ToxSci Advance Access originally published online on December 1, 2005
Toxicological Sciences 2006 90(2):269-295; doi:10.1093/toxsci/kfj062
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Published by Oxford University Press 2005.
REVIEW |
The Toxicology of Ligands for Peroxisome Proliferator-Activated Receptors (PPAR)
* Department of Veterinary and Biomedical Sciences and Center for Molecular Toxicology and Carcinogenesis, The Pennsylvania State University, University Park, Pennsylvania 16802; and Laboratory of Metabolism, National Cancer Institute, Bethesda, Maryland 20892
1 To whom correspondence should be addressed at Center for Molecular Toxicology and Carcinogenesis, 312 Life Sciences Building, The Pennsylvania State University, University Park, PA 16802. Fax: (814) 863-1696. E-mail: jmp21{at}psu.edu.
Received September 21, 2005; accepted November 28, 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONPPAR{alpha} LIGANDSPPAR{gamma} LIGANDSPPARß/{delta} LIGANDSCONCLUSIONS AND FUTURE...REFERENCES |
|---|
Peroxisome proliferator-activated receptors (PPARs) are ligand activated transcription factors that modulate target gene expression in response to endogenous and exogenous ligands. Ligands for the PPARs have been widely developed for the treatment of various diseases including dyslipidemias and diabetes. While targeting selective receptor activation is an established therapeutic approach for the treatment of various diseases, a variety of toxicities are known to occur in response to ligand administration. Whether PPAR ligands produce toxicity via a receptor-dependent and/or off-target-mediated mechanism(s) is not always known. Extrapolation of data derived from animal models and/or in vitro models, to humans, is also questionable. The different toxicities and mechanisms associated with administration of ligands for the three PPARs will be discussed, and important data gaps that could increase our current understanding of how PPAR ligands lead to toxicity will be highlighted.
Key Words: peroxisome proliferator-activated receptors (PPAR); toxicity; carcinogenesis; receptor-mediated toxicology.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONPPAR{alpha} LIGANDSPPAR{gamma} LIGANDSPPARß/{delta} LIGANDSCONCLUSIONS AND FUTURE...REFERENCES |
|---|
Peroxisome proliferator-activated receptor-
(PPAR) was the first of three PPARs to be identified, approximately fifteen years ago (Issemann and Green, 1990
). The PPAR was named in large part because it was thought to mediate the phenomenon of increased hepatic peroxisome volume and density, or peroxisome proliferation. This hypothesis was proven correct when the PPAR-null mouse was constructed and shown to be resistant to peroxisome proliferation in response to administration of the PPAR ligands clofibrate and Wy-14,643 (Lee et al., 1995). Two additional receptors, termed PPARß (also referred to as PPAR
) and PPAR
were identified shortly after the discovery of PPAR (Dreyer et al., 1992; Kliewer et al., 1994), as they shared significant sequence homology with PPAR, but are not involved in mediating peroxisome proliferation. The three PPARs are encoded by separate genes, are differentially expressed in various tissues, and are found in all mammalian species examined to date (Peters et al., 2005). This suggests that PPARs have evolved to function as critical modulators of physiological homeostasis, by responding to endogenous ligands and transcriptionally regulating genes involved in important biological processes including lipid and glucose metabolism and transport, and associated energy homeostasis.
PPARs belong to the soluble nuclear hormone receptor superfamily. The classic mechanism of action of PPARs is similar to that of other members of this superfamily (Fig. 1A). In response to ligand binding, PPARs undergo a conformational change in protein structure allowing for dissociation of co-repressor proteins such as NCoR that inhibit transcription, coincident with heterodimerization with another nuclear receptor, retinoid X receptor- (RXR). Co-activator proteins (e.g., p300/CBP, SRC-1) and histone acetyl transferases (HATs) are recruited to the receptor heterodimer along with RNA polymerase resulting in a relaxing of chromatin to permit transcription of target genes containing the direct repeat-1 (DR1) consensus sequence binding elements for PPARs. Numerous PPAR target genes have been identified in the past fifteen years that are regulated by PPARs (reviewed in Desvergne and Wahli, 1999; Mandard et al., 2004). Changes in gene expression mediated by PPARs facilitate the complex modulation of intracellular processes required to maintain homeostasis in a constantly changing environment. While this classic mechanism of action is certainly critical for mediating many of the effects induced by PPAR ligands, there are a number of alternative mechanisms that must be considered. For example, in addition to the classic mechanism of action, PPAR and PPAR can interfere directly with NF-
B and AP1 signaling through direct protein-protein interactions (Delerive et al., 1999; Wang et al., 2002) (Fig. 1B). PPAR-mediated repression of NF-kB target genes that lead to inhibition of inflammation was recently shown to be due to ligand-dependent SUMOylation of PPAR, which causes PPAR to localize to inflammatory gene promoters in association with co-repressors (Pascual et al., 2005). PPARs can also down-regulate gene expression, through another mechanism that has yet to be elucidated. For example, PPAR is required to mediate down-regulation of apolipoprotein CIII mRNA expression as revealed by studies in fibrate-treated PPAR-null mice (Peters et al., 1997b); however, the molecular events responsible for this transcription inhibition are not completely understood. A relatively unique mechanism of transcriptional repression has also been described for PPARß (Fig. 1C), as PPARß can inhibit transcription of PPAR and PPAR gene expression in vitro (Shi et al., 2002). Studies with fluorescently labeled PPAR and PPAR fusion proteins indicate that PPARs are primarily found in the nucleus, independent of the presence of ligands (Akiyama et al., 2002a). Recent evidence suggests that in the absence of exogenous ligands in adipocytes, PPAR co-localizes with either co-activators (Fig. 1D) or co-repressors (Fig. 1E) on PPAR target gene promoters facilitating constitutive expression of some target genes (e.g., aP2) while repressing expression of others (e.g., glycerol kinase), respectively (Guan et al., 2005a). Administration of a PPAR ligand leads to both the classic transcriptional changes in PPAR target genes by causing conformational changes in PPAR, co-repressor release and co-activator recruitment, in addition to indirectly inducing PGC-1, which destabilizes co-repressor binding to PPAR (Guan et al., 2005a). It is important to note that considerably less is known about these latter mechanisms of action (Figs. 1B–E) as compared to the classic ligand-receptor mechanism (Fig. 1A). Even with the classical mechanism, ligands can also differentially effect co-activator and co-repressor recruitment, thus modulating biological activities and possible toxicities through activating subsets of PPAR target genes similar to what has been found with selective androgen receptor modulators (SARMS) and selective estrogen receptor modulators (SERMS). This illustrates the complexity of how the PPARs can modulate transcriptional activity and suggests that the mechanisms underlying PPAR ligand associated toxicity could be due to any one, or combination of these mechanisms. Adding to this complexity are the facts that secondary and tertiary events that result from ligand activation of PPARs can also contribute to the biological changes that occur, and that different ligands for PPARs elicit different biological responses, that may or may not be directly mediated by PPARs.
|
While many physiological responses induced by PPAR ligands are clearly directly mediated by activating a specific PPAR subtype, there are also examples of effects that could be due to "off-target" mechanisms. For example, PPAR ligands can activate kinases that lead to phosphorylation cascades that can trigger transcriptional changes in addition to directly phosphorylating PPARs (reviewed in Diradourian et al., 2005
; Gardner et al., 2005). This could lead to several levels of downstream regulation that could contribute to PPAR ligand toxicity. For example, there is evidence that PPAR ligands can activate ERKs, which could result in both increased signaling through this pathway, but also lead to phosphorylation of PPARs themselves (Fig. 1F). While it is known that PPARs can be phosphorylated and lead to changes in transcriptional activity based on reporter assays, the functional significance of PPAR phosphorylation on target gene expression in vivo is uncertain. It is possible that phosphorylation of PPARs could alter ligand binding, co-repressor/co-activator recruitment, or heterodimerization with RXR that could all influence PPAR-mediated transcription. This is further complicated by reports suggesting that growth factor-mediated signaling might also interact with other kinase signaling resulting in changes in PPAR transcriptional activity (Fig. 1F). PPAR ligands are also known to have anti-inflammatory effects that could be due to direct inhibition of pro-inflammatory enzymes (Fig. 1G) (Kim et al., 2005a) and inhibition of pro-inflammatory modulators (e.g., iNOS, COX-2, TNF, etc) (Chawla et al., 2001; Welch et al., 2003). Interestingly, the anti-inflammatory action of PPAR ligands can be mediated by both PPAR-dependent (Welch et al., 2003) and PPAR-independent mechanisms (Chawla et al., 2001; Kim et al., 2005a; Welch et al., 2003). It is important to emphasize that in some cases, pharmacological and sometimes suprapharmacological concentrations of ligand are required to cause an anti-inflammatory effect in experimental animal model systems, and thus may not reflect the activity expected with specific receptor activation. This issue will be addressed in more detail later. PPAR ligands are also reported to cause mitochondrial dysfunction (Fig. 1H). A number of studies have shown that PPAR and PPAR ligands can inhibit oxidative phosphorylation (Brunmair et al., 2004; Keller et al., 1992, 1993a; Martinez et al., 2005; Scatena et al., 2003, 2004a,b; Tirmenstein et al., 2002; Zhou and Wallace, 1999), and since some of these effects were demonstrated using isolated mitochondria that do not express PPARs, this suggests that ligand-induced mitochondrial dysfunction is due to a mechanism that is PPAR independent. However, while there is evidence that PPAR ligands can cause mitochondrial dysfunction leading to inhibition of oxidative phosphorylation, the role of this effect, if any, on specific ligand-induced toxicities is unknown. Lastly, in addition to all of the above described mechanisms that could contribute to PPAR ligand-induced toxicity, it is also worth noting that it is possible that PPAR ligands may interact with other receptors (Fig. 1I) or with other unidentified molecular targets (Fig. 1J), and that this putative interaction leads to, or contributes to ligand-specific toxicities.
As noted above, PPAR ligand-induced toxicity could be influenced by any number of mechanisms that may or may not be receptor-dependent. While it is tempting to suggest that a specific PPAR ligand would cause toxicity solely through a receptor-mediated mechanism, this may not always be true. PPAR ligand signaling can be influenced by differences in receptor function (e.g., interactions with other proteins, repressive influences, etc.), in addition to events that are not mediated specifically by the receptor (e.g., kinase activation, enzyme inhibition, anti-inflammatory action, mitochondrial dysfunction, etc.). The use of knockout mouse models has been instrumental in delineating receptor-dependent toxicity. For example, PPAR ligand-induced liver cancer clearly requires PPAR since PPAR-null mice are refractory to the development of hepatocellular carcinomas in contrast to mice expressing a functional PPAR (Hays et al., 2005; Peters et al., 1997a). The observations that ligand-induced enhanced expression of cell cycle regulatory genes, hepatomegaly and replicative DNA synthesis do not occur in the absence of PPAR expression upon treatment with PPAR ligands (Hays et al., 2005; Peters et al., 1997a,1998), demonstrates that non-receptor-dependent mechanisms do not contribute to the mechanisms underlying PPAR ligand-induced hepatocarcinogenesis. Unfortunately, null mouse models have not always been examined for PPAR ligand-induced toxicities, which in turn complicates interpretation of the literature on the mechanisms underlying PPAR ligand-induced toxicities. For many of the examples that will be discussed below, the precise series of molecular events that are initiated by administration of a PPAR ligand, and the physiological relevance to PPAR ligand-mediated events, including toxicity, are not always well characterized. This is due in part to the complexity associated with PPAR signaling on a number of different levels, and to the fact that there is a large diversity in the types of compounds that can act as either endogenous or exogenous PPAR ligands.
There are a variety of endogenous and exogenous compounds that can bind to and activate PPARs. However, the focus of this review is on PPAR ligands that are reported to cause toxicity, and will not include discussion on endogenous ligands since it is unlikely that under ordinary circumstances, activation by endogenous ligands leads to toxicity. Figure 2 illustrates a variety of different synthetic PPAR ligands. In some cases, specificity for a PPAR isoform is high but this is not always true. Synthetic compounds have recently been developed that exhibit high selectivity for one PPAR, as compared to the other two isoforms. For example, L796449 (a phenylacetic acid derivative) preferentially activates human PPAR as compared to PPAR or PPARß (Berger et al., 1999); GW501516, GW0742 and L165041 preferentially activate PPARß as compared to PPAR or PPAR (Berger et al., 1999; Sznaidman et al., 2003); and GW7647 preferentially activates PPAR as compared to PPARß or PPAR (Seimandi et al., 2005). However, some compounds can activate all three PPARs (e.g., pan agonists) or in some cases, preferentially activate two of the three PPARs (e.g., dual agonists). For example, bezafibrate activates all three PPARs and the EC50 for receptor activation is similar for all three PPARs ranging from 55–110 µM (Shearer and Hoekstra, 2003). Muriglitazar, a recently developed dual PPAR/ agonist exhibits high selectivity for these PPAR isoforms with an EC50 of 320 nM and 110 nM for PPAR and PPAR, respectively (Devasthale et al., 2005). In addition to synthetic ligands for PPARs, endogenous compounds have also been shown to bind to and activate PPARs including fatty acids and eicosanoids (Forman et al., 1997; Keller et al., 1993b; Kliewer et al., 1997; Yu et al., 1995). Similar to synthetic PPAR ligands, endogenous compounds that activate PPARs exhibit variation in their specificity of activation.
|
The structure of the ligand binding domain of the PPARs determined using x-ray crystallography (Gampe et al., 2000
; Nolte et al., 1998; Uppenberg et al., 1998; Willson et al., 2001; Xu et al., 1999, 2001, 2002) suggests that features of both the ligand and the receptor modulate relative ligand binding affinity. The acidic groups typically found in most PPAR ligands (Fig. 2) participate in hydrogen bonding within the ligand binding domain, in particular with a key tyrosine residue in the C-terminus AF-2 helix (Gampe et al., 2000; Nolte et al., 1998). However, some PPAR ligands exhibit no binding to PPARß/ because the acidic head group is too large to fit into the ligand binding domain, which is significantly smaller as compared to the ligand binding domain of either PPAR or PPAR (Xu et al., 2001). In contrast, GW501516 has an unsubstituted phenoxyacetic acid group that allows for a better fit into the ligand binding domain of PPARß/. It was also been shown that tyrosine-314 in PPAR and the histidine-323 in PPAR are the primary reason for ligand specificity between these two receptors, as these amino acids facilitate critical hydrogen bonds within the ligand binding domain (Xu et al., 2001). Thus, PPAR ligands activate PPARs due in part to specific functional groups within each molecule, in addition to structural differences in the amino acid composition of the ligand binding domain, that combined allow for hydrogen bonding and other molecular interactions between ligand and the receptor. The review will focus on the different types of toxicity described in the literature to date for the different classes of PPAR ligands, with an emphasis on illustrating what is known regarding the role of the ligand/receptor-mediated mechanism, in addition to receptor-independent events that could contribute to toxicity (Table 1). Some of this discussion centers on the use of null mouse models generated to determine receptor-specificity, which are described in detail elsewhere but summarized in Table 2.
|
|
|
PPAR LIGANDS
|
|---|
|
TOPABSTRACTINTRODUCTIONPPAR{alpha} LIGANDSPPAR{gamma} LIGANDSPPARß/{delta} LIGANDSCONCLUSIONS AND FUTURE...REFERENCES |
|---|
The focus of this review is on PPAR ligands that cause toxicity, with an emphasis on PPAR agonists with a reasonable number of reports to allow for a good discussion. It is noteworthy that there are relatively potent PPAR
ligands that are under investigation as possible next generation hypolipidemic drugs but the toxicity (if any) associated with these recently developed chemicals is currently uncertain. Similarly, a number of PPAR/PPAR dual agonists are under development and in clinical trials. To date, the most well studied PPAR ligands are the fibrate class of hypolipidemic drugs (bezafibrate, clofibrate, fenofibrate, etc.) that have been used therapeutically for many years in humans. Fibrates, while relatively weak PPAR ligands, are capable of activating PPAR at pharmacological doses leading to increased expression of lipid metabolizing enzymes that effectively lower serum lipid levels in humans. Extensive studies have been performed examining the toxic effects of fibrates. Phthalates are another class of PPAR ligands (and in some cases PPARß/ and PPAR ligands as well) that are used as monomers in the production of pliable plastic products such as footware, toys and hospital IV tubing and bags. Phthalates are not functional PPAR ligands until they are metabolized by carboxylesterases to phthalate monoesters, and the relative ability of phthalate monoesters to activate PPAR increases with increasing side-chain length of the esterified moiety (Bility et al., 2004). Phthalate monoesters are very weak PPAR ligands as compared to fibrates and more potent synthetic PPAR ligands (Hurst and Waxman, 2003). Trichloroacetic acid (TCA) and dicholoracetic acid (DCA), metabolites of trichloroethylene (TCE), an industrial solvent and environmental contaminant, can also activate PPAR although their affinities are among the lowest of all PPAR ligands (Maloney and Waxman, 1999). Perfluoroalkyl acids and derivatives are also environmental contaminants that can activate PPAR (Maloney and Waxman, 1999; Shipley et al., 2004). It is important to note that the relative ability to activate PPAR is typically based on cell-based reporter assays and in some cases relative binding assays that rely on ligand-induced structural changes in the receptor that result in coactivator and corepressor binding that can be assayed. However, ligand binding has not been demonstrated for many ligands, particularly low-affinity environmental and industrial chemicals. For example, while it is known that TCA and DCA can activate PPAR as shown by cell-based reporter assays (Maloney and Waxman, 1999), it is not known whether these compounds directly bind to PPAR. It is possible that these, and other chemicals can "activate" PPAR, but do so by displacing other endogenous ligands, which in turn activate the receptor. They could also cause metabolic changes that lead to production of endogenous ligands, similar to the possibility that metabolites of ligands may be direct ligands for PPAR. These possibilities await futher experimentation. While there are no known toxicities associated with endogenous ligands of PPAR, there are a number of toxicities associated with the synthetic PPAR ligands in animal models.
Reproductive and Developmental Toxicity
Studies describing reproductive or developmental toxicity associated with exposure to PPAR ligands are limited. However, considerable research by pharmaceutical companies investigating the reproductive and developmental toxicities associated with the fibrate class of hypolipidemic drugs was likely performed prior to approval by the appropriate agencies. Administration of clofibrate or gemfibrozil to rats and mice during pregnancy causes atypical changes in maternal and fetal liver associated with peroxisome proliferation and induction of lipid catabolism (Cibelli et al., 1988; Stefanini et al., 1989; Wilson et al., 1991), but evidence of overt teratogenesis is lacking (Chhabra and Kurup, 1978; Cibelli et al., 1988; Fitzgerald et al., 1987; Nyitray et al., 1980; Pantaleoni and Valeri, 1974). High doses of clofibrate (>500 mg/kg) during pregnancy are reported to lead to an impaired incidence of successful implantation (Pantaleoni and Valeri, 1974) and reduced maternal weight gain (Cibelli et al., 1988) (which could be due to the induction of lipid catabolism). Thus, there is no strong evidence that either clofibrate or gemfibrozil can cause developmental toxicity/teratogenicity, while administration of clofibrate can cause some signs of maternal toxicity, but the dose required to elicit this response is considerably higher than those used therapeutically.
While administration of fibrates during pregnancy is not associated with teratogenicity, there are numerous reports that some environmental contaminants including phthalates, perfluoroalkyl derivatives and TCE, which are relatively weak PPAR agonists, can cause both reproductive and developmental toxicity. Administration of phthalates, prior to and during pregnancy, in rodent models can cause reduced fertility rates (Peters and Cook, 1973; Singh et al., 1974; Tyl et al., 1988), altered ovulation (Davis et al., 1994), altered development of the male reproductive tract and impaired spermatogenesis (reviewed in Corton and Lapinskas, 2005) and teratogenesis including skeletal, cardiac, and neural tube defects (Gao et al., 2003; Ritter et al., 1985; Shiota and Nishimura, 1982; Singh et al., 1972; Tomita et al., 1982). Exposure to perfluorooctanoic acid (PFOA) or perfluorooctane sulfonate (PFOS) during pregnancy can cause developmental defects including reduced fetal weights, skeletal and cardiac malformations (reviewed in Lau et al., 2004). In contrast, exposure to TCE during pregnancy is associated with cardiac defects (Dawson et al., 1990, 1993; Loeber et al., 1988) but not with changes in fertility, ovulation, or testicular function.
The role of PPAR in mediating the developmental and reproductive toxicity associated with exposure to the relatively weak PPAR ligands, phthalates, PFOA, PFOS, or TCE is not clear. Since administration of fibrates during pregnancy does not result in overt teratogenesis while exposure to the relatively weak PPAR ligands does, this argues against the hypothesis that activation of PPAR during development is a central mechanism of action leading to teratogenesis. Consistent with this view, administration of DEHP during organogenesis causes neural tube defects in both wild-type and PPAR-null mice (Peters et al., 1997c), demonstrating that PPAR is not required to mediate this effect. Indeed, there is good evidence that DEHP and other phthalates may cause a functional zinc deficiency, which could explain some of the developmental and/or reproductive alterations induced by phthalate exposure by a mechanism independent of PPAR (Curto and Thomas, 1982; Parkhie et al., 1982; Peters et al., 1997c; Thomas et al., 1982). While cardiac malformations are not reported with fibrate exposure during development, it is of interest to note that they are found after exposure to phthalates, PFOA, PFOS, and TCE. However, administration of both clofibrate and PFOS, but not PFOA, during gestation results in significant perinatal mortality (Butenhoff et al., 2004; Lau et al., 2003; Nyitray et al., 1980). Whether or not PPAR mediates either of these effects has not been examined to date. The observations that clofibrate, gemfibrozil, PFOA exposures do not cause altered ovulation while phthalate esters do, argues against the notion that activation of PPAR is a central mechanism leading to altered female reproductive toxicity. It has been hypothesized that activation of PPAR leads to inhibition of aromatase and increased expression of 17ß-HSD IV, which combined could contribute to reduced estradiol levels and anovulation (Lovekamp-Swan and Davis, 2003). However, mechanistic studies demonstrating a direct role of PPAR in modulating expression of these gene products are lacking. In contrast, ligand activation of PPAR was found to increase, rather than inhibit, aromatase expression (Liu et al., 1996a,b), and administration of DEHP actually increases serum estradiol in vivo (Eagon et al., 1994). In males, there is good evidence that PPAR mediates phthalate-induced reproductive toxicity as DEHP exposure causes testicular toxicity and impaired spermatogenesis in wild-type mice, but not in similarly treated PPAR-null mice (Ward et al., 1998). However, PPAR-null mice maintained on a DEHP diet after all wild-type mice had died, also exhibit signs of testicular toxicity including tubule lesions and diffuse tubular aspermatogenesis (Ward et al., 1998). This suggests that DEHP can also lead to male reproductive toxicity through mechanisms that do not require the PPAR. This also emphasizes the need for animal models that do not exhibit PPAR-induced liver toxicities so that extrahepatic toxicities can be examined after long periods of ligand exposure that would otherwise cause morbidity and mortalities due to hepatocarcinomas (see below).
Collectively, these studies indicate that ligand activation of PPAR is not likely a general mechanism of action leading to developmental toxicity. However, it is still possible that some weak PPAR ligands could cause developmental defects by activating PPAR. Many questions remain regarding the role of PPAR in development, and whether or not this receptor mediates developmental/reproductive toxicity induced by some environmental toxicants, whether there are species differences in the responses resulting from exposure to these chemicals, and whether the environmental concentrations are capable of activating the receptor.
Hepatotoxicity
The most classic hallmark of liver toxicity induced by activation of PPAR is hepatocarcinogenesis. However, this subject has been extensively reviewed and will not be considered in detail. It is worth noting though, that numerous studies examining hepatocarcinogenesis caused by ligand activation of PPAR provide excellent examples demonstrating receptor-specificity, species differences in receptor activity, and relative ligand binding/activation, which are critical factors that mechanistically influence this ligand/receptor-mediated effect (reviewed in Klaunig et al., 2003; Peters et al., 2005). Other than the pleiotropic response associated with PPAR ligand-induced liver cancer (peroxisome proliferation, increased hepatocellular proliferation, hepatomegaly, increased expression of numerous target genes, etc.), there is little evidence suggesting that PPAR ligands are hepatotoxicants. No increases in serum levels of alanine aminotransferase or aspartate aminotransferase are observed in rats treated with clofibrate (Tanaka et al., 1992), and humans treated with fenofibrate only exhibit marginal increases (<3-fold) in these markers of liver toxicity (Balfour et al., 1990).
In contrast to the well-documented hepatocarcinogenic effect of PPAR ligands, there is also evidence that PPAR ligands can be used to prevent liver toxicity. Treatment of ethanol-fed mice with a PPAR agonist can reverse ethanol-induced fatty liver despite continued ethanol consumption (You and Crabb, 2004), and PPAR-null mice exhibit exacerbated fatty liver and symptoms of cirrhosis in response to chronic ethanol consumption (Nakajima et al., 2004). Activation of PPAR also influences acute liver toxicity. Pre-treating mice or hepatocytes with PPAR ligands can prevent acute hepatotoxicity induced by acetaminophen, bromobenzene, carbon tetrachloride, and chloroform (Anderson et al., 2004; Manautou et al., 1994, 1998; Nicholls-Grzemski et al., 1992; Shankar et al., 2003). That this protective effect of PPAR ligands against acute liver damage requires the receptor was demonstrated by studies showing the lack of protection in PPAR-null mice (Anderson et al., 2004; Chen et al., 2000; Shankar et al., 2003). Interestingly, caloric restriction can protect against thioacetamide-induced liver toxicity, and this effect requires PPAR as exacerbated liver toxicity is found in similarly treated PPAR-null mice (Corton et al., 2004). These findings suggest that although chronic ligand activation of PPAR in liver can lead to hepatocarcinogenesis in rodents, paradoxically, ligand activation can also prevent acute liver damage by alcohol and environmental liver toxicants. The protective effect of PPAR ligand activation in liver is likely the result of receptor-mediated induction of fatty acid catabolism and/or through anti-inflammatory/antioxidant activity associated with PPAR ligands, which could significantly reduce the ability of these agents (ethanol, carbon tetrachloride, etc.) to induce steatosis and related liver toxicity.
Recent findings also suggest that activation of PPAR may modulate liver toxicity by interfering with aryl hydrocarbon receptor (AhR)-dependent signaling. Expression of CYP1A2 mRNA and enzyme activity in liver are both decreased in rats treated with ciprofibrate (Gallagher et al., 1995). Similarly, decreased expression of CYP1A1 and CYP1A2 mRNA and protein is found rat liver after clofibrate treatment, and this effect appears to be due to reduced turnover of the AhR that mediates induction of CYP1A1 and CYP1A2 (Shaban et al., 2004). These combined observations suggest that PPAR ligands could potentially inhibit bioactivation and/or detoxification of chemical carcinogens/toxicants catalyzed by CYPs through mechanisms that are not clearly defined. However, another recent report suggests that PPAR ligands can induce CYP1A1. Increased mRNA encoding CYP1A1 is found in CaCo2, HepG2, and human keratinocytes in response to 200 µM Wy-14,643, and this increase parallels increased EROD activity observed in CaCo2 cells treated with Wy-14,643 (Seree et al., 2004). These findings suggest that there could be a species difference in the effect of PPAR ligands on the expression of CYP1A1; PPAR ligands inhibit CYP1A1/CYP1A2 expression in rats whereas PPAR ligands induce expression of CYP1A1 in human cell lines. Further work is necessary to determine whether PPAR ligand-induced changes in CYP1A1/CYP1A2 expression modulate bioactivation and/or detoxification of chemical toxicants/carcinogens.
Lastly, fibrate therapy is also associated with cholelithiasis (gallstones). Humans treated with clofibrate, bezafibrate, or fenofibrate have a significantly increased incidence of cholelithiasis (Bateson et al., 1978; Caroli-Bosc et al., 2001; Mazzella et al., 1990; Raedsch et al., 1995). Increased cholelithiasis is not consistently observed with all PPAR ligands as etofibrate and gemfibrozil treatment are not as lithogenic as compared to bezafibrate or clofibrate, respectively (Mazzella et al., 1990; Raedsch et al., 1995). One mechanism that may underlie the increased incidence of cholelithiasis is that fibrates can increase the lithogenicity of bile acids, due to increased biliary output of cholesterol and reduced output of bile acids, which can be measured clinically by the cholesterol saturation index (Carey, 1978). Bezafibrate, fenofibrate, and ciprofibrate all increase the cholesterol saturation index indicative of increased lithogenicity (Angelin et al., 1979; Leiss et al., 1986; Palmer, 1985). The possibility that the increased lithogenicity of bile that leads to increased cholelithiasis is mediated by PPAR is suggested by the finding that reduced bile output by ciprofibrate and bezafibrate requires a functional PPAR in mice (Hays et al., 2005; Post et al., 2001). However, since the relative ability of different PPAR ligands to cause cholelithiasis varies, other mechanisms contributing to this effect cannot be excluded.
Muscle Toxicity
Fibrate administration can cause myopathy in humans, and in rare cases lead to rhabdomyolysis. Myopathies and/or rhabdomyolysis have been reported to occur in patients treated with clofibrate (Bridgman et al., 1972; Langer and Levy, 1968; Rush et al., 1986), fenofibrate (Alsheikh-Ali et al., 2004; Barker et al., 2003; Blane and Pinaroli, 1980; Clouatre et al., 1999), bezafibrate (Gorriz et al., 1995; Heidemann and Bock, 1981; Kanterewicz et al., 1992; Rumpf et al., 1984; Vita et al., 1993), ciprofibrate (Bourrier et al., 1990; Delangre et al., 1990), and gemfibrozil (Bermingham et al., 2000; Chow and Chow, 1994; Layne et al., 2004; Magarian et al., 1991; Zimetbaum et al., 1991). The mechanisms underlying fibrate-induced myopathy or rhabdomyolysis are uncertain. There is some evidence that the mechanism involves the relative concentration of the drug in blood, since there are notable differences in the relative incidence of this toxicity induced by different fibrates and the fact that myopathy is more common in patients with kidney failure or hypoalbuminemia (Hodel, 2002). This toxicity in humans underscores the efforts to develop new high affinity ligands for PPAR. It is of interest to note that increased expression of lipoprotein lipase in skeletal muscle leads to severe myopathy in mice (Levak-Frank et al., 1995), since lipoprotein lipase is a known PPAR target gene (Heller and Harvengt, 1983; Schoonjans et al., 1996). Whether or not PPAR is required to mediate myopathy or rhabdomyolysis has not been examined to date.
Extra-hepatic Carcinogenesis
In addition to the well-described role of PPAR ligands in causing hepatocellular carcinomas in rodents, PPAR ligands have also been linked to other malignancies. The "Tumor Triad" has been described in rats treated with PPAR ligands since they result in liver cancer, Leydig cell tumors, and pancreatic acinar cell tumors. However, only nine of fifteen PPAR ligands examined to date (clofibrate, DEHP, HCFC-123, fenofibrate, gemfibrozil, methylclofenapate, PFOA, tibric acid, and Wy-14,643) result in the Tumor Triad, while the other six (cinnamyl anthranilate, nafenopin, butyl benzyl phthalate, DINP, perchloroethylene, and trichloroethylene) only result in a subset of carcinogenic endpoints (e.g., liver cancer and pancreatic acinar cell tumors but not Leydig cell tumors, etc.) (reviewed in Klaunig et al., 2003). Additionally, Leydig cell tumors and pancreatic acinar cell tumors have only been reported to occur in rats, and have not been observed in mice; although the number of chemicals examined in mice to date is less as compared to those examined in rats (reviewed in Klaunig et al., 2003). Hypothetical mechanisms have been postulated that could mediate PPAR ligand-induced Leydig cell and pancreatic acinar cell tumors via a PPAR-dependent pathway, but this has not been clearly demonstrated to date. If PPAR were required to mediate PPAR ligand-induced Leydig cell and/or pancreatic acinar cell tumors, this would suggest that there could also be a species difference between rats and mice since mice appear to be refractory to these effects.
Ligand activation of PPAR can protect against cisplatin-induced acute renal failure. There are a number of molecular events that could contribute to this protection including induction of fatty acid oxidation (Li et al., 2004b), repression of endonuclease G that leads to reduced apoptosis and necrosis (Li et al., 2004a), and anti-inflammatory activities of Wy-14,643 (Li et al., 2005b). That these effects are receptor-dependent was demonstrated by the use of PPAR-null mice, which are not protected from cisplatin-induced acute renal failure after administration of PPAR ligands.
There is one report suggesting that PPAR ligands may potentiate breast cancer cell proliferation as well. Wy-14,643 and clofibrate treatment increases the growth of human breast cancer cell lines (Suchanek et al., 2002). Whether or not this effect is mediated by PPAR and/or other receptor-independent mechanisms has not been determined to date.
In many cases, it has not been established whether hepatic PPAR is responsible for non-hepatic toxicity and carcinogenicity. The severe hepatomegaly and carcinogenicity found in rats and mice could affect other tissues. This is especially noteworthy, since humans do not exhibit the liver toxicities found with the rodent models treated with PPAR ligands. It is thus important to find suitable animal models to predict potential toxicities in humans. Ideally, a liver-specific PPAR-null mouse could be used for this purpose but to date, this model has not been available. However, a PPAR-humanized mouse was recently produced that does not exhibit the liver toxicities found in wild-type mice that might be of value in determining off-target toxicities of PPAR ligands (Cheung et al., 2004).
|
PPAR LIGANDS
|
|---|
|
TOPABSTRACTINTRODUCTIONPPAR{alpha} LIGANDSPPAR{gamma} LIGANDSPPARß/{delta} LIGANDSCONCLUSIONS AND FUTURE...REFERENCES |
|---|
Thiazolidinediones are the most widely studied PPAR
ligands, and were found to improve insulin sensitivity in diabetic animals more than ten years before PPAR was cloned. Troglitazone (Rezulin) was the first drug approved for this use, followed by rosiglitazone (Avandia) and pioglitazone (Actos). The mechanism of action of thiazolidinediones was not known until 1995 when Lehmann et al. reported that thiazolidinediones were high affinity ligands for PPAR (Lehmann et al., 1995). Endogenous ligands for PPAR have also been described (e.g., 15deoxy
12,14prostaglandin J2, 15-PGJ2), and additional synthetic ligands for the receptor are under development. However, acute liver toxicity found in patients prescribed troglitazone led to the withdrawal of Rezulin from the market along with peaked concerns about the safety of PPAR activators. Phthalate monoesters are known to activate PPAR, but recent findings demonstrate that this class of chemicals can also activate PPAR (Bility et al., 2004; Hurst and Waxman, 2003). To date, it has not been clearly established whether PPAR mediates any of the toxicities associated with PPAR ligands. Surprisingly, while PPAR ligands can significantly improve insulin sensitivity and PPAR is required to mediate this effect, the precise mechanisms leading to improved insulin sensitivity are uncertain (Knouff and Auwerx, 2004; Rangwala and Lazar, 2004). PPAR, while abundantly expressed in adipose tissue, is not highly expressed in skeletal muscle and is almost absent in normal liver, tissues where appreciable alterations in glucose transport and fatty acid metabolism are found after treatment with PPAR ligands, suggesting that PPAR ligands mediate their effects by secondary changes induced by PPAR activity in adipose tissue such as production of adipokines such as leptin, resistin, and adiponectin (Bouskila et al., 2005). Whether or not this type of regulation also influences PPAR ligand-induced toxicities is unknown.
Reproductive and Developmental Toxicity
It is of interest to note that there is strong evidence that PPAR is required for development since targeted disruption of this gene results in embryo lethality, due in part to defective placental development (Barak et al., 1999; Kubota et al., 1999; Rosen et al., 1999). This demonstrates that the receptor itself is essential for developmental processes, and suggests that ligand activation during development could influence specific molecular events modulated by PPAR. There are limited published reports describing the outcome of developmental or reproductive toxicity testing for specific PPAR ligands. However, rosiglitazone and pioglitazone are classified as category B drugs for use during pregnancy since animal studies have shown no evidence of altered development and limited analysis in humans are consistent with these findings (Kalyoncu et al., 2005; Yaris et al., 2004). It has been hypothesized that phthalate monoesters can lead to anovulation by activating PPAR leading to inhibition of aromatase expression (Lovekamp-Swan and Davis, 2003), a mechanism that has also been linked to activation of PPAR as described above. By contrast, the more specific PPAR ligand troglitazone has been shown to improve ovulation in humans with polycystic ovary syndrome (Azziz et al., 2001; Mitwally et al., 1999), suggesting that ligand activation of PPAR leads to changes that facilitate ovulation rather than inhibit this process. The mechanisms underlying these effects have not been defined, and it is currently unclear whether PPAR is required to mediate these effects or whether there are species differences between these responses. PPAR may also regulate trophoblast differentiation, as troglitazone exposure leads to differentiation of human trophoblasts (Schaiff et al., 2000), and this may be mediated by PPAR-dependent regulation of mucin expression (Shalom-Barak et al., 2004). Interestingly, the "natural" PPAR ligand 15-PGJ2 diminishes trophoblast differentiation, suggesting that there could be ligand-specific effects. Whether this is due to differential recruitment of co-effectors (e.g., histone acetyltransferases, histone deacetylases, etc.) has not been determined to date. Further, whether exogenous ligand activation of PPAR during placental development impairs embryo or fetal development has not been reported in the literature to date.
Hepatotoxicity
One of the most studied toxic effects of a PPAR agonist is hepatotoxicity. Early clinical trails with troglitazone indicated elevations in serum alanine aminotransferase (ALT) of greater than three times the upper normal range in 1.9% of 2510 of treated patients (Watkins and Whitcomb, 1998). Subsequent case reports demonstrated liver toxicity in humans treated with troglitazone (Gitlin et al., 1998; Herrine and Choudhary, 1999; Neuschwander-Tetri et al., 1998). Additionally, two case reports suggest that liver toxicity may occur with rosiglitazone, although clinical trials examining hepatotoxicity demonstrated similar low incidence of elevated ALT levels (Al-Salman et al., 2000; Forman et al., 2000). The mechanism of PPAR ligand-induced liver toxicity is poorly understood. Since troglitazone induces CYP3A4 (Dimaraki and Jaffe, 2003; Ramachandran et al., 1999), it has been hypothesized that potentially toxic quinones derived from CYP3A4-dependent metabolism could cause liver damage (Neuschwander-Tetri et al., 1998; Yamamoto et al., 2002). Since clinical trials demonstrate a large difference in the ability of troglitazone to cause liver toxicity versus rosiglitazone or pioglitazone (Al-Salman et al., 2000; Watkins and Whitcomb, 1998), the hepatotoxicity may result from genetic differences in the population in the induction of CYP3A4 or the metabolic activation of troglitazone that results in these rare cases of hypersensitivity. It should also be noted that under normal circumstances, PPAR is not expressed to any appreciable degree in normal liver; however, it is expressed at functionally significant levels in steatotic livers (Matsusue et al., 2003). Thus, it cannot be ruled out that hepatotoxicity may be the result of abnormal target gene expression in patients with fatty livers.
Surprisingly, while PPAR ligands can cause liver toxicity, recent findings suggest that PPAR ligands can protect against liver damage. Thiazolidinedione administration improves the histological features of nonalcoholic steatohepatitis and restores serum ALT levels in patients with nonalcoholic steatohepatitis (Caldwell et al., 2001; Neuschwander-Tetri et al., 2003). Pioglitazone can also prevent carbon tetrachloride-induced fibrosis (Kon et al., 2002), and acute liver toxicity from ethanol and lipopolysaccharide exposure (Ohata et al., 2004). One mechanism that may contribute to the protective effect of PPAR ligands against liver toxicity, in particular fibrogenesis, is inhibition of stellate cell activation. PPAR ligands can inhibit stellate cell activation, including suppressing expression of collagen and smooth muscle actin, and these effects appear to require PPAR as they do not occur in the presence of a PPAR antagonist (Miyahara et al., 2000). Given the paradoxical observations that PPAR ligands may cause hepatotoxicity and yet protect against chemical-induced liver toxicity, this demonstrates that the mechanisms underlying the effects of PPAR ligands in liver are largely unknown. Examination of liver cell-specific PPAR-null mice could be useful for delineating the role of PPAR in liver.
Cardiac Toxicity
The effect of PPAR ligands on heart function is somewhat controversial. Thiazolidinediones cause cardiac hypertrophy and administration of a number of different thiazolidinediones including rosiglitazone to F344 rats causes a significant increase in heart weight (Oguchi et al., 2000). Deleting PPAR expression in cardiomyocytes demonstrates that cardiac hypertrophy does not always require PPAR expression in these cells, as hypertrophy (albeit somewhat different) was observed in both wild-type and cardiac myocyte-specific PPAR-null mice (Duan et al., 2005). In other models, PPAR ligands inhibit cardiac myocyte hypertrophy (Asakawa et al., 2002; Yamamoto et al., 2001), which is consistent with the ability of PPAR to interfere with NF-B signaling (Wang et al., 2002). Since congestive heart failure can be negatively influenced by cardiac hypertrophy, these opposing findings suggest that PPAR ligands may or may not be of benefit to diabetics using thiazolidinediones, as diabetics are at particularly high risk for congestive heart failure. The specific mechanism mediating changes in cardiac myocyte hypertrophy has not been elucidated, but could involve PPAR/NF-B signaling. There is also evidence that PPAR ligands increase the incidence of heart failure in type II diabetics (Delea et al., 2003). Thus, there is current concern regarding the administration of PPAR ligands to type II diabetics because they are known to cause peripheral edema, which could exacerbate congestive heart failure (Nikolaidis and Levine, 2004). Several mechanisms have been postulated that could lead to PPAR ligand induced edema, which in turn contributes to cardiac toxicity. For example, PPAR ligands have calcium channel blocking activity that may or may not be receptor-dependent (Song et al., 1997; Zhang et al., 1994). However, recent work strongly suggests that the edema resulting from PPAR ligands is mediated by a receptor-dependent mechanism. Deleting PPAR from the collecting duct of the nephron eliminates PPAR ligand-induced weight gain due to water accumulation (Guan et al., 2005b; Zhang et al., 2005), and this effect appears to be mediated by PPAR-dependent regulation of epithelial sodium channel expression (Guan et al., 2005b). This suggests that PPAR-dependent regulation of critical regulatory proteins in the kidney secondarily influences fluid retention that modulates cardiac function. It will be of great interest to determine whether PPAR ligand-induced edema is altered in individuals with mutations in PPAR (Barroso et al., 1999; Gurnell et al., 2000) or splice variants of PPAR (Sabatino et al., 2005).
Carcinogenesis
The effect of PPAR ligands on carcinogenesis is controversial. There are a number of studies suggesting that ligand activation of PPAR can potentiate tumorigenesis (Table 3). However, it is unknown if the mechanisms mediating these effects are receptor-dependent or -independent pathways, or whether there are species differences as suggested by the long-term troglitazone bioassay (Heaney et al., 2002), and/or whether they are due to ligand-specific effects. Further complicating this issue is the larger body of evidence that PPAR ligands induce differentiation, inhibit cell growth, induce apoptosis and/or inhibit tumorigenesis in a variety of cancer models (Table 3). Similar to the uncertainty associated with the mechanisms that potentiate tumorigenesis by PPAR ligands, the mechanisms leading to attenuation of carcinogenesis by PPAR ligands are unclear. While it is known that PPAR ligands can induce differentiation, inhibit cell growth, induce apoptosis and/or inhibit tumorigenesis in a variety of cancers, there is only limited evidence that these effects are mediated by both receptor-dependent and receptor-independent mechanisms. For example, the use of null mice to demonstrate the requirement of chemopreventive activity of PPAR ligands is lacking for the most of these models. However, heterozygous PPAR-null mice have been used to demonstrate that PPAR is required to mediate inhibition of gastric cancer in a mouse model (Lu et al., 2005). This mouse model has also been used to demonstrate a potential tumor suppressive effect of PPAR in azoxymethane-induced colon cancer (Girnun et al., 2002) and in 7,12-dimethylbenz[a]anthracene-induced mammary and skin cancer (Nicol et al., 2004). The link between PPAR and carcinogenesis has not been established but clues are beginning to emerge. In colon, ß-catenin expression, known to be involved in colon carcinogenesis, is enhanced in PPAR heterozygous mice (Girnun et al., 2002). Expression of mRNA encoding differentiation-related target genes (keratin 19, keratin 20 and Kruppel-like factor 4 [KLF4]) that is induced by PPAR ligands in colonic epithelium is significantly lower in PPAR heterozygotes (Drori et al., 2005). This suggests that PPAR mediates the induction of differentiation-related target genes by binding to PPREs and modulating transcription as previously described (Fig. 1A). There is also evidence that the induction of KLF4 by PPAR ligands may require PPAR, but is also mediated by the MEK/ERK pathway in colon cancer cells (Chen and Tseng, 2005). It also remains a possibility that the anti-inflammatory effects of PPAR ligand could be involved in suppression of carcinogenesis (Ricote et al., 1998a,b, 1999). Indeed, cytokine expression is known to enhanc