PPARs are a unique set of fatty acid regulated transcription factors controlling both lipid metabolism and inflammation.

Cells are constantly exposed to a large variety of lipids. Traditionally, these molecules were thought to serve as simple energy storing molecules. More recently it has been realized that they can also initiate and regulate signaling events that will decisively influence development, cellular differentiation, metabolism and related functions through the regulation of gene expression. Multicellular organisms dedicate a large family of nuclear receptors to these tasks. These proteins combine the defining features of both transcription factors and receptor molecules, and therefore have the unique ability of being able to bind lipid signaling molecules and transduce the appropriate signals derived from lipid environment to the level of gene expression. Intriguingly, the members of a subfamily of the nuclear receptors, the peroxisome proliferator-activated receptors (PPARs) are able to sense and interpret fatty acid signals derived from dietary lipids, pathogenic lipoproteins or essential fatty acid metabolites. Not surprisingly, Peroxisome proliferator-activated receptors were found to be key regulators of lipid and carbohydrate metabolism. Unexpectedly, later studies revealed that Peroxisome proliferator-activated receptors are also able to modulate inflammatory responses. Here we summarize our understanding on how these transcription factors/receptors connect lipid metabolism to inflammation and some of the novel regulatory mechanisms by which they contribute to homeostasis and certain pathological conditions. This article is part of a Special Issue entitled: Translating nuclear receptors from health to disease.


PPARs are a unique set of fatty acid regulated transcription factors controlling both lipid metabolism and inflammation
Tamas Varga, Zsolt Czimmerer, Laszlo Nagy To cite this version:

Introduction
Early in the '90s it was observed that several distinct compounds with similar chemical properties were able to trigger both an increase in size and number of hepatic and renal peroxisomes in rodent cells. Treatment of rodents with these compounds also increased the rate of β-oxidation of fatty acids and caused hepatomegaly and carcinogenesis, which did not occur in humans [1]. These compounds included certain herbicides, phtalate plasticizers and, most importantly, the fibrate class of hypolipidemic drugs [2,3], were later termed as peroxisome proliferators. The search for the pharmacophores whose activation by peroxisome proliferators caused the above effects led to the identification of PPARα, a novel member of the nuclear receptor hormone superfamily [1]. Based on sequence homology, the genes for two other members of the PPAR subfamily, PPARγ and PPARβ/δ, were later cloned from the mouse genome [4][5][6]. All three subtypes of the PPAR subfamily were found to be highly expressed in tissues relevant to energy homeostasis, and since certain dietary fatty acids and their metabolic derivatives were found to activate PPARs, the idea was instantly formulated that PPARs were regulators of metabolism. The validity of this suggestion was thoroughly proven in human and mouse studies revealing that PPARs are indeed master regulators of metabolism [7].
Later, better understanding of the expression pattern, activity and biology of these transcription factors indicated that they also have diverse functions outside of the realm of metabolism. Not surprisingly, among these non-canonical PPAR functions the regulation of inflammation has received the most attention and has been the focus of A C C E P T E D M A N U S C R I P T ACCEPTED MANUSCRIPT 5 (such as cytokine release), the reaction of the surrounding tissue to inflammatory agents and to immune mediators as well as the resolution/repair phase are all targets of regulatory mechanisms. Interestingly, several regulatory mechanisms of the inflammatory reaction were demonstrated to be influenced by lipid molecules. Due to the fact that PPARs are specialized receptors to detect fatty acid derived signal molecules, they are key candidate for being the receptors that transduce a fraction of the lipid mediated inflammatory signaling events. Indeed, several instances were identified in which certain fatty acid derived molecules were shown to activate PPARs and modulate inflammation [9,10]. The best example of this might be the case of eicosanoids. It has been known for a long time, that eicosanoids, the products of the essential fatty acid metabolism, are potent inflammatory agents that act locally to modulate inflammation. The simplified picture suggests that eicosanoids are pro-inflammatory mediators that act locally to enhance vasodilatation and increased permeability of venules. However, a few examples of anti-inflammatory eicosanoids, such as lipoxins, have also been described (for a review see [11]). Additionally, it was found that a gradual shift in the eicosanoid profile of inflammatory reactions resulted in diminished production of the initially predominant pro-inflammatory leukotrienes and increased release of pro-resolution lipoxins in the course of the inflammatory reaction [12]. Importantly, certain eicosanoids can signal not only via their own cell surface receptors, but also via PPARs, suggesting that the regulatory effects of these eicosanoids are partly mediated by PPARs. Recently, emerging evidence has suggested that another class of fatty acid molecules may also have dual potential to activate PPARs and modulate inflammation. Dietary fatty acids have been regarded solely as energy source for a long time, but they are now also recognized A C C E P T E D M A N U S C R I P T ACCEPTED MANUSCRIPT 6 as regulators of inflammation (often via the modulation of eicosanoid synthesis). For instance, saturated fatty acids and different classes of polyunsaturated fatty acids (PUFAs) were demonstrated to modulate inflammation [13]. Again, it is highly probable that PPARs play major roles in transducing the signals derived from dietary intake of lipids to the level of immune regulation.
The role of PPARs in inflammation is especially relevant in the case of metabolic syndrome and atherosclerosis. These are diseases of lipid (and glucose) metabolism with an underlying inflammatory component. Metabolic syndrome is a cluster of symptoms (impaired glucose tolerance, high blood pressure, dyslipidemia and abdominal obesity) that are often associated and significantly increase cardiovascular risk. Currently two main competing theories are proposed that could explain the emergence of impaired glucose tolerance in these patients. The lipotoxicity theory [14] proposes that when the fat storing capacity of the adipose tissue is chronically overloaded, fatty molecules will be deposited in other tissues, including muscle and liver. This would cause impaired insulin signaling in these tissues and, as a result, would lead to impaired glucose tolerance. The alternative explanation suggests that chronic caloric overload initiates an inflammatory response that is originated in the adipose tissue [15,16]. The inflammatory reaction make adipocytes and immune cells residing in the adipose tissue produce cytokines and adipokines (such as tumor necrosis alpha (TNFα)) that could lead to impaired glucose tolerance in remote tissues of the body [17]. Regardless of the nature of the primary mechanism, chronic inflammation seems to be an important component of metabolic disease. Synthetic PPARγ agonists (the glucose sensitizing thiazolidinedione (TZD) drugs) are potent drugs that improve insulin sensitivity in metabolic syndrome

ACCEPTED MANUSCRIPT
7 [18]. These drugs act primarily by modulating lipid and glucose metabolism, but they also have well documented anti-inflammatory effects as well. It is possible that PPARγ ameliorates metabolic syndrome solely by improving metabolic activities of target tissues involved in carbohydrate and lipid metabolism, as well as fat storage. It is also possible, however, that the beneficial effects of PPARγ activation on insulin sensitivity are mediated, at least partly, by its anti-inflammatory activities.
Due to their role in the above medical conditions, the metabolic roles of PPARs have been the target of intensive research. As their functions in inflammation slowly emerged, an increasing number of studies were devoted to dissect the role of PPARs in different types of inflammation, as well ( fig. 1.).

General biology of nuclear receptors
As already noted, PPARs belong to the nuclear receptor hormone superfamily. These proteins are transcription factors that are not only able to bind to DNA and regulate gene expression but they also serve as intracellular receptors by binding lipid molecules. This superfamily emerged in the early metazoan evolution and underwent an intensive evolutionary divergence [19,20]. As a result, the human and the mouse genome contain genes for 48 or 49 different nuclear receptors, respectively. A series of gene duplication events during early vertebrate evolution produced (among other subfamilies) the three members of the PPAR subfamily. There are several ways to group and categorize the divergent members of the superfamily. Originally those nuclear receptors whose ligands were identified were called "classic" nuclear receptors. These are typically endocrine A C C E P T E D M A N U S C R I P T

ACCEPTED MANUSCRIPT
8 receptors that bind their ligands with high affinity (such as thyroid hormone receptors for the thyroid hormones or the estrogen receptors for estrogens). Other nuclear receptors, whose ligands remained unknown, were classified as orphan nuclear receptors. To compound the naming system, the cognate ligands of some orphan nuclear receptors (including PPARs) were originally unknown but were later identified. Upon the discovery of their cognate ligands this later group of orphan nuclear receptors became "adopted" and was often referred to as adopted nuclear receptors [21]. This classical grouping system of nuclear receptors has several weaknesses. Maybe the biggest source of inconsistency in this classification derives from the fact that those nuclear receptors are categorized as orphans whose ligand has not been found yet. Since novel ligands are found and characterized all the time, this classification is, by its very nature, always temporary. Reflecting on the weaknesses of the above classification principle based on the presence of endogenous ligands, a phylogenetic classification model was introduced [22] in which the main organizing principle was sequence similarity. In this new nomenclature nuclear receptors received novel acronyms that reflected their position in the phylogenetic tree. Accordingly, PPARα, PPARβ/δ and PPARγ were renamed as NR1C1, NR1C2 and NR1C3, respectively. Although it would be more accurate to refer to PPARs according to this new naming system in the literature (including this review), we will use their trivial names to avoid confusion. One can assume that nuclear receptors that are very close on the phylogenetic tree and are therefore highly related show similar ligand binding properties and fulfill similar cellular roles. As their novel nomenclature indicates, the three different subtypes of PPAR show relatively high similarity. It would be tempting to assume therefore that they have similar functions. However, this is

ACCEPTED MANUSCRIPT
10 PPARs were originally described as orphan nuclear receptors, but soon a plethora of potential endogenous ligands were described. There are two major class of assays used to identify ligands. One class is represented by in vitro transactivation assays. In these cellular assays nuclear receptors are expressed ectopically in cells and the candidate ligands are used to activate the receptors. A marker of the transactivation activity of the nuclear receptor is then measured. This marker could be e.g. a known target gene for the corresponding receptor, or a signal that derives from an expression construct where a PPAR target sequence linked to a core promoter and a reporter (e.g. luciferase) gene. PPARs is also reflected in the size of the ligand binding pocket of the PPAR proteins.
The ligand binding pocket of PPARs is characteristically larger than that of classic nuclear receptors. It is possible that the unusually large ligand binding pocket enables PPARs to bind such a variety of different ligands.
Molecules that were found to bind physically to PPARs include polyunsaturated fatty acids (PUFAs) such as certain ω3-polyunsaturated fatty acids (e.g. α-linolenic acid with C18:3, or docosahexaenoic acid with C22:6), and certain ω6-polyunsaturated fatty acids (such as linoleic acid with C18:2 and arachidonic acid with C20:4). Certain saturated fatty acids (such as myristic acid with C14:0 and stearic acid with C18:0) were also found A C C E P T E D M A N U S C R I P T PPARs also have synthetic ligands that can easily be used to interrogate the transcriptional activities of the PPARs in cells that express different subtypes of PPARs.
Good examples are the hypolipidemic drugs clofibrate and fenofibrate, and the potent synthetic ligand Wy-14643 for PPARα, the thiazolidinedione (TZD) group of antidiabetic drugs (including troglitazone, pioglitazone, ciglitazone and rosiglitazone (formerly known as BRL 49653)) for PPARγ and GW-501516 for PPARβ/δ. PPARγ also has a specific synthetic antagonist, called GW-9662. Certain nonsteroidal anti-inflammatory drugs (NAIDS) were also shown to activate PPARγ [39]. These drugs (including indomethacin, ibuprofen and fenoprofen) are cyclooxygenase inhibitors, but they have certain effects that could not be ascribed to inhibition of cyclooxygenases.

DNA recognition by PPARs
A C C E P T E D M A N U S C R I P T

ACCEPTED MANUSCRIPT
14 PPARs form heterodimers with their obligate partners, the members of another subfamily of nuclear receptors, RXRs (of which three subtypes exist, RXRα, RXRβ and RXRγ).
According to the simplified model, upon ligand binding PPAR/RXR heterodimers recognize and bind to specific DNA sequences, called PPAR response elements (PPRE) [40]. This PPRE is a direct repeat of six nucleotide long core recognition motives (AGGTCA) that are separated by a single nucleotide. Because of the orientation and distance of the two hexameric motives, the PPRE is also called DR1. There are some additional features of PPAR target sites, such as the presence of an extended 5-half site, the presence of adenine as the separator single nucleotide or a slightly imperfect hexameric motif [41]. It is important to note, however, that the exact features of the consensus PPREs were calculated from DNA sequences to which PPARs were shown to bind and on which they acted as transactivators. The consensus sequence was derived from the 5' region of known PPAR target gene transcriptional start sites that showed robust regulation by PPARs. Due to the fact that PPARs (and other nuclear receptors) can bind to DNA in both orientation and at an unpredictable distance to their target genes, it is possible that the fine features of the DR1 elements were calculated from sequences that were not perfectly representative to all PPREs. Novel experimental approaches, such as "ChIP-on-chip" or "ChIP-seq" that are based on determining the sequence of the DNA binding sites of chromatin immune-precipitated transcription factors hold the potential to identify and fully characterize PPREs at the whole genome level [42,43].
All three PPAR subtypes are believed to bind to canonical DR1 elements. Due to the fact that certain cell types express more than one PPAR subtype, the question arises what A C C E P T E D M A N U S C R I P T ACCEPTED MANUSCRIPT 15 determines PPAR binding to a certain DR1 element in these cells [44]. Results showed that the 5' flanking nucleotides of the core DR1 elements played an important role in determining the PPAR subtype specificity of PPREs. Still, these fine additional sequence features cannot make PPREs entirely subtype specific. Accordingly, there are only a very few, if any, PPAR responsive genes that can only be regulated by one subtype of PPARs.
A recent example of a subtype specific response element was shown to be present in the fatty acid binding protein 4 (FABP4/aP2) gene [45] that is under the exclusive control of PPARγ in macrophages but not in adipocytes.
A further twist in the recognition of DR1 elements by PPAR-RXR heterodimers is the fact that the heterodimeric partner, RXR, is also a nuclear receptor and has its own cognate ligand (e.g. 9-cis retinoic acid or various fatty acids

Experimental models for studying PPAR functions
One of the reason why PPARs have been studied very intensively is that synthetic ligands of PPARα and PPARγ have been commonly used to treat metabolic diseases, such as dyslipidemia and type 2 diabetes, respectively, that affect millions of patients worldwide.
As a result, the early studies focused on the metabolic actions of PPARs. Concomitant to A C C E P T E D M A N U S C R I P T PPAR functions can also be studied in animal models of disease. There are genetically modified mouse strains available for each PPAR. Whole body knockout of PPARα causes relatively minor phenotypes in unchallenged animals [47,48]. PPARβ/δ knockout animals display high embryonic mortality on inbred background [49]. PPARγ -/animals are not viable and die in utero due to the deleterious effects of PPARγ deficiency in placentation. As a corollary, deficiency in PPARγ (and partly in PPARβ/δ) can be studied in heterozygous animals or in mice in which the PPARγ gene is disrupted by a Cre/lox

ACCEPTED MANUSCRIPT
17 mediated deletion only in certain cell types [50,51].  How are the molecular mechanisms of PPAR activity translated into anti-inflammatory effects? It is possible that PPARs regulate the expression of genes (either directly or indirectly, through transrepression) that have direct inflammatory roles. Alternatively, it is also possible that PPARs modulate inflammatory processes indirectly by altering lipid metabolism. According to this scenario, PPARs would directly modify the intra-, and extracellular pool of lipid molecules available in the body, and this altered lipid environment would initiate secondary regulatory processes. Such a mechanism is described in human dendritic cells (DCs) [54] where activation of PPARγ leads to the generation of retinoic acid, a molecule that regulates DC phenotype. It must be noted that the three PPAR subtypes alter intra-, and extracellular lipid homeostasis in distinct ways.

Mechanism of the inflammatory actions of PPARs
PPARα is activated and provides energy from fatty acid catabolism during starvation and cold acclimatization, PPARγ is activated in the well fed state and regulates the synthesis of fatty acids and related lipids, while PPARβ/δ ensures, among other, that fatty acids can provide energy for working muscles. How the three subtypes of PPAR that alter lipid metabolism in three distinctly different ways and generate distinct classes of lipid molecules can have similar anti-inflammatory roles in diseases suggest that the regulatory circuits of PPARs in metabolism and inflammation are, at least partly, uncoupled. It is also probable that the observed anti-inflammatory effects of PPARs are not exclusively mediated through their capacity to alter whole body lipid homeostasis and direct molecular regulatory mechanisms are, at least partly, accountable for their antiinflammatory effects. In the next sections we provide a short summary of the biology of the different subtypes and present a subjective account of the inflammatory models in which PPARs were implicated (table 2.) We limited our discussion to those models where molecular details of the regulatory mechanisms are accumulating.

PPARγ
There are two distinct isoforms of PPARγ, termed PPARγ1 and PPARγ2, which are transcribed from the same gene. PPARγ2 differs by an extra N terminal motif (28 amino acids in human or 30 in mouse) [55]. PPARγ2, which has a stronger transcriptional activity, is expressed at a high level almost exclusively in the adipose tissue, while of PPARγ that were approved as drugs (for the treatment of type 2 diabetes) raised the appealing prospect that the same drugs could easily be used to modulate inflammation as well.

Mechanisms of PPARγ mediated gene regulation
The canonical PPAR activity is the ligand dependent transactivation, in which liganded PPARγ forms heterodimers with RXR, and the PPARγ/RXR heterodimers recruit a large protein complex of co-activators required for the regulation of transcription. With the help of these co-activators, PPARγ/RXR heterodimers bound to PPREs in the enhancers of target genes will modulate the activity of the basal transcription machinery. This expression. The removal of the inhibitory complex is normally carried out by the ubiquitine proteasome system. In cells that receive concomitant PPARγ ligand and LPS tratements, a fraction of the liganded PPARγ will be SUMOylated on lysine K365. The SUMOylated PPARγ will not be able to bind its regular heterodimerization partner, RXR.
Instead, it will bind to the repressor complex located on the promoter of inflammmatory genes. The binding of PPARγ to these repressor complexes will block the ubiquitination and hence the efficient removal of the repressors. As a result, ligand bound PPARγ will maintain the repression on the promoter of inflammatory genes, such as iNOS2, even in the presence of active TLR4 signaling. Other forms of transrepression also exist. PPARγ was shown to bind directly other transcription factors, such as NF-κB or activator protein 1 (AP-1) [68,69], interfering with the DNA binding capacity of these transactivators.
Ligand activated PPARγ was also demonstrated to modulate p38 mitogen activated protein (MAP) kinase activity [70]. Although the above mechanisms can explain a subset of the anti-inflammatory effects of PPARγ, it cannot be excluded that positive transcriptional regulation of inhibitory proteins, rather than trans-repression of other transcription factors, play important roles in ligand-induced repression. Furthermore, the in vivo relevance and the contribution of these proposed mechanisms to the inhibition of gene expression remains to be further established.

Alternative activation of PPARγ
The consequence of ligand mediated activation of PPARγ can be modulated by phosphorylation. The first study described the phosphorylation of PPARγ2 at serine-112 by MAP kinases. This phosphorylation, which reduced the transcriptional activity of PPARγ [71], was observed upon exposure of cells to serum. Controversially, it was also found that insulin potentiated the ligand dependent activation of PPARγ via the action of MAP kinases, as witnessed by the enhanced expression of the robust PPARγ target gene, aP2 (also known as FABP4) [72]. PPARγ1 could also be phosphorylated at serine-82 (which corresponds to serine-112 of PPARγ2) by epidermal growth factor (EGF) and platelet derived growth factor [73]. Again, this phosphorylation event attenuated the transcriptional activity of the ligand bound PPARγ. The explanation for the observed divergent effect of PPARγ phosphorylation at the relevant serine is lacking.
Recently a novel mechanism for the posttranslational modification of PPARγ was described [74]. In this study the cdk5 mediated phosphorylation at serine-273 of PPARγ2 led to the dysregulation of the expression of a number of metabolism related target genes in adipocytes. Treatment of adipocytes with the synthetic PPARγ agonist, rosiglitazone, hindered this phosphorylation and normalized gene expression. Importantly, the genes whose expression was perturbed upon serine-273 phosphorylation were known to be relevant in metabolic regulation but did not necessarily belong to the "canonical" genes that are robustly regulated by PPARγ agonists. This is a completely novel angle of In two other early studies human monocytes, the human myelomonocytic cell lines THP1 and HL60 and murine lymph node macrophages [31,58] were investigated. It was found that lipid components of oxidized low density lipoprotein (oxLDL) were able to activate PPARγ in HL60 cells differentiating along the macrophage lineage, while the parent LDL It was also shown that PPARγ was expressed in cytokine treated human monocyte derived DCs (moDCs) [86][87][88]. Our laboratory and others demonstrated that activation of PPARγ in monocyte derived DCs led to an altered immune phenotype characterized by

ACCEPTED MANUSCRIPT
29 increased phagocytic capacity, antigen processing and lipid antigen presenting capacity [89][90][91] and [54]. Although the full picture of how PPARγ activation led to these phenotypic changes in DCs is not fully understood, important regulary mechanisms were described. In this model system, monocytes were isolated from peripherial blood, and Retinoic acid could activate its own receptors, retinoic acid recpetors (RARs), that are

ACCEPTED MANUSCRIPT
30 also members of the nuclear receptor superfamily. In short, a concerted transition from an active PPARγ signaling to an active RAR signaling occurred [54]. PPARγ expression was also detected in B cells [98]. B cell response to a various stimuli, including LPS stimulation or antigen receptor crosslinking, was also found to be modulated by PPARγ [99]. Interestingly, PPARγ +/animals which showed no difference in the T cell compartment when compared to wild type mice exhibited enhanced B cell proliferative responses to stimulation. Contrary to earlier findings, B cells of PPARγ +/animals exhibited increased viability.
PPARγ expression was also described in isolated primary natural killer (NK) cells and in NK cell lines [100]. Treatment of NK cells with 15d-PGJ2 or ciglitazone was found to cause a general attenuation of NK cell functions. The IFN-γ production, CD69 expression and the cytolytic activity of NK cells were investigated in PPARγ ligand treated (15d-PGJ2 and ciglitazone) and untreated cells. The inhibition of IFN-γ production was mediated by a PPARγ, while the cytolytic activity of NK cells was inhibited by a PPARγ independent mechanism. The chemically induced inflammation model was used in mice heterozygous for PPARγ [101], in mice with colonic epithelium specific [102], macrophage specific [103] or T cell specific PPARγ deletion [104], or in wild type mice in which PPARγ ligand treatment was applied to modulate inflammation [105]. All above animal models showed that deficiency in PPARγ resulted in an increased susceptibility to disease. PPARγ deficient animals showed more severe disease symptoms even in the absence of an exogenously administered PPARγ ligand. This suggested that either unliganded PPARγ had an activity

ACCEPTED MANUSCRIPT
34 that protected mice from the disease, or alternatively, an endogenous ligand with a potent PPARγ agonist activity was produced in these animals. These results raised the possibility that PPARγ agonists could be used to ameliorate human IBD.
Experimental autoimmune encephalomyelitis is an animal model of brain inflammation, in which the role of PPARγ in the regulation of inflammation can be studied. EAE in rodents is accompanied by demyelination. This characteristic of EAE makes it be a useful model for human multiple sclerosis (MS) and acute disseminated encephalomyelitis (ADEM). The first experiments demonstrated that the endogenous ligand 15d-PGJ2 and the TZD drugs troglitazone and ciglitazone were able to ameliorate experimentally induced EAE [106,107]. The decrease in disease severity and duration was, at least partly, due to a decrease in IL-12 production and Th1 cell differentiation. In line with the above results, PPARγ heterozygous mice developed an exacerbated disease [108] in the same EAE model. A recent study suggested that PPARγ activation suppressed central nervous system inflammation in the EAE model via the reduction of Th17 T cell differentiation [109]. This study demonstrated both the beneficial effects of the PPARγ agonist treatment in wild type mice and the deleterious effects of the CD4 T cell specific deletion of PPARγ.
Experimental autoimmune myocarditis is another model in which the potential involvement of PPARγ was investigated. When autoimmune myocarditis was induced in Lewis rats by immunization with cardiac myosin, administration of synthetic PPARγ ligands ameliorated disease severity. It was suggested that the inhibition of the expansion of autoreactive T cells and a shift in the Th1/Th2 balance were responsible for the beneficial effects of the agonist treatment [110]. PPARγ in macrophages involved in lesion formation [113].
Unexpectedly, macrophage PPARγ was shown to be a regulator of insulin sensitivity in two metabolic studies [80,114]. It has been known for long from both murine studies and human medical practice that ligand activation of PPARγ improves insulin sensitivity.
Originally, it was assumed that the beneficial effects of ligand activation were mediated by PPARγ in one of the major target tissues of glucose and lipid metabolism, such as muscle, fat or liver. Interestingly, the loss of PPARγ in hematopoietic cells was found to lead to insulin resistance. This suggested that inflammation is a component in developing insulin resistance in the mouse and macrophage PPARγ deletion exacerbated this

A C C E P T E D M A N U S C R I P T
ACCEPTED MANUSCRIPT 36 inflammation. It must be added, however, that a similar study [115] found preserved glucose tolerance in high-fat-fed C57BL/6 mice transplanted with PPARγ -/-, PPARβ/δ -/or LXRα -/-(liver X receptor alpha, another nuclear receptor with a role in lipid metabolism) hematopoietic cells. It was also found that the main site of the insulin sensitizing activity of ligand activated PPARγ was, in fact, adipose tissue [116]. Because of these contradicting results further studies are needed to clarify the contribution of macrophage PPARγ to the development of insulin resistance.

PPARγ and inflammation in humans
Human PPARγ mutations and single nucleotide polymorphisms (SNPs) were associated with metabolic and inflammatory diseases. Barroso

PPARα
PPARα expression was predominantly found in the liver, but was also found to be expressed in cardiac myocytes, proximal tubular epithelial cells of kidney, skeletal muscle, large intestine epithelium, endothelial and smooth muscle cells as well as immune cells including macrophages, lymphocytes and granulocytes [5,[123][124][125][126]. It is a key regulator of peroxisomal and mitochondrial β-oxidation of fatty acids, ketone body synthesis and systemic lipid metabolism. Similarly to PPARγ, there is an accumulating body of data suggesting that PPARα is not exclusively a metabolic regulator, but also have potent anti-inflammatory activities.

Mechanisms of PPARα activation
It is important to note that PPARα is the only subtype of PPARs whose candidate endogenous ligands are indeed most likely bone fide ligands. As already mentioned, dietary fatty acids can bind to and activate PPARα. Consequently, there is an intriguing possibility that our diet directly influences our immune system by activating transrciption factors and therefore regulating gene expression. The classical model described in the PPARγ section for agonist activated PPAR transactivation is also applicable to PPARα.
Similarly to PPARγ, there is hardly any immunologically relevant gene whose positive regulation by PPARα can be explained by the classical, agonist mediated transactivation model. The only notable exception may be the case of IL-4 regulation in lymphocytes

ACCEPTED MANUSCRIPT
39 [127]. In this experiment, gemfibrozil treatment attenuated the symptoms of experimental autoimmune encephalomyelitis (EAE) in mice. It was observed that the beneficial effect of the gemfibrozil could be detected only in wild type but not in IL-4 knockout animals.
Because it was known that gemfibrozil treatment enhanced IL-4 production, the idea was release. Treatment of aortic explants with fenofibrate strongly decreased IL-6 production.
It was found that liganded PPARα directly bound key transcription factors that are known to regulate IL-6 expression, such as the NF-κB subunit p65, c-Jun and c-AMP response element binding protein-binding protein (CBP).
Interestingly, a new twist in the mode of transrepression was described by Bougarne et al [128]. Glucocorticoid receptor alpha (GRα) is yet another nuclear receptor that has a well documented anti-inflammatory activity. Its ligands, glucocorticoids, are important drugs in treating inflammatory conditions. Both GRα and PPARα can inhibit NF-κB mediated inflammatory gene expression by transrepressing NF-κB. It was found that cells that received simultaneous PPARα and GRα ligand treatments exhibited an increased, additive transrepression of NF-κB. At the same time, ligand activated PPARα could negatively interfere with the transactivation capacity of GRα on its target sequences.
These results raised the questions if PPARα, activated by endogenous ligands, is also able

A C C E P T E D M A N U S C R I P T
ACCEPTED MANUSCRIPT 40 to potentiate GRα in vivo, and if PPARα agonists could be used in combination therapies with glucocorticoids to alleviate inflammatory conditions.

Alternative activation of PPARα
Several posttranslational modifications were described for PPARα. Insulin mediates the phosphorylation of Ser12 and Ser21, which enhance the transactivation capacity of PPARα [129]. Other phosphorylation mechanisms, by the p38 MAPK or protein kinase C (PKC) pathways, were also described. Interestingly, inhibition of PKC had a dual effect on PPARα. PKC inhibition decreased transactivation capacity of PPARα, but enhanced its transrepression activity [130]. .

Cellular model systems to study the inflammatory functions of PPARα
The expression of PPARα has been reported in several immunologically relevant cell types. In human monocyte derived macrophages it was found that ligand activation of PPARα induced apoptosis [126]. This pro-apoptotic effect was even more pronounced if phosphorylation [131]. As mentioned above, ligand activation of PPARα in murine T lymphocytes revealed that IL-4 and IL-5 are possible PPARα target genes [127].
Langerhans cells were also shown to express PPARα and pharmacological activation of PPARα inhibited langerhans cell maturation.

PPARα in animal models of inflammatory diseases
The first report that suggested that PPARα could control inflammation [9]studied leukotriene B4 (LTB4) induced inflammation in wildtype and PPARα knockout mice.
LTB4 is a locally generated lipid inflammatory agent that initiates and coordinates inflammation by activating its cell surface receptor. Alternatively, it can also bind to and Interestingly, PPARα showed a dual role in the LPS induced endotoxic shock model [132]. Similarly to PPARγ, it is possible that PPARα activity also modulates obesity associated inflammation either through its metabolic activity or anti-inflammatory effects. Induction of obesity with high-fat diet in PPARα knockout or wildtype mice suggested that PPARα protected against obesity-induced chronic inflammation in the liver. Plasma markers of liver injury and inflammation, including serum amyloid A and alanine aminotransferase activity, were increased in high-fat diet fed PPARα -/-, but not in wild-type animals [149].

PPARβ/δ
Among the PPARs, PPARβ/δ has been the least studied subtype so far. However, important advances were made in recent years in understanding the metabolic and inflammatory functions of PPARβ/δ. PPARβ/δ is expressed almost ubiquitously [59], with the highest level of expression found in colon, small intestine, liver and keratinocytes. PPARβ/δ is a general regulator of fatty acid oxidation in many tissues.

The role of PPARβ/δ
The involvement of PPARβ/δ in the regulation of lipid metabolism has been well established based on knockout and overexpression studies in transgenic mice [49,150,151]. PPARβ/δ knockout mice are smaller, both pre-, and postnatally, than wildtype animals. Reduced offspring numbers were also found, due to a placental defect.
Overexpression of a constitutively active PPARβ/δ in white adipose tissue reduced adiposity, most probably to the enhanced level of fatty acid oxidation. The metabolic pathways that were under the regulation of PPARβ/δ included fatty acid metabolism, mitochondrial respiration and programming of the muscle fiber type. The analysis of the full body knockout of PPARβ/δ revealed that it also regulated the inflammatory reaction during skin wound healing [150] and [151]. Recently, several studies reported the role of PPARβ/δ in different types of inflammation.

Cellular model systems to study the inflammatory functions of PPARβ/δ
Several inflammatory cell types express PPARβ/δ. PPARβ/δ activity in macrophages was thoroughly studied due to the connection of PPARβ/δ to atherosclerosis. Several  [152,153]. The main form of PPARβ/δ activity in these cells is proposed to be a ligand-independent transrepression, in which unliganded PPARβ/δ binds and sequesters the repressor molecule B-cell lymphoma-6 (BCL-6). Upon PPARβ/δ ligand activation, however, PPARβ/δ releases BCL-6, which then can repress the expression of inflammatory genes. The involvement of PPARβ/δ in atherosclerosis was also investigated [154]. It was found that PPARβ/δ deficiency in hematopoietic cells protected against atherosclerosis.

PPARβ/δ in animal models of inflammatory diseases
Interestingly, it seems that the loss of PPARβ/δ in foam cells reduced atherosclerotic lesion areas not by the modulation of lipid metabolism but the regulation of the A C C E P T E D M A N U S C R I P T ACCEPTED MANUSCRIPT 49 inflammatory component of atherosclerosis. The expression of several inflammatory genes was decreased in PPARβ/δ deficient macrophages, including MMP9 (matrix metalloproteinase 9) and IL-1β. According to the proposed mechanism for the observed anti-inflammatory role of PPARβ/δ, a ligand independent transrepression of BCL-6 was proposed. Also a ligand independent transrepression mechanism was proposed for the protective effect of PPARβ/δ in DSS induced colitis. The level of IFN-γ, TNF-α and IL-6 was increased in PPARβ/δ deficient animals [155].
Interestingly, PPARβ/δ also influenced the development of alternatively activated macrophages (M2 macrophage). Arg1, an M2 marker gene that is under PPARγ regulation in murine myeloid cells [80], was also regulated by PPARβ/δ [156,157]. In fact, a complementer regulation by both PPARγ and PPARβ/δ was needed to ensure Arg1 expression. Moreover, the M1 macrophage mediated uptake of Leishmania major was impaired in cells that received PPARβ/δ ligand treatment. These results suggested again that PPARγ and PPARβ/δ, but not PPARα, are regulators of the M2 macrophage development.
Another animal model in which PPARβ/δ shows overlapping functions with other PPARs is EAE [143]. In the EAE model, PPARβ/δ -/animals displayed more inflammatory foci in the central nervous system. This was partly due to an expanded population of CD4+ cells that produced both IFN-γ and IL-17. PPARβ/δ ligand treatment of wildtype CD4+ cells in serum-free medium resulted in a decreased production of IFN-γ by wildtype but not PPARβ/δ -/cells.
Recently, another aspect of the anti-inflammatory role of PPARβ/δ was revealed [158].
PPARβ/δ was found to be necessary for the timely clearance of apoptotic cells. PPARβ/δ A C C E P T E D M A N U S C R I P T

ACCEPTED MANUSCRIPT
50 deficiency caused a delay in the uptake, while PPARβ/δ ligand treatment in wild-type mice caused an enhanced uptake of apoptotic cells. C1q, a component of the classical complement activation pathway was found to be a direct target gene of PPARβ/δ. As a result of the abnormal sensing of apoptotic cells in PPARβ/δ -/animals, these mice developed a lupus erythematosus-like autoimmune disease.

Summary and perspectives
PPARs