Peroxisome proliferator-activated receptor-α and liver cancer: where do we stand?
- First Online:
- Cite this article as:
- Peters, J.M., Cheung, C. & Gonzalez, F.J. J Mol Med (2005) 83: 774. doi:10.1007/s00109-005-0678-9
- 451 Views
The peroxisome proliferator-activated receptor-α (PPARα), first identified in 1990 as a member of the nuclear receptor superfamily, has a central role in the regulation of numerous target genes encoding proteins that modulate fatty acid transport and catabolism. PPARα is the molecular target for the widely prescribed lipid-lowering fibrate drugs and the diverse class of chemicals collectively referred to as peroxisome proliferators. The lipid-lowering function of PPARα occurs across a number of mammalian species, thus demonstrating the essential role of this nuclear receptor in lipid homeostasis. In contrast, prolonged administration of PPARα agonists causes hepatocarcinogenesis, specifically in rats and mice, indicating that PPARα also mediates this effect. There is no strong evidence that the low-affinity fibrate ligands are associated with cancer in humans, but it still remains a possibility that chronic activation with high-affinity ligands could be carcinogenic in humans. It is now established that the species difference between rodents and humans in response to peroxisome proliferators is due in part to PPARα. The cascade of molecular events leading to liver cancer in rodents involves hepatocyte proliferation and oxidative stress, but the PPARα target genes that mediate this response are unknown. This review focuses on the current understanding of the role of PPARα in hepatocarcinogenesis and identifies future research directions that should be taken to delineate the mechanisms underlying PPARα agonist-induced hepatocarcinogenesis.
KeywordsPeroxisome proliferator-activated receptor-α (PPARα)HepatocarcinogenesisFibratesHuman cancer
The phenomenon of peroxisome proliferation was initially described in 1960 when it was discovered that administration of the hypolipidemic drug, clofibrate, resulted in hepatomegaly that was accompanied by an increase in the number and size of intracellular peroxisomes . Thus, clofibrate and other structurally diverse chemicals were termed “peroxisome proliferators” by Reddy and Krishnakantha, based on their ability to induce this effect . By 1980, Reddy and coworkers discovered that chronic, long-term administration of these chemicals to rats also resulted in liver cancer . These classic observations laid the foundation for the focus of future research to elucidate the mechanisms underlying the effect of peroxisome proliferators. The discovery of nuclear receptors aided this pursuit and it was soon hypothesized that the effects induced by peroxisome proliferators were the result of a receptor-mediated mechanism . It was not until 1990 that a receptor was cloned and shown to be activated by this class of chemicals  and thus termed the peroxisome proliferator-activated receptor (PPAR). Soon after this discovery, two additional related receptors were cloned that shared significant homology in their encoded DNA and protein structures with PPAR; these are now known as PPARα, PPARβ/δ, and PPARγ [6, 7].
The PPARs belong to the superfamily of ligand-activated, soluble nuclear receptors. In response to ligand activation, PPARα heterodimerizes with another nuclear receptor, RXRα, and after conformational changes that lead to recruitment of transcriptional coregulators, the receptor complex modulates target gene expression resulting in diverse physiological responses . Endogenous fatty acids and fatty acid derivatives were shown to bind to and activate PPARs, suggesting that these compounds represent potential natural endogenous ligands [9–12]. Interestingly, genetic disruption of the peroxisomal β-oxidation enzyme acyl CoA oxidase (ACO) leads to spontaneous and sustained activation of PPARα, thus providing evidence that fatty acid derivatives represent natural ligands for PPARα . This suggests that under normal circumstances, PPARs modulate lipid homeostasis through their activation by endogenous signaling molecules. For example, ligand activation of PPARα leads to increased expression of target genes that regulate fatty acid mobilization from adipose tissue and increased hepatic fatty acid catabolism mediated by both peroxisomal and mitochondrial enzymes . Because of this key regulatory mechanism, PPARα is known to be a central modulator of lipid homeostasis that occurs in mammalian species and is activated in response to fasting when lipids need to be mobilized as an energy source. In addition to and possibly related to this pathway, PPARα has also been implicated in controlling a number of other physiological processes including cell proliferation, apoptosis, inflammation, oxidative stress, and others [14, 15]. Given this central role of PPARα in such critical physiological processes including the acquisition of energy from fatty acid substrates in the form of high-energy phosphate, it is of great interest to understand why prolonged activation of this pathway also leads to hepatocarcinogenesis in susceptible species. This is of great concern since new ligands are under development that target PPARs in an effort to inhibit or prevent diseases including dyslipidemias, diabetes, atherosclerosis, and cancer [14, 15]. The focus of this review is to update our current understanding of the role of PPARα in liver cancer.
Physiological role of PPARα
Numerous chemicals have been shown to bind to and activate PPARα and to be hepatocarcinogenic in rodent models. The fibrate class of hypolipidemic drugs and phthalates represents two of the more extensively studied classes of PPARα agonists that lead to rodent hepatocarcinogenesis. Drugs that reduce serum lipids do so through the receptor-mediated mechanism described above and may mimic endogenous ligands. One could argue that structural similarities with fibrates and/or fatty acids explains why unrelated chemicals like phthalates activate the same receptor. The fact that lipid lowering occurs in response to fibrate (PPARα agonists) treatment in both rodents and humans demonstrates that PPARα likely has similar roles in both species. It is less clear why this pathway leads to liver cancer in rodents, although the evidence for many of the key events described below indicates that some of these pathways may not occur in humans. This raises the question of what is the mode of action for the carcinogenic properties of PPARα agonists.
Mode of action
There are series of key events that are thought to mediate the hepatocarcinogenic effects of PPARα agonists: (1) receptor binding and activation, (2) induction of key target genes, (3) increased cell proliferation and inhibition of apoptosis, (4) oxidative stress that either results in oxidative damage to DNA and/or contributes to increased signaling for cell proliferation, and (5) clonal expansion of initiated tumor cells . The strength of evidence supporting a causal relationship for each of these key regulatory steps to the underlying mechanisms of PPARα agonist-induced hepatocarcinogenesis and the specificity of these events for this mode of action is varied. In this context, the key events with direct links to PPARα will be briefly discussed.
The strength of evidence supporting a causal relationship between receptor activation and hepatocarcinogenesis is strong because studies using null mice definitively demonstrate that without the receptor, liver cancer does not occur after long-term administration of PPARα agonists [28, 29]. However, these studies have only been performed with two representative chemicals and the duration of ligand exposure was only for approximately 1 year. The complete lack of tumors in PPARα-null mice is a striking observation and strongly supports a causal link to liver cancer induced by PPARα agonists. It is also worth noting that all chemicals that cause induction of peroxisome proliferation, induction of fatty acid metabolizing enzymes, hepatomegaly, and ultimately liver cancer after prolonged administration are typically able to bind to and activate PPARα. This characteristic also supports the weight of evidence suggesting that ligand activation of PPARα is specific for this class of chemicals. A number of approaches have been used to measure receptor–ligand binding, which typically involve direct competitive binding assays and/or indirect cofactor recruitment [9, 11, 12, 31, 32]. Receptor activation is more commonly measured using reporter constructs containing either PPRE-luciferase with or without cotransfection of a PPARα expression vector in the presence or absence of ligand. Using these methods, a broad range in the ability of different chemicals to bind to and activate PPARα have been reported, ranging from relatively weak agonists (phthalate monoesters) to strong agonists (Wy-14,643, GW2331).
The hallmark feature resulting from administration of PPARα ligands in rats and mice is peroxisome proliferation. The increase in peroxisome volume and number is due in part to the increased expression of peroxisomal enzymes including ACO, bifunctional enzyme, and thiolase, which are targeted to the peroxisome by a relatively conserved sequence in their carboxy termini [33, 34]. Although it is clear that peroxisome proliferation occurs in response to treatment with PPARα agonists, the role of this organelle in the hepatocarcinogenic response appears more associative than causative. This is based on the observation that the degree of peroxisome proliferation observed in rats fed peroxisome proliferators does not correlate well with tumorigenicity , although it is plausible that there exists a threshold above which peroxisome proliferation could be of sufficient magnitude to lead to oxidative damage to macromolecules. Therefore, peroxisome proliferation is a common feature found in rodents treated with PPARα agonists and represents a highly associative endpoint .
In response to ligand activation, a number of lipid-metabolizing enzymes are induced, including peroxisomal ACO. There are also several other peroxisomal oxidases that could also lead to increased production of H2O2 as a result of peroxisome proliferation . Most importantly, administration of PPARα agonists causes a marked increase in ACO expression and its resultant generation of intracellular H2O2 through fatty acid oxidation, which under normal non-induced circumstances can be detoxified by catalase. Catalase expression increases only approximately twofold in response to PPARα agonists, and this increase may be insufficient to detoxify the large increase in H2O2 resulting from the relatively large increase in ACO activity , although catalase has a very high turnover number. Increased cellular H2O2 could potentially react with metals and generate highly reactive hydroxyl radicals or react with lipids and produce lipid peroxides. Thus, the induction of ACO increases intracellular levels of H2O2 and could lead to (1) oxidative stress (which could influence cell proliferation), and/or (2) the generation of lipid peroxides or free radicals that could damage DNA or proteins . Increased oxidative stress resulting from elevated expression of ACO could activate signaling pathways that lead to increased cell proliferation as will be discussed later. It is noteworthy that lipofuscin accumulates in rodent liver after treatment with PPARα agonists, a finding that is consistent with higher oxidative damage to proteins [37, 38], although the mechanism of lipofuscin formation is not well understood [38, 39]. Treating rodents with antioxidants can also inhibit PPARα agonist-induced liver cancer [40, 41], but this change does not always correlate with reduced markers of oxidative damage . Interestingly, administration of PPARα agonists can also lead to decreased levels of antioxidant enzymes [42, 43]. The evidence demonstrating that increased DNA damage occurs in response to PPARα agonists due to oxidative stress is unclear. Some studies report little or no change in markers of oxidative DNA damage (e.g., 8-hydroxydeoxyguanosine residues) in response to PPARα agonists (reviewed in [30, 44]), although there are also reports suggesting that this could occur . Increased expression of long-patch base excision DNA repair enzymes, proteins that remove oxidized DNA lesions, is found in response to PPARα agonists, which is consistent with the hypothesis that increased PPARα-dependent expression of ACO and subsequent generation of H2O2 lead to oxidative DNA damage . Studies using an ACO-null mouse fed PPARα agonists could effectively examine this hypothesis; unfortunately, ACO-null mice exhibit spontaneous liver hepatocarcinogenesis, which prevents the long-term analysis . Nevertheless, it would still be of interest to determine the effect of PPARα-agonist treatment on relative cell proliferation, inasmuch as increased oxidative stress induced by ACO could theoretically enhance cell proliferation signaling. Unfortunately, the relationship between oxidative stress induced as a result of increased ACO expression and downstream signaling that could contribute to increased cell proliferation is currently unclear (reviewed in ). Oxidative stress resulting from increased ACO expression could also participate in oxidative signaling induced from activated Kupffer cells and lead to increased cell proliferation (Fig. 2). It was hypothesized that PPARα-independent activation of Kupffer cells leads to increased cytokine signaling that contributes to increased cell proliferation , but the fact that PPARα-null mice exhibit no changes in replicative DNA synthesis argues against this idea . Furthermore, TNFα, which is one of the main cytokines produced from Kupffer cells, appears not to be involved in this putative signaling because TNFα-null and TNFα receptor-null mice are not resistant to PPARα agonist induced changes in liver cell proliferation as would be predicted [50, 51]. Thus, there remains some uncertainty in establishing a causal link between increased ACO expression (and associated oxidative stress) and liver cancer resulting from long-term treatment with PPARα agonists, but there is a strong weight of evidence showing an associative relationship with high specificity .
Activation of PPARα also leads to increases in hepatocellular proliferation and inhibition of apoptosis, and when this occurs in a DNA-damaged cell, it is thought to lead to proliferation of initiated cells that ultimately progress to a liver tumor. That PPARα is required to mediate increased cell proliferation and inhibition of apoptosis is strongly supported by observations in PPARα-null mice that are refractory to all of these changes in response to long-term ligand-feeding studies [28, 29, 52]. Whereas it is clear that PPARα agonists lead to increased cell proliferation and inhibition of apoptosis, the specific target genes mediating these events remain unidentified (Fig. 2). Increased cell proliferation and inhibition of apoptosis are clearly causally linked to PPARα agonist treatment and hepatocarcinogenesis. In their absence, it would be highly unlikely that liver cancer would result, since they are required to allow for a cell with DNA damage and gene mutations to proliferate and ultimately form a liver tumor. Since increased cell proliferation can influence both initiation and promotion events, the precise role of these changes is less clear. However, the strength of the evidence is strong that causally link changes in cell proliferation and apoptosis to PPARα agonist-induced hepatocarcinogenesis .
As noted above, it is well established that PPARα mediates peroxisome proliferator-induced liver cancer in rodents . However, there is considerable controversy as to whether the administration of drugs that are ligands for PPARα to humans causes liver cancer. This is significant because PPARα agonists such as gemfibrozil, fenofibrate, bezafibrate, and ciprofibrate have been prescribed to human patients for the treatment of hyperlipidemias for more than 30 years . Yet, there is no epidemiological evidence that these drugs lead to an increased risk for liver cancer (see below), suggesting that humans are resistant to fibrate drug-induced hepatocellular carcinomas. Indeed, a number of experimental observations suggest that there is a species difference between rodents and humans in the response to PPARα agonists, including differences in receptor activation, peroxisome proliferation, changes in cell proliferation and/or apoptosis, and induction of target genes. However, it should be noted that fibrates are relatively weak PPARα ligands, and in humans, cancers typically have a latency period of more than 20 years.
Activation of PPARα by agonists occurs in both rodents and humans, and is the basis for the therapeutic use of fibrates to treat dyslipidemias. However, higher concentrations of ligands may be required to fully activate the human PPARα as compared to the mouse PPARα, based on observations made with reporter gene assays examining receptor activation [11, 15, 54–58]. At this point it is worth reiterating that there is a wide range of receptor affinities for different PPARα ligands ranging from relatively weak (e.g., phthalate monoesters) to relatively strong (e.g., Wy-14,643, GW2331).
Marked hepatic peroxisome proliferation is always observed in rodents fed PPARα agonists, but the limited reports from humans suggest that this phenomenon does not occur in human liver. For example, a three- to fivefold increase in hepatic peroxisome number is found in rats treated twice daily with clofibrate at a dose (200 mg/kg) equivalent to about 14 times the human therapeutic dose . Hanefeld et al., who examined the similar phenomenon in humans treated with clofibrate (≤7 years), found only a marginal 50% increase in liver peroxisome number with an insignificant change in peroxisome volume . In contrast, others have found no change in hepatic peroxisome proliferation in humans treated with clofibrate, gemfibrozil, or fenofibrate for about 1–7 years [61–64]. It is important to note that for most of these studies, drug treatment was stopped for about 12 h before biopsy, which in addition to the half-life of peroxisomal proteins, could have influenced these findings. Comparative studies examining other species, in particular nonhuman primates, also suggest that there may be a difference in the extent of peroxisome proliferation in response to PPARα agonists between rodents and humans, but some reports do not support this view (reviewed in [30, 65]). For example, some have reported that the induction of peroxisome proliferation and/or increased markers associated with this response do not occur in nonhuman primates treated with bezafibrate, ciprofibrate, nafenopin, or DEHP [66–69], whereas others have shown the presence of peroxisome proliferation in this species treated with ciprofibrate or DL-040 [70, 71]. Some of the disparity in these reports may be due to differences in the concentration of PPARα agonists used. A recent study confirmed the lack of peroxisome proliferation in nonhuman primate liver from animals treated with therapeutic doses of either fenofibrate or ciprofibrate for 15 days, but a twofold increase in peroxisome number at doses four to five times the therapeutic exposure was observed . Thus, as compared to rodent models in which three- to fivefold increases in peroxisome number are found in response to PPARα agonist treatment, comparable increases in peroxisome proliferation are not reportedly found in humans and nonhuman primates.
PPARα-mediated peroxisome proliferation is also associated with an increase in the expression of peroxisomal ACO and other enzymes present in this subcellular organelle. As described above, the induction of ACO could lead to increased intracellular oxidative stress. A rapid and robust increase in ACO expression (15-fold) is found in rat liver within 24 h of treatment with ciprofibrate (250 mg/kg) or clofibrate (500 mg/kg) , and this effect is essentially sustained in rodent liver with prolonged administration of these or related PPARα agonists . In contrast, despite the presence of measures indicative of PPARα activation in human patients treated with fenofibrate, bezafibrate, or gemfibrozil for 8 weeks (e.g., reductions in plasma low-density lipoprotein (LDL) cholesterol and triglycerides and increased expression of hepatic Apo AI), no significant increase in ACO expression in liver was found in patients treated with the different PPARα agonists as compared to controls . This finding is consistent with other comparative studies examining ACO expression in nonhuman primates, which also exhibit no or marginal increases in ACO expression in response to administration of PPARα agonists (reviewed in [30, 65]). In addition, several studies have comparatively examined this effect using rodent and human primary hepatocytes or hepatoma cell lines exposed to this class of chemical. Collectively, these studies typically report that rodent-derived cells or cell lines exhibit large increases in ACO expression in contrast to low or no expression of this enzyme in cultured primary hepatocytes or cell lines from human liver (reviewed in [30, 65]). Consistent with reports examining peroxisome proliferation, the induction of ACO expression in human liver by PPARα ligands appears to be considerably less than the marked ≥15-fold induction observed in rodent liver that occurs within 24 h of treatment.
In addition to peroxisome proliferation, increased liver weight is also associated with administration of PPARα ligands to rodent models, and this effect is likely the result of both hypertrophy and hyperplasia (the latter due to both increased cell proliferation and inhibition of apoptosis). Liver weight dramatically increases in rodents within 1 day after treatment with PPARα ligands, becoming up to three to four times normal after long-term treatment . That this increase in liver weight is due in part to increased replicative DNA synthesis that accompanies induction of the S phase of the cell cycle is suggested by the observed increase in bromodeoxyuridine (BrdU) labeling index found in rodent liver treated with PPARα agonists [35, 38, 75, 76]. In addition, the increase in liver cell growth found with administration of PPARα agonists is also due in part to inhibition of apoptosis . Comprehensive in vivo evidence from humans treated with PPARα agonists examining their effect on liver hyperplasia that is causally linked to the hepatocarcinogenic effect of these chemicals is not available. However, there is one report from a human study showing that clinical hepatomegaly did not occur in 12 human patients treated with fenofibrate for 4–86 months . The majority of evidence suggesting that humans do not respond to PPARα agonists by increased hepatic cell proliferation is based on comparisons between nonhuman primates and rodent models and/or in vitro analysis using rodent and human primary hepatocytes or hepatoma cell lines exposed to PPARα ligands. Primary human hepatocytes do not represent the most appropriate model for examining cell proliferation inasmuch as they do not exhibit large increases in replicative capacity as compared to what is found in vivo , likely due to the absence of optimal growth factors and/or contributions made from nonparenchymal cells . Despite this disadvantage, the collective findings from this in vitro approach still show that rodent primary hepatocytes exhibit increased markers of cell proliferation after exposure to PPARα ligands, and human hepatocytes are refractory to this effect (reviewed in [30, 65]). Results from studies examining increases in liver weight and/or markers of cell proliferation in nonhuman primates are also consistent with the hypothesis that humans do not respond to the same extent as rodents (reviewed in [30, 65]). For example, administration of clofibrate, ciprofibrate, or fenofibrate for either 2 or 13 weeks caused no significant increases in relative liver weight in nonhuman primates when dosed at therapeutic levels [72, 79], although doses of ciprofibrate that were five to nine times higher than the therapeutic range did cause a significant increase in liver weight when given for 15 days . However, in contrast to studies in rodents, nonhuman primates treated with either fenofibrate or ciprofibrate exhibit no evidence of increased mitotic figures, Ki-67-positive cells, or mRNA markers of enhanced cell proliferation (e.g., proliferating cell nuclear antigen (PCNA), cyclin D, c-myc) in liver, even at doses four to nine times higher than the therapeutic range .
Epidemiological and clinical studies, although not conclusive, provide further support for the hypothesis that it is unlikely that exposure to PPARα agonists represents a human cancer hazard. Results from the Helsinki Heart Study did not find an association between gemfibrozil treatment and liver cancer [80, 81]. Similarly, no association between clofibrate treatment and liver cancer was observed in a study conducted by the World Health Organization [82–84]. These studies did not show an association between liver cancer and treatment with clofibrate or gemfibrozil, but there are limitations with these studies that will be discussed further. Lastly, in vivo analysis of nonhuman primates treated chronically with PPARα agonists have also revealed largely negative data sets, suggesting that this species may be refractory to PPARα ligand-induced liver cancer (reviewed in [30, 65]).
There are numerous studies that have focused on delineating the mechanism/mode of action for PPARα agonists, how these chemicals cause liver cancer in rodents, and whether humans are essentially refractory to the hepatocarcinogenic effect of these chemicals. Despite the wealth of information available on this subject, there are several important gaps in the data that pertain to PPARα and its role in the pleiotropic response to its agonists. In 2000, a variant of the human PPARα gene (L162V) was identified by one laboratory  and subsequently confirmed by two other independent laboratories [86, 87]. Interestingly, analysis of reporter activity using the L162V variant, which contains a mutation in the DNA-binding domain of the receptor, showed significantly greater ligand-induced activity as compared to the wild-type receptor [85, 86]. Furthermore, humans with this variant form of PPARα are reported to be more responsive to bezafibrate as shown by a more pronounced decrease in total serum cholesterol . Other studies have also suggested that this variant may be associated with differences in serum lipid profiles [86–91], but this relationship is not always found . In addition, another variant of PPARα found in a Japanese population is associated with alterations in serum lipid levels . Combined, these results give good reason to suggest that there is some heterogeneity in the human population with respect to the functional coding sequence of PPARα and some evidence suggesting that there could be functional differences in the receptor activity associated with these variants. To date, any potential relationship between these variants of PPARα and PPARα agonist-induced liver cancer has not been examined, but this clearly deserves further evaluation. In addition to polymorphic variants of PPARα, there is also a truncated splice variant of human PPARα that appears to be expressed in most humans and reportedly functions as a dominant negative . The functional significance of this variant and any potential interaction with wild-type PPARα, given recent observations showing increasingly complicated interactions between different nuclear receptors and other signaling pathways [95, 96], has never been extensively examined.
The expression level of PPARα in human liver has also been suggested to underlie the reported species differences in response to PPARα ligands. Analysis of 20 human liver lysates obtained from samples that were frozen within 10 h of isolation revealed relatively variable findings, but consistently showed significantly lower levels (10- to 20-fold) of PPARα in human liver as compared to mouse liver . The limitations from these studies include the following: (1) no measurement of other proteins susceptible to degradation was performed to demonstrate that protein quality was not significantly different between samples, given the length of time before freezing and (2) only 20 samples were examined. However, work by others is consistent with these findings and show that some humans may express appreciably lower levels of PPARα as compared to rodents [98, 99]. Interestingly, expression of PPARα in human liver appears to be variable and, at least in one case, may be comparable to rodent liver . Further work is necessary examining PPARα expression in human liver, with an emphasis on determining the relative presence of polymorphic proteins and truncated mutants and their functional significance.
Whereas expression levels of PPARα in liver may explain, in part, the observed species difference, recent work with “humanized” PPARα mice suggests that there are inherent differences in the receptor that may be more critical. Mice that express the human PPARα but lack expression of mouse PPARα (PPARα-humanized mice) exhibit a phenotype that is remarkably consistent with the reported literature describing a large species difference. For example, administration of Wy-14,643 to PPARα-humanized mice results in increased expression of ACO and other target genes known to regulate lipid homeostasis . These mice also exhibit lowering of serum lipids upon ligand treatment, similar to wild-type mice expressing the mouse PPARα. In contrast, hepatomegaly, enhanced BrdU labeling, and increased expression of genes associated with the S phase of the cell cycle [PCNA, cyclin-dependent kinases (CDKs), cyclins] are found in wild-type mice, but these changes do not occur in humanized PPARα-humanized mice fed Wy-14,643 . These results demonstrate that there are distinct differences in the biological effects mediated by mouse and human PPARα. Although both mouse and human PPARα mediate changes in lipid metabolizing enzymes that are essential for the lipid-lowering properties associated with fibrates and other PPARα agonists, the changes in cell proliferation events only occur with the mouse isoform of PPARα. Long-term studies also demonstrate diminished tumorigenicity in humanized PPARα mice fed Wy-14,643 (C. Cheung et al., unpublished data), which provides stronger support for the hypothesis that inherent differences exist between human and mouse PPARα. Determining the molecular mechanisms that underlie this observed difference will likely be paramount in our understanding of PPARα agonist-induced liver cancer and likely explain why humans appear to be refractory to this event.
Transcriptional events mediated by coactivators could also represent a level of regulation for modulating PPAR-dependent gene expression that has not been examined extensively. It is well accepted that nuclear receptors such as PPARs undergo conformational changes after ligand binding that lead to recruitment of coactivator proteins that have histone acetyltransferase activity, which in turn allows the receptor complex to activate transcription of target genes [18, 101]. There are large numbers of proteins that function as coactivators, and many appear to be recruited as complexes [18, 101]. This complicated level of regulation is only beginning to be explored, and at this time the influence of differential coactivator recruitment cannot be excluded in having a role in PPARα-dependent liver carcinogenesis. For example, PBP is required for ligand-induced changes in liver gene expression mediated by PPARα . Thus, it is possible that differences in expression pattern, or polymorphisms or mutations in PBP or other PPARα coactivators in liver, could lead to functional differences in target genes that are regulated by PPARα and influence liver carcinogenesis. There is also evidence that the conformational change induced by specific PPARα ligands could differ and lead to preferential recruitment of specific coactivators , which in turn could lead to differential regulation of target gene expression by PPARα. In addition, there are several other mechanisms, including phosphorylation of PPARα and/or coactivators, differences in DNA response elements, and trans-repressive effects, that could be influenced by coactivator regulation and subsequently cause differences in a particular species response to ligand activation of PPARα.
The presence of polymorphisms in the promoters of PPARα target genes has also been hypothesized to explain the observed specific differences in the response to PPARα agonists. Functional analysis of a PPRE (AGGTCAGCTGTCA) in the human ACO promoter showed that PPARα binds to the PPRE in the presence of RXRα and that treatment with 500 μM ciprofibrate results in a marked increase in a reported construct driven by a 7.8-kb fragment of the human ACO promoter [102, 103]. In contrast, others have shown, using reporter assays and site-directed mutagenesis, that the human ACO PPRE is nonresponsive to PPARα agonism [104, 105]. Thus, there is some uncertainty regarding the functional significance of the ACO PPRE found in human populations. Although the role of ACO in PPARα agonist-induced hepatocarcinogenesis is unclear but still associative, differences in responsiveness to PPARα suggest that polymorphisms in response elements of this and other target genes may be functionally significant and should be examined in greater detail. Of particular interest are the target genes that regulate progression into the S phase of the cell cycle that likely lead to increased cell proliferation and subsequent liver cancer. It is clear that PPARα is required to mediate changes in many cell cycle regulatory proteins in liver  and that the human receptor does not modulate similar changes in gene expression ; yet, whether these changes are mediated by direct interactions of PPARα with any of these target genes or whether these changes reflect secondary or tertiary changes in gene expression has not been delineated. It is possible that some of these changes could be due to polymorphisms in promoters and/or functional differences between human and rodent PPARα.
Many of the studies examining species differences in the response to PPARα agonism in liver have focused on conventionally used PPARα ligands that have low or moderate affinity for PPARα, such as Wy-14,643 or ciprofibrate. However, recent advances in nuclear receptor ligand discovery has led to the identification of highly selective agonists for PPARα with affinities for activation in the nanomolar range [15, 56] as compared to the micromolar range for the above-mentioned compounds. In addition, some of these agonists have greater affinity for human rather than mouse or rat PPARα . Although there are reports demonstrating a specific difference in the response to more potent PPARα agonists , there are essentially no reports that have examined the effects of more potent PPARα agonists in vivo, in particular in nonhuman primate models. Demonstrating the lack of increased cell proliferation in “nonresponsive” species in response to very potent PPARα ligands would provide stronger evidence supporting the hypothesis that humans are refractory to PPARα agonist-induced hepatocarcinogenesis.
Another criticism that has been raised regarding the human and correlative animal studies examining the hepatocarcinogenic effect of PPARα ligands is that the length of treatment is insufficient to draw strong conclusions. For example, assays performed in rodents typically run from 1 to 2 years (at least half of the life span of a rodent), and essentially 100% incidence of liver cancer is observed in these models. By contrast, the human studies described above that correlated fibrate treatment with cancer incidence were performed in less than 10 years. This raises the possibility that the relatively shorter experimental periods may have missed identification of a potential relationship between PPARα agonist treatment and liver cancer. However, other studies show considerably diminished responses in ligand activation between humans and rodents (e.g., no induction of ACO in human liver after 8 weeks of fibrate treatment  vs. 15-fold induction of hepatic ACO observed within 24 h of fibrate treatment in rats  or no significant peroxisome proliferation in human liver [61–64] vs marked peroxisome proliferation in rodent liver ). In addition, there is good evidence showing that nonhuman primates are refractory to increased markers of cell proliferation that are causally linked to PPARα agonist-induced liver cancer , whereas rodent models consistently exhibit increased hyperplasia. However, whether or not longer treatment periods would demonstrate increased cell proliferation in nonhuman primate or human models in vivo has not been critically evaluated. Given the potential role of increased oxidative stress on cell proliferative events, this hypothesis should be more carefully examined. In the absence of PPARα expression, hepatic BrdU labeling indices are not changed in response to Wy-14,643 treatment , suggesting that oxidative stress resulting from PPARα-independent events (e.g., activation of Kupffer cells ) does not significantly influence cell proliferation, but this has not been examined after long-term treatments.
In conclusion, the mode of action for PPARα agonist-induced liver cancer is relatively well established. There is good evidence showing a profound species difference in the response to PPARα agonism in liver, with rodent models consistently showing enhanced sensitivity as compared to nonhuman primate and human models, which typically show a diminished response. Collectively, there is good reason to suggest that humans are refractory to PPARα agonist-induce liver cancer, but there are clearly some data gaps that should be filled to specifically delineate the mechanisms underlying the species differences.
This study was supported in part by the National Institutes of Health grants CA89607 and CA97999 (J.M.P.).