this assay, MEHP treatment resulted in a maximal activation of mPPARα and hPPARα to a similar extent, despite having a slightly higher affinity for the mouse than for the human receptor, as previously observed (Bility et al. 2004;Hurst and Waxman 2003). To further analyze what could be the functional consequence of the subtle conformational differences of mPPARα versus hPPARα, we tested their ability to inter act with coregulators in the absence of ligand or upon binding of Wy14643 or MEHP ( Figure 7B). In the presence of Wy14643, mPPARα and hPPARα shared a similar pat tern of interaction with coregulators, releasing NCoR (nuclear receptor corepressor 1) and efficiently recruiting Med1 (mediator complex subunit 1), PGC1α (peroxisome proliferator activated receptorγ coactivator 1α), and to a lower extent p300 (E1A binding protein p300). In contrast, MEHP induced a partial release of NCoR with both the mouse and the human receptor but failed to induce the liganddependent recruitment of the coactiva tors Med1, p300, and PGC1α to mPPARα. In contrast, all three coactivators were effi ciently recruited by the human receptor in the presence of MEHP, at a level very close to that obtained with Wy14643. Together with the experiments in the humanized PPARα mouse model, these in vitro observations sug gest that subtle differences in the mPPARα and hPPARα receptors contribute to a species specific PPARαdependent metabolic response to DEHP.

DEHP protects from diet-induced obesity in WT mice via hepatic PPARα activation.
DEHP is the most widely used industrial plas ticizer, and human exposure to this pollutant is high through the daily use of polyvinyl chlo ride products (NTPCERHR 2000). Mice fed either CD or HFD supplemented with DEHP were partially protected from dietinduced obesity because they gained 30% less weight than did their controls without modifying their feeding behavior. This reduced weight gain resulted solely from reduced fat mass and was associated with a metabolic improvement, which includes lower levels of tri glycerides in the liver and the blood, smaller adipocytes, and enhanced glucose tolerance. Using a combina tion of metabolic phenotyping, metabo lomics, gene expression profiling, and genetically engi neered mouse models, we could attribute this phenotype primarily to a PPARαdependent activation of fatty acid catabolism in the liver. DEHP induced the expression of genes con trolling βoxidation of fatty acids in the liver Figure 6. DEHP protects from obesity in mouse but not human models. PPARα WT and PPARα-humanized mice were fed HFD and exposed to vehicle or 500 mg/kg/day DEHP as described in Figure 2  as well as that of FGF21, a novel PPARα tar get important for lipid oxidation and ketogen esis (Badman et al. 2007;Inagaki et al. 2007). We demonstrated that PPARα was required for the protective effect of DEHP by showing that both the protection from dietinduced obesity and the induction of oxidative genes in the liver were completely abolished in PPARα null mice. Although our results clearly demon strate DEHPmediated metabolic interference via PPARα, other metabolic pathways-such as those driven by CAR (constitutive andros tane receptor) and RevErbα (nuclear receptor subfamily 1, group D, member 1)-could also be involved (Eveillard et al. 2009).
Unlike what has been hypothesized from the ability of MEHP to activate PPARγ and promote adipogenesis in cellular models (Bility et al. 2004;Grun and Blumberg 2007;Heindel 2003;Hurst and Waxman 2003), we found that DEHP expo sure does not cause PPARγ activation and fat accumulation in the adipose tissue of WT mice. The reasons for this observation remain unclear because MEHP has been detected as the major metabolite in the blood of rats treated with doses of DEHP in the same range as those used in the present study (Akingbemi et al. 2004). In those animals, circulating MEHP concen trations were around 10 µM (Akingbemi et al. 2004), a level at which PPARγ can be activated by MEHP in cellular assays ). Moreover, in our previous study , the levels of MEHP in the blood were sufficient to induce the hepatic activation of PPARα, an activation that occurs at similar concentrations for PPARα and PPARγ. Thus, the most likely hypothesis is that MEHP can not enter adipocytes in sufficient concentra tions to activate PPARγ.
DEHP exacerbates diet-induced obesity in a humanized PPARα mouse model. It is well established that DEHP is a potent inducer of hepato cellular and peroxisome prolifera tion in rodents through PPARα activation by MEHP (Rusyn et al. 2006) and that these proliferative processes do not occur in humans after PPARα activation (Holden and Tugwood 1999). To overcome this limita tion and to evaluate the effects of DEHP in a model more relevant to humans, we adminis tered DEHP to mice humanized for PPARα (i.e., expressing the human PPARα receptor in the mPPARα null background) that do not exhibit hepatocellular proliferation but whose PPARαdependent oxidative functions remain intact (Cheung et al. 2004;Yang et al. 2008). Strikingly, the metabolic actions of DEHP were abolished in this mouse model, which actually tended to be more sensitive to diet induced obesity than were untreated controls. This unexpected phenotype may relate to two main cellular defects. First, it is possible that hepatocellular proliferation and the resultant hepatomegaly itself are important metabolic contributors. Second, and more specifically, genes responsible for fatty acid oxidation are not upregulated in the liver of DEHPtreated humanized PPARα mice, whereas they are markedly induced in response to synthetic PPARα agonists.
Although MEHP induced a similar trans activation of mPPARα and hPPARα in cel lular models, it actually induced stronger interactions with coactivators with the human receptor than with the mouse receptor in GST pulldowns. It is therefore surprising that the induction of PPARα target genes by DEHP exposure is blunted in PPARαhumanized mice. However, interactions with coregula tors depend highly on the promoter context (Guan et al. 2005), and the regulation of genes crucial to metabolic homeo stasis could potentially involve promoterspecific interac tions with coregulators. Because activation of PPARα was recently reported to repress Let7C microRNA in WT but not in human ized PPARα mice (Shah et al. 2007;Yang et al. 2008), the blunted response to DEHP in PPARαhumanized mice may also reflect a dependency on this microRNAmediated response. However, these hypotheses do not exclude the possibility that secondary MEHP metabolites could also play important roles in the metabolic actions of DEHP by differen tially activating hPPARα and mPPARα.

Considerations regarding the metabolic consequences of DEHP exposure in humans.
The extrapolation of the action of DEHP from animal models to humans is definitely a complex issue that depends both on the rela tive levels of exposure and on the conservation of the underlying molecular mechanisms of action between species. Our studies involved relatively high levels of DEHP exposure, which were chosen according to previous reports of DEHP administration in rodents (David et al. 1999;Lapinskas et al. 2005). These doses are 2 to 3 orders of magnitude greater than esti mated human exposures when normalized to body mass (NTPCERHR 2000). However, the extrapolation of biological observations from rodents to humans is not linear to body mass, and levels of mouse exposure similar to those used herein lead to plasmatic MEHP concentrations in the low micromolar range (Akingbemi et al. 2004). Moreover, individ uals requiring frequent blood transfusion or dialysis are subjected to repetitive acute expo sures to high levels of DEHP because of the leaching of the compound from plastic bags and tubing in direct contact with biological fluids. Under such circumstances, the plasma levels of MEHP can also reach the micromo lar range in humans (NTPCERHR 2000), suggesting that the metabolic consequences of DEHP exposure described here may extend to humans. Thus, the most relevant observation is that DEHP slightly promotes weight gain in PPARαhumanized mice. Consistent with this observation, the urinary concentrations of phthalate metabolites have been recently posi tively correlated with obesity and insulin resis tance in humans (Hatch et al. 2008;Stahlhut et al. 2007). Our results clearly demonstrate that these effects are most likely not linked to activation of PPARγ and adipo genesis in the adipose tissue but potentially to low hepatic oxidative metabolism.

Conclusion
Altogether, our results demonstrate that expo sure to the environmental pollutant DEHP has farreaching metabolic consequences that rely on hepatic oxidative metabolism via PPARα activation. Furthermore, our study also high lights a speciesspecific relationship between exposure to DEHP and dietinduced obesity.