help button home button Endocrine Society Endocrine Reviews
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Copyright Permission
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Desvergne, B.
Right arrow Articles by Wahli, W.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Desvergne, B.
Right arrow Articles by Wahli, W.
Endocrine Reviews 20 (5): 649
Copyright © 1999 by The Endocrine Society

Peroxisome Proliferator-Activated Receptors: Nuclear Control of Metabolism1

Béatrice Desvergne and Walter Wahli

Institute of Animal Biology, University of Lausanne, CH-1015 Lausanne, Switzerland


    Abstract
 Top
 Abstract
 I. Introduction
 II. Molecular Aspects
 III. Physiological Aspects
 IV. Conclusions
 References
 

I. Introduction
II. Molecular Aspects
A. PPAR isotypes: identity, genomic organization and chromosomal localization
B. DNA binding properties
C. PPAR ligand-binding properties
D. Alternative pathways for PPAR activation
E. PPAR-mediated transactivation properties
III. Physiological Aspects
A. Differential expression of PPAR mRNAs
B. PPAR target genes and functions in fatty acid metabolism
C. PPARs and control of inflammatory responses
D. PPARs and atherosclerosis
E. PPARs and the development of the fetal epidermal permeability barrier
F. PPARs, carcinogenesis, and control of the cell cycle
IV. Conclusions


    I. Introduction
 Top
 Abstract
 I. Introduction
 II. Molecular Aspects
 III. Physiological Aspects
 IV. Conclusions
 References
 
THE REGULATION of lipid and carbohydrate metabolism is central to energy homeostasis in higher multicellular organisms. It involves control systems that are sensitive to stimuli such as the availability of food, physical activity, stress, light, and temperature. The coordination of the responses to signals triggered by these stimuli must occur on several levels to ensure a well adapted energy balance, ranging from hypothalamic functions in the brain to the direct control by lipids and carbohydrates of their own fate. Another important role for lipids is the ability of some of their metabolites, such as leukotrienes or prostaglandins, to be secreted and act as potent mediators in many biological processes that participate in the diverse responses to endogenous and exogenous challenges that the organism faces. In this article, we will concentrate on the role of lipids and their derivatives in the genetic control of their own systemic transport, cellular uptake, storage, mobilization, and use. Fine tuning of these metabolic processes is a hallmark of healthy organisms.

Lipid homeostasis depends on factors that are able to transduce metabolic parameters into regulatory events representing the fundamental components of the general control system. Such factors may modulate the catalytic activity of individual enzymes by allosteric interactions, as do citrate and palmitoyl-coenzyme A (CoA), which activate and inhibit the lipogenic enzyme acetyl-CoA carboxylase, respectively. Alternatively, these factors may participate directly in the transcriptional control of genes encoding proteins involved in key metabolic steps. Several transcription factors that sense lipid levels in animal cells have received much attention in recent years. The adipocyte determination and differentiation factor/sterol regulatory element-binding proteins (ADD/SREBPs)2 are intracellular membrane-bound transcription factors whose activity is regulated by the cellular sterol content. In situations of sterol depletion, the active portion of SREBPs is released by proteolytic cleavage, enters the cell nucleus, and stimulates transcription of genes participating in three pathways of lipid metabolism: cholesterol biosynthesis, uptake of circulating fatty acids and cholesterol, and fatty acid biosynthesis (1 ). Another class of transcription factors comprises the liver X receptors [LXRs, i.e., NR1H3, according to the unified nomenclature of nuclear hormone receptors (2 )], whose ligands are oxidized derivatives of cholesterol (oxysterols). Analysis of LXR{alpha}-deficient mice revealed an essential function of this receptor as a major sensor of dietary cholesterol in the liver and an indispensable regulator of cholesterol homeostasis (3 ). Finally, the peroxisome proliferator-activated receptors (PPARs; NR1C) on which this review is focused belong to the steroid/thyroid/retinoid receptor superfamily, like LXRs, and are nuclear lipid-activable receptors that control a variety of genes in several pathways of lipid metabolism, including fatty acid transport, uptake by the cells, intracellular binding and activation, as well as catabolism (ß-oxidation and {omega}-oxidation) or storage. In addition to being indeed activated by fatty acids, they respond to fibrate hypolipidemic drugs and to insulin sensitizers. Rapid progress has been made in the exploration of PPAR biology, which indicates new mechanisms for the regulation of lipid metabolism and functions. In this review, we will first describe molecular aspects concerning the genes that encode PPARs, their structure, and their mechanism of action. The second part concentrates on physiological aspects related to PPAR expression, target genes, and functional consequences of their activation, leading the discussion to the most recent developments in the understanding of their possible physiological roles.


    II. Molecular Aspects
 Top
 Abstract
 I. Introduction
 II. Molecular Aspects
 III. Physiological Aspects
 IV. Conclusions
 References
 
A. PPAR isotypes: identity, genomic organization, and chromosomal localization
Three related PPAR isotypes have been identified in vertebrates, including Xenopus, mouse, rat, hamster, and human (4 5 6 7 8 9 10 11 12 13 14 15 ). They were named PPAR{alpha} (NR1C1), PPARß (NR1C2), and PPAR{gamma} (NR1C3) when the group of three was originally found in Xenopus (15 ), shortly after the characterization of a first PPAR in the mouse (5 ). With respect to this isotype nomenclature established with the Xenopus PPARs, the mammalian PPAR{alpha} and PPAR{gamma} were easily identified, while the third isotype was less clearly homologous to PPARß and was alternatively called PPAR{delta}, FAAR, or NUCI. Some evidence such as the expression pattern and the ligand pharmacological profile argue for these Xenopus and mammalian isotypes as being homologs. The analysis of the chicken PPAR ß-like isotype also suggests that ß and {delta} are indeed homologous since the chicken sequence falls about half way, in terms of similarity, between that of Xenopus and mammals (K. Umesono, personal communication). Below, we refer to this third isotype as PPARß until additional data provide a final answer to this still open question of isotype identity.

Phylogenetic studies have shown that PPARs form a subfamily of the nuclear receptor superfamily, along with the receptors for thyroid hormone, retinoic acid (RA), vitamin D, ecdysone, and the orphan receptors Rev-ErbA{alpha} (=ear1; NR1D1) and E75 (NR1D3, from Drosophila), the two latter being the closest relatives of the PPARs (16 ). The ancestral genes in this subfamily appeared more than 500 million years ago (17 ), and a more recent second period saw the duplication of the ancestral thyroid hormone receptor (TR) gene into two genes, TR{alpha} (NR1A1) and TRß (NR1A2), and of the ancestral retinoic acid receptor (RAR) gene into three genes, RAR{alpha} (NR1B1), RARß (NR1B2), and RAR{gamma} (NR1B3). Similarly, the three PPAR loci, {alpha}, ß, and {gamma}, appeared during this second period (16 ). Although it is not known whether the duplication events that produced the isotypes occurred exactly at the same time for the three receptors, they were likely contemporaneous to the appearance of the early vertebrates (18 ), as is indeed suggested by the chromosomal location of the TR, PPAR, and RAR genes. Since homologous isotypes have been found both in Xenopus and mammals for each of the three groups of genes, PPAR, RAR, and TR, the Xenopus-mammalian lineage divergence event can be used as a starting time point to determine the speed of evolution up to the present time within each of the three groups. In light of the amino acid sequence differences between the Xenopus and mammalian homologs, it appears that the PPAR genes have evolved 2–3 times faster than the RAR and TR genes. The possible relationship between this relatively rapid evolution and some particularities of PPAR ligand-binding properties will be discussed later.

The chromosomal localization of the PPAR genes has been defined in human and mouse. The human (h) PPAR{alpha} was mapped on chromosome 22 slightly telomeric to a linkage group of six genes and genetic markers that are located in the general region 22q12-q13.1 (13 ). The hPPAR{gamma} gene is located on chromosome 3 at position 3p25, close to RARß and TRß, which are at positions 3p24 and 3p21, respectively (14 ). Furthermore, the gene is within 1.5 megabases (Mb) of D3S1263, which is a suitable polymorphic marker that could be used for linkage analysis to evaluate a potential contribution of PPAR{gamma} to lipid metabolism-related diseases (19 ). Finally, the hPPARß has been assigned to chromosome 6, at position 6p21.1-p21.2 (20 ). In the mouse, PPAR{gamma} is located on chromosome 6 at position E3-F1, while PPAR{alpha} and PPARß are found on chromosome 15 and 17, respectively (21 ).

The mouse and hPPAR genes characterized to date reveal a common organization of the translated region in six coding exons with the following distribution: one exon for the N-terminal A/B domain, two exons for the DNA-binding domain (DBD)–one for each of the two zinc fingers–, one exon for the hinge region, and two exons for the ligand-binding domain (LBD). The mouse PPAR{alpha} gene spans at least 30 kb and comprises a total of eight exons, with two exons corresponding to the 5'-untranslated region and the last exon of the LBD comprising the 3'-untranslated region (22 ). For the PPARß gene, only a partial organization in Xenopus, which corresponds to the six exons of the translated region, has been reported so far (23 ). The human and mouse PPAR{gamma} genes extend over more than 100 kb of genomic DNA and give rise to three mRNAs, PPAR{gamma}1, PPAR{gamma}2, and PPAR{gamma}3, that differ at their 5'-end as a consequence of alternate promoter usage and splicing (Fig. 1Go). PPAR{gamma}1 is encoded by eight exons, comprising two {gamma}1-specific exons for the 5'-untranslated region, A1 and A2, and the six coding exons that are common to all three mRNAs. PPAR{gamma}2 is encoded in seven exons, the first one, exon B, comprising the {gamma}2 5'-untranslated region and encoding additional N-terminal amino acids specific of {gamma}2. On genomic DNA, this {gamma}2-specific exon is located between the second mPPAR{gamma}1 exon (A2) and the first common exon (19 24 25 ). A third mRNA, PPAR{gamma}3, encodes the same protein as PPAR{gamma}1 but is controlled by an alternative promoter located in the region flanking exon A2 in 5' (Fig. 1Go) (26 ).



View larger version (20K):
[in this window]
[in a new window]
 
Figure 1. hPPAR{gamma} isoforms. The genomic organization of the hPPAR{gamma} gene is shown. Two PPAR{gamma} isoforms are produced by the differential use of three promoters and alternative splicing of the three 5'-exons A1, A2, and B1. Exons 1–6 are common to all three transcripts. A scheme of the three mRNAs obtained is drawn, and the size of the proteins obtained is indicated. Note that transcription from the promoters {gamma}1 and {gamma}3 results in the same protein of 477 amino acids. In the human gene, the splicing event between exon A2 and exon 1 generates an AUG translation initiation codon six nucleotides upstream of the one used in other species, which is located at the very beginning of exon 1. Therefore, the hPPAR{gamma}1 protein has two additional amino acids at its N terminus compared with the rodent PPAR{gamma}1 protein. The PPAR{gamma}2 protein of 505 amino acids is produced by transcription from the promoter {gamma}2. The number of additional N-terminal amino acids in PPAR{gamma}2 vs. PPAR{gamma}1 is 28 in humans and 30 in rodents.

 
B. DNA binding properties
The DBD is the most conserved domain between all nuclear receptors, and indeed is the hallmark of the superfamily. It is formed by two zinc finger-like motifs folded in a globular structure that can recognize a DNA target composed of 6 nucleotides. In most cases, nuclear hormone receptors bind as dimers to two copies of such a core motif, which constitute a functional hormone response element. The spacing of the two motifs and their relative orientation (i.e., direct repeat, palindrome, or inverted palindrome configurations) determine which receptors bind to a given hormone response element (27 ). Members of the TR/RAR subfamily to which PPARs belong recognize preferentially the core hexanucleotide motif AGGTCA and are also characterized by the ability of forming a heterodimer with the 9-cis-retinoic acid receptor, RXR (NR2B). In fact, PPARs bind neither as homodimer nor as monomer but strictly depend on RXR as DNA-binding partner. Herein, the PPAR/RXR heterodimer will from now on be designated as PPAR:RXR.

1. PPAR response elements (PPREs) and PPAR:RXR binding properties. PPRE was first characterized by using synthetic oligonucleotides and was defined as a direct repeat of two core recognition motifs AGGTCA spaced by one nucleotide, thus also called DR1 (28 ). The first natural PPRE, found in the promoter of the acyl-CoA oxidase gene (15 29 ), and all natural PPREs subsequently identified fulfill these DR1 criteria, which allow PPREs to be discriminated from other direct repeat response elements of the TR/RAR class of receptors, such as the one recognized by vitamin D receptor (VDR) (NR1I1) (DR3), TR (DR4), and RAR (DR2, DR5). However, the detailed analysis of the CYP4A6 and malic enzyme genes PPRE, together with a sequence comparison of 19 native PPREs and subsequent mutational analyses, defined additional PPRE determinants (30 31 32 33 ). The three following properties can be added to the initial PPRE definition as a DR1: an extended 5'-half-site, an imperfect core DR1, and an adenine as the spacing nucleotide between the two hexamers, giving the following consensus sequence PPRE: 5'-AACT AGGNCA A AGGTCA-3'. These particularities most likely add discriminating parameters that contribute to PPAR:RXR binding selectivity vs. homo- and heterodimers of other members of the superfamily, some of which also recognize a DR1 type element (see below).

The PPRE structure as an extended direct repeat motif imposes a polarity to the bound heterodimer. PPAR interacts with the upstream extended core hexamer of the DR1, whereas RXR occupies the downstream motif (31 34 ). This represents a reversed polarity as compared with VDR:RXR and TR:RXR bound to DR3 and DR4, respectively, where RXR occupies the upstream core hexamer of the direct repeat. This difference in binding polarity between PPAR:RXR and VDR:RXR or TR:RXR is, at least in part, determined by the 5'-extended half-site in the PPRE. Receptors binding as monomers, such as NGFI-B (NR4A1), ROR (NR1F), and RevErbA{alpha}, also require an AT-rich 5'-extended binding site (35 ). Interaction of these receptors with the 5'-flank is thought to involve the receptor region immediately C-terminal to the second zinc finger, called carboxy-terminal extension (CTE). While PPARs are unable to bind DNA as monomers, it has been demonstrated that the CTE region of PPARs in PPAR:RXR is indeed responsible for the recognition of the 5'-flank of the DR1 in PPREs (36 ). The inability of PPAR to bind as a monomer has been attributed to the N-terminal region since deletion of the A/B domain of PPAR{alpha} allowed the truncated receptor to bind to a PPRE as a monomer. Limitation of the DNA binding capacity of PPAR by its A/B domain might reflect evolutionary changes that allow PPAR to functionally diverge from its monomeric cousins (36 ). While the three-dimensional structure of the DBD and CTE region of PPAR has not yet been solved, some of its properties can be inferred from detailed biochemical studies and structural analyses of RAR:RXR and TR:RXR bound to direct repeat sequences (37 38 39 40 41 42 43 44 45 ). Structural and biochemical analyses of RAR:RXR bound to a DR1 element demonstrate that the crucial amino acids for heterodimerization within the 5'-positioned receptor (RAR) are located in the second zinc finger, outside the first knuckle called D box, while the 3'-positioned receptor (RXR) contributes to the dimerization interface via its CTE region. Exchanging the specific PPAR D box, which has only three amino acids instead of five in other members of the superfamily, with that of RXR did not alter PPAR:RXR binding to a PPRE (G. Krey and W. Wahli, unpublished observations), consistent with the exclusion of the D box of the 5'-positioned receptor from the dimerization interface. The recent crystal structure analyses of the Rev-Erb DBD dimer bound on a DR2 (Rev-DR2) further confirm the presence of bonds between the tip of the second zinc finger of the upstream receptor and the GRIP box (VRFGRIPK residues) contained in the CTE region of the downstream receptor (46 ). Interestingly, PPAR{alpha}, which possesses the same GRIP box as Rev-Erb, is also capable of binding as PPAR:RXR on a DR2 if the spacing sequence between the half-sites corresponds to that found in Rev-DR2 (AGGTCATCAGGTCA) in opposition to an alternative DR2 (AGGTCAGGAGGTCA) to which it does not bind. Furthermore, transcriptional activation by PPAR:RXR can be obtained through the Rev-DR2 that contains the conserved 5'-extended sequence which is recognized by dimers of Rev-ErbA and ROR{alpha} (47 ). Therefore, a possibility of cross-talk exists between PPAR:RXR and these receptors on 5'-extended DR2 elements (36 ). The polarity of PPAR:RXR on such elements and the functional consequences of the formation of this complex have not yet been evaluated.

The fact that some tissues express more than one PPAR isotype raises the question of PPAR isotype-specific PPRE recognition. Assessment of the relative DNA-binding capabilities of the three PPAR isotypes ({alpha}, ß, {gamma}) to 16 native PPREs led to the classification of PPREs into three functional groups: strong, intermediate, and weak elements, which correlates with the level of the PPRE conformity to the consensus element (32 ). Surprisingly, the number of identical nucleotides in the core DR1 region is rather homogeneous across the different elements, and it is mainly the number of identities of the 5'-flanking nucleotides, rather than that of the stricto-sensu core DR1, which determines the binding strength of a given PPRE. In all cases, PPAR{gamma} binds more strongly than do PPAR{alpha} and PPARß and is thus less dependent on a well conserved 5'-flanking extension. In contrast, conservation of the 5'-flank is particularly essential for PPAR{alpha} binding and therefore contributes to isotype specificity. The PPAR DNA-binding activity is also modulated by the isotype of the RXR heterodimeric partner. Binding of PPAR:RXR to strong elements is reinforced when RXR{gamma} is the partner, whereas heterodimerization with RXR{alpha} is more favorable for binding to weak elements. However, it remains to be seen how these in vitro observations translate into selective recognition of the PPREs within their natural genomic and chromosomal environment.

2. Hormonal cross-talk occurring at the level of DNA binding. Direct repeat elements with a 1-bp spacing are also recognized by RAR:RXR, as well as RXR, androgen receptor-related protein-1, hepatocyte nuclear factor 4 (HNF-4) (NR2A), and chicken ovalbumin upstream promoter-transcription factor (NR2F) homodimers (48 49 ). Accordingly, HNF-4 and chicken ovalbumin upstream promoter-transcription factor homodimers can displace PPAR:RXR from its binding site and thus compete with PPAR signaling (50 51 52 53 ). Evaluation of the biological significance of the competition between PPAR and other members of the superfamily for binding to PPRE requires the consideration of at least two parameters. First, the subtle sequence determinants that we described above are important for nuclear receptor discrimination. A recent study shows that single point mutations applied to the core DR1 motif differently affect the binding affinity of HNF-4, androgen receptor-related protein-1, RAR:RXR, and PPAR:RXR (54 ). Thus, the conjunction of a core recognition motif that deviates from the consensus with a specific 5'-flanking sequence, as seen in many natural PPREs, may result in preferential binding of PPAR:RXR (30 31 ). Second, the relative amount of each type and isotype of nuclear receptor within a cell is of great importance in such a cross-talk.

Functional PPREs are almost exclusively represented by DR1-like elements. In addition to binding to the Rev-DR2 dicussed above, another exception is the ability of PPAR:RXR to recognize an estrogen response element (ERE) (55 ). Although an ERE-containing reporter plasmid can be transactivated by PPAR:RXR, no natural ERE-containing gene has been identified that is coactivated by estrogen receptor (ER) (NR3A) and PPAR:RXR. On the contrary, competition for the ERE leads to a PPAR:RXR-dependent repression of the ER-mediated transactivation of the vitellogenin gene A2 promoter as seen in transfection experiments. Thus the possibility of a hormonal cross-talk through an ERE exists (55 56 ), and genes might be found to be coregulated by ER and PPAR:RXR, and consequently by estrogens, fatty acids, and 9-cis-RA, in a natural cell physiological context and in a promoter- and cell type-specific manner.

RXR is a common DNA binding partner to many nuclear receptors of the steroid/thyroid receptor superfamily, including PPAR. Consequently, competition between these receptors for their common partner can occur. Reciprocal negative interactions between the PPAR and TR signaling pathways, through a mechanism of RXR sequestration, was indeed demonstrated in transfection assays (57 58 ). In these experiments, the relative amount of PPARs and TRs, respectively, determined which receptor was dominant, i.e., which signaling pathway inhibited the other. In vivo, such competition is likely to occur only when the amounts of RXR are limiting. Whether the relative amounts of TR, PPARs, and RXR meet these conditions in any tissue in vivo is so far unknown. A similar competition has been proposed to occur among PPAR isotypes (59 ). As a consequence, if several PPAR isotypes are coexpressed in a single cell type and if RXR amounts are limiting, there is a possibility of differential activity of the expressed isotypes.

Interestingly, it was observed that expression of RXR abolished PPAR{alpha} stimulation of the PRL promoter in pituitary GH4C1 cells (60 ). Analysis of this phenomenon revealed that stimulation of the PRL promoter by PPAR{alpha} was mediated by protein-protein interaction rather than binding of PPAR:RXR to the promoter. The mechanism proposed is a ligand-dependent association of PPAR{alpha} with the transcription factor GHF-1, which stimulates transcription and implies that PPAR{alpha} would act similarly to a coactivator in this specific situation. Overexpression of RXR is thought to titrate out PPAR{alpha} and therefore suppress its association with GHF-1 and consequently its stimulatory effect.

C. PPAR ligand-binding properties
One of the reasons for the present infatuation for PPARs lies in their particular ligand binding properties, making them attractive therapeutic targets. As we describe below, PPAR moved from the status of orphan receptor to that of generous host, capable of specifically interacting with more than one ligand, including some important natural compounds such as fatty acids. This section will end with information gained from the x-ray crystal structure of the PPAR{gamma} and PPARß LBD, which provides a link between structural and functional viewpoints.

1. Tools for PPAR ligand identification. The first molecules able to activate PPAR were identified in cultured cells cotransfected with a GRE- or ERE-containing reporter gene together with an expression vector encoding the chimeric receptor GRDBD-PPARLBD or ERDBD-PPARLBD. Compounds that trigger a stimulation of the reporter gene expression when added to the culture medium have been categorized as PPAR activators; the first identified were the fibrate hypolipidemic agents known to induce peroxisome proliferation in rodents, followed soon after by fatty acids (5 10 15 61 ). However, since activation might result from indirect events such as production of a metabolite of the test compound, release of an endogenous ligand, or activation of a cell surface-initiated signaling pathway, these compounds had to be tested further for direct binding to the PPARs.

As could be anticipated from transactivation assays, classical competition assays using radioligands first identified some PPAR{alpha} and PPAR{gamma} ligands with a relatively broad structural diversity (62 63 64 65 66 67 ). Additional techniques have then been adapted or developed to allow the screening of a large number of compounds (for review and technical aspects, see Ref. 68 ). The Scintillation Proximity Assay (SPA) is an equilibrium method that uses scintillation to measure the interaction between a molecule prebound to a fluomicrosphere and a radioactive ligand (69 ). It has been recently adapted to the evaluation of PPAR ligands in competition assays (70 ). The Differential Protease Sensitivity assay (DPSA) relies on a ligand-dependent reduction of PPAR sensitivity to enzymatic proteolytic cleavage (71 ). The Ligand Induced Complex (LIC) assay detects ligand-dependent binding of limiting amounts of PPAR:RXR to a PPRE (72 ). Based on the hypothesis that ligand binding to PPAR would induce interaction of the receptor with transcriptional activators, we have developed a novel sensor assay, termed Coactivator-Dependent Receptor Ligand Assay (CARLA) in which we measure the ability of a compound to induce PPAR-SRC1 interaction (73 ). Because of its strong interaction with PPARs, p300/CBP (cAMP response element-binding protein) can also be used in CARLA (74 ). In addition to ligand identification, these assays revealed three peculiarities of PPAR ligand binding properties that have important consequences for PPAR biology. First, in contrast to TR, RAR, VDR, ER{alpha}, or GR (glucocorticoid receptor) (NR3C1), PPARs accommodate several types of ligand, and the above-mentioned in vitro assays have demonstrated that most of the known PPAR activators are bona fide ligands (75 ). Second, and as a corollary, most of the molecules that specifically bind to PPAR do so with a rather low affinity as compared with the affinity of classical hormones for their cognate receptor. Third, there is some overlap in ligand recognition by the different PPAR isotypes, some ligands binding to more than one isotype although with different affinities. Although the known natural PPAR ligands fit well in our present understanding of PPAR functions, the question remains open whether, in addition, highly selective natural ligands exist with much higher affinity for each of the PPAR isotypes. Below is a presentation of the main PPAR ligands, natural and synthetic, discovered so far.

2. PPAR{alpha}, PPARß, and PPAR{gamma} ligands. The identification of unsaturated fatty acids as PPAR ligands (Table 1AGoGo) provides firm evidence that at least part of the PPAR-dependent transcriptional activity of fatty acids results from a direct interaction of the nuclear receptor with these molecules. These fatty acids bind all three PPARs, with PPAR{alpha} exhibiting the highest affinity, at concentrations that are in agreement with their circulating blood levels. In contrast, the very long chain fatty acid, erucic acid (C22:1), which is a weak ligand, appears more selective for hPPARß than for hPPAR{alpha} and hPPAR{gamma}, as measured in transfection assays using chimeric GR-PPAR proteins (76 ). Compared with the unsaturated fatty acids, saturated fatty acids are poor PPAR ligands in general (66 72 73 ), whereas phytanic acid, a dietary branched-chain, isoprenoid-derived fatty acid, efficiently binds PPAR{alpha} (77 ). The discovery that some fatty acids can act as hormones that control the activity of transcription factors demonstrated for the first time that fatty acids are not merely passive energy-providing molecules but are also metabolic regulators. This finding opens novel perspectives for deeper understanding of energy metabolism and therapeutic interventions. Future investigations examining the differential tissue distribution of PPAR isotypes with respect to qualitative and quantitative fatty acid content of each tissue might be very informative for further understanding the specific roles of PPAR isotypes and their fatty acid-mediated activation.


View this table:
[in this window]
[in a new window]
 
Table 1. PPAR ligands

 

View this table:
[in this window]
[in a new window]
 
Table 1A. Continued

 
Eicosanoids are a class of fatty acids mainly derived from arachidonic acid, either via the lipoxygenase pathway leading to the formation of leukotrienes (LTs) and hydroxyeicosatetraenoic acids (HETEs) or via the cyclooxygenase pathway producing prostaglandins (PGs). Several of these eicosanoids are activators of the different PPARs, and some are indeed ligands (see Ref. 78 ). 15-Deoxy-{Delta}12,14-PGJ2, which is a PGD2 derivative, is a ligand for PPAR{gamma} (63 64 73 ) and 8(S)-HETE, a compound associated with phorbol ester-induced inflammation, is a ligand for PPAR{alpha}, whereas the 8(R)-isomer shows a much weaker binding (66 72 73 ). Leukotriene B4 (LTB4), a chemotactic inflammation mediator, binds Xenopus PPAR{alpha} in classical binding assays, in CARLAs, and in LIC assays (65 73 79 ). Because binding affinities of molecules such as 8(S)-HETE and LTB4 for PPAR{alpha} are in the submicromolar range, or micromolar range for fatty acids, one might dismiss at first glance these interactions as it would seem unlikely that tissue concentrations of these ligands can reach the levels required for receptor activation in vivo. However, the nuclear localization of 5-lipoxygenase in some cell types supports the idea that in the nucleus local concentrations of eicosanoids, such as LTB4, can reach high levels and that intranuclear action of endogenous leukotrienes is feasible (80 81 ). Furthermore, the methods for assessing ligand dissociation constant (Kd) also deserve critical attention. The physical separation of bound and unbound molecules, often used in these techniques, leads to an equilibrium disturbance and often underestimates the Kd. To circumvent this problem, fluorescence-based methods that have been used to measure binding of retinoids to proteins are currently also applied to measure PPAR-ligand binding affinities (68 ). In this assay, the measurement of binding affinity is performed by optical means, which do not require the physical separation of bound and unbound molecules. This appoach has identified cis-parinaric acid as PPAR{gamma} ligand (82 ), and the affinity of many PPAR{alpha} ligands, such as fatty acids and LTB4, is found to be in the nanomolar range in such analyses (83 ).

The CARLA as well as the DPSA and the LIC assays have proven effective tools for the identification of interesting synthetic compounds as ligands (Table 1GoB). These include some hypolipidemic agents such as fibrates, of which clofibrate and the potent Wy-14,643 compound preferentially bind PPAR{alpha}. Thiazolidinediones (TZDs), which includes troglitazone, pioglitazone, and BRL 49653, now called rosiglitazone, are a class of antidiabetic drugs that are structurally derived from clofibric acid but selectively bind PPAR{gamma} (62 84 ). This functional association of a key regulator of lipid metabolism and an antidiabetic drug has important implications with respect to the pharmacological use of these compounds, on the one hand, and for the link that it emphasizes between lipid and glucose metabolism on the other. This link is further reinforced by the discovery of a novel series of antihyperglycemic and antihyperlipidemic agents that are PPAR{gamma} agonists (85 ). Other synthetic compounds that bind to PPARs include the arachidonic acid analog ETYA and some agonists and antagonists of the leukotriene membrane receptors (75 ). Intriguingly, the screenings for natural and synthetic ligands were not very successful in identifying PPARß ligands. Bezafibrate is a Xenopus PPARß-specific ligand (66 73 ), but its activity is much weaker on the mammalian PPARß. In transfection assays, the rat PPARß can be activated by the nonmetabolizable, substituted fatty acid {alpha}-bromopalmitate. However, nonspecific toxic effects of {alpha}-bromopalmitate are found at doses close to those required for PPARß activation (S. Basu-Modak, P. Escher, B. Desvergne, and W. Wahli, unpublished data). A novel series of fibrate derivatives, non-TZD compounds, was recently described as human-specific PPARß agonist and will aid in the functional analyses of this elusive PPAR subtype (74 ).

The recent advent of the combinatorial chemistry technology opens new opportunities for the identification of PPAR ligands. Instead of relying on classical large and diverse compound libraries, Brown et al. (86 ) designed a solid phase synthesis of biased chemical libraries of fibrates (so-called focused library) based on the observation that fibrates have activity on the three PPARs. Screening of the library identified a pool of compounds with activity on each of the three PPAR isotypes, of which the compound GW 2433 exhibits a high, although not selective, affinity for xPPARß. Thus, this approach may offer possibilities to develop selective and potent ligands for the three PPAR isotypes and has already provided a source of information about the ligand preferences of the three PPAR isotypes.

3. Species specificity in ligand recognition. Species difference in ligand recognition, already mentioned above for bezafibrate and xPPARß vs. mammalian PPARß, was first investigated with the PPAR{alpha} isotypes from Xenopus, mouse, and human, which differentially respond to two PPAR{alpha} ligands, Wy-14,643 and ETYA (67 ). Two amino acid residues in helix 3 of the LBD are responsible for the preferential responsiveness of Xenopus and hPPAR{alpha} to ETYA and of mouse PPAR{alpha} to Wy-14,643 (67 ). This identification of structure-function relationships involved in PPAR ligand binding specificity is of interest for drug development and may now be extended to additional compounds. These species differences, which have not been described for other nuclear receptors, raise two issues. The first, of practical importance, is that toxicological tests of PPAR ligands in whole animals must take into account possible species differences. The second is that the species-related ligand binding specificity may be linked to the speed of evolution of the PPAR genes. This might reflect an adaptation of the PPAR signaling pathways to nutritional patterns that can differ from species to species.

4. x-Ray crystal structure of the PPAR LBD. Although the LBD is less well conserved than the DBD between nuclear receptors, structural analyses of this domain performed with liganded RAR and TR LBDs and unliganded RXR LBD have revealed a common structural tridimensional fold, which consists of an antiparallel {alpha}-helical sandwich of 12 helices (helix 1 to helix 12) organized in three layers with a central hydrophobic pocket. Upon ligand binding, the swinging of helix 12 or activation function helix (HAF) closes the ligand binding pocket like a lid, in a so-called "mouse trap model" (87 88 89 ).

The x-ray crystal structure of the human apo-PPAR{gamma} LBD and apo-PPARß reveals an overall fold very similar to that of the above mentioned LBDs from helix 3 to the C terminus (90 91 92 ). However, some distinct differences are apparent. The core AF-2 activation domain in the apo-PPARs is folded against the ligand binding pocket in a conformation similar to that observed in the holoforms of PPAR and other nuclear receptors. An additional helix, called helix 2', which is found between the first ß-strand and helix 3, together with a placement of helix 2 that differs from other nuclear receptor tertiary structures, provides an easy access to the hydrophobic pocket for ligands. The region between helix 2' and helix 3, corresponding to the {Omega} loop in RAR is extended and is the most thermally mobile loop and participates in the structural changes occurring upon ligand binding. The ligand binding cavity is buried in the bottom half of the LBD and is particularly large, {approx}1300 Å3, of which the ligand occupies only about 30–40%. It is thus larger and more accessible than in other known LBDs [compare with the cavity in TR {approx} 600Å3, most of this volume being occupied by the T3 molecule {approx} 530 Å3 (89 )]. The T-shaped cavity in PPAR{gamma} comprised one region–the horizontal bar of the T–of 20 Å in length which lies parallel to helix 3; a second cavity region of 16 Å in length is orthogonal to the first and extends to the C-terminal AF2. The main amino acids involved in bonds with rosiglitazone are depicted in Fig. 2Go. Two histidine residues, H323 and H449, participate in the fixation of the TZD head group and are proposed to permit similar links with {alpha}-substituted carboxylic acids (91 ). The Y-shaped cavity in PPARß comprises three arms of about 12 Å in length, the left arm being rather polar in character. Eicosapentaenoic acid occupies this pocket in two distinct conformations, with the acid group and eight first carbon units fitting in the left arm oriented toward the AF-2 helix, while the hydrophobic tail either bends upward or downward into the up or bottom arm of the pocket, respectively (90 ). The same network of hydrogen bonds as seen with roziglitazone and PPAR{gamma} forms between eicosapentaenoic acid and PPARß AF-2 (see Fig. 2Go). These characteristics also explain that PPARß ligands are preferentially unsaturated fatty acid, given the requirement of a flexible hydrocarbon tail, and have an optimal length, long enough for sufficient stabilizing hydrophobic interactions and short enough for being docked inside the cavity (90 ). In conclusion, these key interactions associated with the relatively free non-specific interactions that the hydrophobic part of ligands can develop within the large cavity explain the promiscuous behavior of PPAR with respect to ligand binding. One consequence might be that different functional activities of a ligand might reflect different binding conformation in the cavity (90 ). Together with the easy access provided by the extra helix 2' and the tertiary placement of helix 2, these characteristics define PPAR as a nuclear receptor that has evolved to bind to multiple natural ligands with relatively low affinity, as reported above (91 ).



View larger version (76K):
[in this window]
[in a new window]
 
Figure 2. General structure and LBD of PPARs. A, Scheme indicating the four-domain structure of PPARs, which is the same as for the other members of the nuclear receptor superfamily. Coordinates of the boundaries of each domain are given for the hPPARs. The number inside each domain corresponds to the percentage of amino acid sequence identity of human PPARß and PPAR{gamma} relative to PPAR{alpha}. The region whose amino acid sequence is shown in panel B is underlined, and the coordinates are according to the following Genbank accession numbers: PPAR{alpha}, L02932; PPARß, L05792; PPAR{gamma}, X90563. DNA indicates DBD; ligand indicates LBD; aa, amino acids. B, Sequence alignment of the human (h), mouse (m), and Xenopus (x) LBD of PPAR{gamma} (PPARg), PPARß (PPARb), and PPAR{alpha} (PPARa). The secondary structure adopted by hPPAR{gamma} is indicated above the sequence in boxes for {alpha}-helices (H) and arrows for ß-strands (s). Amino acids that are involved in determining the ligand entry site, ligand binding, dimerization, and interaction with SRC-1 are indicated by colored filled boxes, for hPPAR{gamma} as well as for the other PPARs when conserved. Residues involved in specific interactions with the carbonyl group, the benzene group, and the sulfur atom of rosiglitazone are indicated by a black dot, a triangle, and an asterisk, respectively. These data are derived from the x-ray crystal structure of the LBD of the hPPAR{gamma} isotype in a complex containing the ligand rosiglitazone (BRL 49653) and 88 amino acids of human SRC-1 (91 ). Results obtained from the crystal structure of the LBD of the hPPARß isotype containing the fatty acid eicosapentaenoic are consistent with those of PPAR{gamma} (88 ). Individual residues for ligand entry and ligand binding to PPARß are indicated by the corresponding colored empty boxes. The dimerization PPAR:RXR interface is postulated, taking into account the residues conserved in RXR that contribute to the interface of a PPAR:PPAR homodimer observed in crystal solution (91 ). The mutations previously characterized and cited in the text are indicated by vertical lines below the sequence of helix 3, helix 10, and HAF (helix containing the activation function 2 core).

 
Taking into account the position and nature of the key residues bridging hPPAR{gamma} and its ligand, it is of interest to relate the effects of already described substitutions/mutations to these structural features (see Fig. 2Go). In PPAR{gamma}, four amino acids, namely aspartic acid 243 (D243) at the N terminus of the first ß-sheet, arginine 288 (R288), glutamic acid 291 (E291), and glutamic acid 295 (E295) in helix 3, are determinants of the ligand entry site (91 ). E291 and E295 are conserved in all known PPARs, whereas D243 and R288 are only conserved in the {gamma}-isotype, suggesting that these latter positions might be involved in isotype- and species-specific ligand selectivity. It has been observed that an experimental mutation of mPPAR{alpha} in which E282 (marked with a vertical line in Fig. 2BGo), which corresponds to E291 in PPAR{gamma}, is replaced by a glycine results in a 4-fold loss of PPAR{alpha} transcriptional response to Wy-14,643 and ETYA (93 ). Xenopus, human, and mouse PPAR{alpha} respond differently to these two compounds. Two amino acids in helix 3 determine the preference for ETYA in the Xenopus and human receptor, isoleucine 272 (I272) and threonine 279 (T279), whereas these two positions (marked by vertical lines in Fig. 2BGo) are occupied by phenylalanine (F272) and methionine (M279) in the mouse receptor, which has a preference for Wy-14,643 (67 ). After substitution in the mouse receptor of F to I and M to T, the mouse receptor loses its preference for Wy-14,643 in favor of ETYA. Interestingly, M279 or T279 in PPAR{alpha} corresponds to R288 in PPAR{gamma}, which supports the idea of a role of this position for ligand selectivity.

Finally, the crystal structure also revealed coiled-coil interaction between helix 10 from two PPAR{gamma} molecules forming a homodimer, reinforced by salt bridges involving helices 9 and 10. While PPAR{gamma} homodimers do not seem to occur in vivo, these observations are similar to those described for RXR homodimers (87 ) and are likely to reflect the contacts involved in PPAR:RXR. Consistent with these observations, a deletion comprising helix 10 and HAF of PPAR{alpha} impairs heterodimerization with RXR{alpha} (71 ). Similarly, a leucine to arginine substitution at position 433 (marked by a vertical line in Fig. 2BGo) in helix 10 of PPAR{alpha} also abolishes heterodimerization with RXR (58 ).

D. Alternative pathways for PPAR activation
1. PPAR{alpha} and PPAR{gamma} are phosphoproteins. Several nuclear hormone receptors, including PPARs, are regulated by phosphorylation in addition to ligand-dependent activation. PPAR{alpha} was first shown to be a phosphoprotein in primary rat adipocytes in culture. Treatment of these cells with insulin increases PPAR{alpha} phosphorylation. In parallel, transfection studies in CV-1 cells and HepG2 cells revealed that insulin increases by nearly 2-fold the transcriptional activity of PPAR{alpha}, as well as that of PPAR{gamma} (94 ). This insulin effect occurs through the phosphorylation of two microtubule-associated protein (MAP) kinase sites, identified at positions 12 and 21 in the A/B domain of hPPAR{alpha} (95 ). Using cotransfections in a Chinese hamster ovary cell line that expresses the insulin receptor, Zhang et al. (96 ) demonstrated a synergistic effect between insulin treatment and PPAR{gamma} ligand-dependent activation on the expression of the target gene aP2. Because this effect was partially inhibited by the addition to the culture medium of a MAP kinase inhibitor, the insulin effect was correlated with PPAR{gamma} phosphorylation observed in vitro upon exposure to purified MAP kinase (96 ). Contrasting results were obtained when exploring the role of growth factors in modulating PPAR activity (97 98 ). The epidermal growth factor (EGF) and platelet-derived growth factor (PDGF) decrease the transcriptional activity of PPAR{gamma} while increasing PPAR phosphorylation through MAP kinase signaling. A unique MAP kinase target site, which can be used by both extracellular signal-regulated protein kinase (ERK) and c-Jun N-terminal kinase (JNK) (99 ), was mapped at serine 82 in the N-terminal domain of mPPAR{gamma}1, which corresponds to serine 112 of mPPAR{gamma}2. Substitution of this serine by alanine leads to a loss of PDGF-mediated repression of PPAR{gamma} activity (97 98 ). Comparable MAP kinase-dependent PPAR{gamma} phosphorylation and inhibition of activity were obtained in 3T3-L1 cells with PGF2{alpha}, an arachidonic acid derivative that acts through a membrane receptor and has a potent inhibitory effect on adipogenesis (100 ). At the molecular level, the mutant PPAR{gamma}2 Ser112Asp exhibits a decreased ligand binding affinity and coactivator recruitment. Limited protease digestion of this mutant also results in an altered digestion pattern as compared with that of the native form of PPAR{gamma}2, when performed in the absence of ligands. These differences suggest that the N terminus, more specifically the phosphorylation status of the serine 112, plays a role in the conformation of the unliganded receptor, thereby regulating the affinity of PPAR{gamma} for its ligands (101 ).

How the same changes in phosphorylation can lead to an activation or an inhibition of PPAR{gamma} signaling, depending upon the nature of the triggering signal, insulin or growth factor, respectively, is unclear but likely involves either specific pleiotropic actions or use of different kinase pathways by these hormones acting on metabolic processes. For example, the insulin-mediated up-regulation of phosphoenolpyruvate carboxykinase (PEPCK) gene expression is independent of the Ras/MAP kinase pathway but relies on phosphatidylinositol 3-kinase (102 ). Another example is the inhibition of PPAR{alpha} activity by GH through the Janus kinase 2-signal transducer and activator of transcription 5b (JAK2-STAT5b) pathway. This inhibition requires a nuclear and transcriptionnally active STAT5b molecule and likely occurs via an indirect mechanism (103 ).

Thus, the remaining open questions concerning PPAR phosphorylation emphasize the importance that posttranslational site-specific modifications may have in the cross-talk between cell membrane signaling and nuclear effectors.

2. Activation of PPAR:RXR by RXR agonists. Another alternative activation pathway of PPAR:RXR occurs through ligand binding to RXR. PPAR forms a permissive heterodimer with RXR, meaning that either partner can regulate the transcriptional activity of the DNA-bound complex by interacting with its cognate ligand, on its own or when both partners are liganded. Indeed, cotransfection studies have shown that both members of the PPAR:RXR complex can mediate a response in the presence of their respective ligand. Furthermore, cotreatment of the cells with both ligands results in an additive effect. Thus, the natural RXR ligand 9-cis-RA as well as synthetic RXR-selective compounds such as LG 1069 and LG 100268 can activate a PPRE-driven reporter gene in a PPAR:RXR-dependent manner (28 61 104 ). Other examples are the liver fatty acid binding protein (L-FABP) gene and the ApoAII gene, known PPAR target genes, which are also responsive to 9-cis-RA (105 106 ). In vivo, this ability of PPAR:RXR to transduce 9-cis-RA signal has been associated with the observation that RXR-selective agonists display antidiabetic activities comparable to those obtained with a TZD, which specifically binds PPAR{gamma} (107 ). This in vivo action of 9-cis-RA is not PPAR{gamma} specific since it also activates PPAR{alpha}-inducible genes (108 ). The physiological relevance in the whole organism of 9-cis-RA pathways has been recently underscored by the identification of two 9-cis-retinol dehydrogenases that might participate in the synthesis in vivo of 9-cis-RA (109 110 ). However, interpretation of the above-mentioned results, obtained after 9-cis-RA stimulation in relation to the potential clinical use of RXR agonists, calls attention to the three following points: 1) in addition to acting through PPAR{gamma}:RXR and PPAR{alpha}:RXR, RXR agonists in vivo might also recognize and act via NGFIB:RXR and LXR:RXR permissive heterodimers (111 112 113 ); PPARß:RXR-mediated effects can also not be excluded; 2) in the absence of ligands, RXR forms inactive tetramers and addition of RXR-specific ligands preferentially directs the formation of homodimers rather than heterodimers (114 ). Thus, the formation of heterodimers, and the subsequent signaling, also depends on the presence of ligands for the heterodimerization partner (115 ); 3) RXR homodimers, whose formation is favored by 9-cis-RA, bind and transactivate through response elements corresponding to DR1 (48 116 ). In the context of the PPAR:RXR signaling pathway, how much of the 9-cis-RA response is PPAR dependent and how much is relayed by the formation of RXR homodimers remain to be determined. Such a question is brought up, for example, by the apparent contradictory observations concerning apoCIII gene expression whose down-regulation by fibrates is mediated by PPAR{alpha} (117 ), whereas the same gene is up-regulated by RXR-specific agonists (118 ). While little RXR homodimers are seen binding on PPREs in vitro, the situation in vivo remains to be explored. The use of a more comprehensive panel of RXR agonists, partial agonists, and antagonists should be used for investigating the complex network built up by RXR within nuclear receptor signaling. This is becoming possible with the recent characterization of the LG 100754 compound, which is a specific RXR:RXR antagonist but RAR:RXR and PPAR:RXR agonist (119 120 ).

In contrast to its role in PPAR:RXR, RXR can only work in subordination when heterodimerized with RAR, i.e., the liganded RXR is transcriptionally active only when RAR is itself liganded (Refs. 121 122 123 and references therein). A similar situation may apply for heterodimers between RXR and either TR or VDR. This would prevent retinoids from activating VDR or TR pathways, in contrast to that mediated by PPAR. It is not yet known whether these different transactivation properties reflect the different polarity with which PPAR:RXR, on the one hand, and RAR:RXR, TR:RXR, and VDR:RXR, on the other, bind to their respective response elements.

E. PPAR-mediated transactivation properties
PPAR-mediated transactivation results from the combination of PPAR:RXR binding to a PPRE and ligand activation of this complex. The conformational change of PPAR triggered by ligand binding or by other activation processes, such as phosphorylation, is believed to generate a conformation with new protein-protein interacting surfaces, that will allow specific contacts with a coactivator(s). Subsequently, this complex transduces regulatory action to the basal transcriptional machinery (see Fig. 3Go).



View larger version (32K):
[in this window]
[in a new window]
 
Figure 3. Model for transcriptional activation by PPARs. 1, Scheme of a PPAR-responsive promoter presented in a linear form with a PPRE, two binding sites for transcription factors (white box), the TATA box, and the transcription start site. The same region organized in a repressive chromatin structure is also shown. A hypothetical PPAR:RXR/corepressor complex, which is not bound to DNA, is activated by a ligand that results in a dissociation of the corepressors from the ligand-activated PPAR:RXR complex. 2, The activated PPAR:RXR complex binds to the PPRE producing a change in chromatin structure indicated by histone H1 release. PPRE-bound PPAR:RXR targets a coactivator-acetyltransferase complex to the promoter. 3, The promoter chromatin at the transcription initiation site region is modified by the coactivator-acetyltransferase complex, which acetylates the histone tails (Ac) thereby producing a transcriptionally competent structure. Acetylation of histones is selectively enriched at the transcription initiation region, involving one to two nucleosomes. 4, Additional transcription factors (e.g., Sp1, NF1) and the basal transcriptional machinery, including the RNA Pol II initiation complex, are recruited to the accessible promoter and transcription is initiated.

 
1. Delineation of activation domains in PPARs. The ligand-dependent activation domain, called AF2, is found in the LBD and is only transcriptionally active in response to ligand binding. Sequence alignment and mutation analyses have helped to locate a potent core activation domain at the very C terminus of the xPPAR{alpha} LBD. This region has some amphipathic helix characteristics and consists of two overlapping motifs, I and II, containing the core sequence {Phi}XE/D{Phi}{Phi} ({Phi} represents hydrophobic residues and X represents residues with long side chains). Motif II terminates at the last residue of the receptor C terminus while motif I terminates four residues before the C terminus. Both motifs are important for PPAR ligand-dependent transcriptional activity, and motif II might act indirectly by stabilizing an optimal conformation of motif I (G. Krey, A. Hihi, and W. Wahli, unpublished data).

While the ligand-mediated activity of PPAR is not affected by the deletion of its A/B domain, a ligand-independent activation function AF1 has been defined within this domain, for both PPAR{gamma} and PPAR{alpha}. The PPAR{gamma} AF1 activity was assayed in transfection analyses using chimeric transcription factors fusing the Gal4DBD with the PPAR{gamma}1 or PPAR{gamma}2 A/B domain. While both A/B domains are active, the mPPAR{gamma}2 N terminus, which possesses an extra 30 N-terminal amino acids, is 5- to 10-fold more potent in transcriptional activation than the mPPAR{gamma}1 N terminus (124 ). Interestingly, mutation of the MAP kinase site at serine 82/112 described above precludes phosphorylation of this site and increases the activity of Gal4-PPAR{gamma} AF1 (125 ), suggesting that phosphorylation at serine 82/112 not only affects ligand binding affinity as discussed above (101 ) but also directly regulates the AF1 activity. Similarly, hPPAR{gamma}2 with a proline-to-glutamine conversion at position 113 (given as position 115 in the original article), when overexpressed in murine fibroblasts, is defective in serine 112 phosphorylation and accelerates differentiation of the cells into adipocytes when compared with wild-type PPAR{gamma}2 (126 ). Insulin, in contrast, enhances the transcriptional activity of the PPAR{gamma} AF1 (124 ), as well as induces the activity of the PPAR{alpha} AF-1 domain (95 ), suggesting that insulin also induces phosphorylation of PPAR via an alternative site or a mechanism that may involve posttranslational modification of an auxiliary factor. No regulated activity of the N-terminal domain of PPARß has been reported so far.

2. PPAR interaction with cofactors. There is strong evidence for the crucial inhibitory or stimulatory role played by molecules that provide a bridge between DNA-bound transcription factors and the transcription initiation machinery (127 128 ). The nuclear receptor corepressor (N-CoR) and the silencing mediator for retinoid and thyroid hormone receptor (SMRT) are corepressor proteins that interact with unliganded nuclear receptors, mediating a repressive signal to the promoter to which the complex binds. Interaction with corepressors requires the CoR box, a structural motif in the N-terminal part of the LBD, which has been described in TR and RAR. PPAR{gamma} interacts strongly with N-CoR and SMRT in solution but not when bound to a PPRE as a PPAR:RXR complex (34 129 ). In association with the absence of a conserved CoR box in PPARs (34 ), these data provide an explanation for the absence of transcriptional repression by unliganded PPAR via its response element, in contrast to the repression activity of unliganded TR and RAR.

The first described nuclear receptor coactivator, steroid receptor coactivator 1 (SRC-1) (130 ), can interact with the PPAR LBD in solution. This interaction is ligand dependent, and we used this property to develop the CARLA screening assay (73 ). Two PPAR binding domains in SRC-1 have been identified (131 ). Furthermore, the liganded PPAR{gamma} LBD has been cocrystallized with the region of SRC-1 (aa 623–710) encompassing two LXXLL motifs (where X is any amino acid), characterized as consensus sequence found in nuclear receptor-associated factors (132 ). The x-ray crystal structure shows that in the presence of the ligand rosiglitazone, the two LXXLL motifs of a single SRC1623–710 molecule interact separately with the AF2 helix (HAF, see Fig. 2BGo) of each receptor molecule of a dimer, making a stable ternary complex: two PPAR{gamma} LBDs and one SRC1 molecule (91 ). In this structure, the LXXLL helix is oriented by a conserved glutamic acid of HAF (E471, see Fig. 2BGo, vertical line) and a conserved lysine in helix 3 (K301) of the LBD, allowing the placement of the LXXLL coactivator motif into the hydrophobic pocket formed by helices 3, 4, 5, and HAF of PPAR{gamma} (93 ). Importantly, distinct amino acids C-terminal to the core LXXLL motif are required for PPAR{gamma} activation in response to different ligands (133 ). In agreement with this structure, a site-directed AF2 mutant of PPAR (E471A) that has lost transcriptional capability also fails to interact with SRC-1 or CBP (134 ). Functional assessment of the importance of SRC-1 in PPAR-mediated transactivation comes from the overexpression of the nuclear receptor-interacting domain of SRC-1, which inhibits PPAR-dependent transactivation, whereas overexpression of the full-length SRC-1 potentiates ligand-dependent transcription by PPAR{gamma}:RXR (34 131 ). In addition, microinjection of SRC-1-directed antibodies inhibits the TZD-dependent activation of a PPRE reporter gene in Rat-1 cells, this inhibition being rescued by the coinjection of a plasmid expressing the full-length SRC-1 (135 ). In contrast, PPAR{alpha} is still transcriptionally active in SRC-1 knock-out (KO) mice, suggesting that its activity can also be mediated through interaction with other nuclear receptor coactivators (136 ).

Some studies have indicated that liganded PPAR might preferentially directly interact with CBP or the related protein p300 (137 138 139 ). This preference for CBP compared with SRC-1 has been confirmed using fluorescence energy transfer as an approach for quantitating such interactions (134 ). CBP/p300s are very large proteins that are essential for the transactivation function of many transcription factors including nuclear receptors (140 141 142 ), AP-1 (143 ), cAMP response element-binding protein (143 144 ), basic helix-loop-helix factors (145 ), STATs (146 147 148 ), and nuclear factor-{kappa}B (NF-{kappa}B) (149 ). Evidence from gene inactivation in mice and from the Rubinstein-Taybi syndrome in humans demonstrates that CBP/p300s are limiting factors (150 ). Thus, they are viewed as integrators of multiple signaling pathways, linking membrane receptor signaling and nuclear receptor activation pathways, as well as being a key limiting factor for which all the above mentioned pathways must compete. In parallel to these functional interferences between distinct pathways, the ability of CBP/p300 to simultaneously make contacts with more than one transcription factor might explain some synergy between these factors (127 151 152 ). Clarifying the specific contacts that PPAR may have with CBP/p300 is thus of great importance for understanding the molecular mechanism of the wide range of PPAR action. The first observations along this line showed that deletion of the 20 C-terminal amino acids of PPAR{alpha} abolished interaction with CBP, whereas the mutated receptor still binds Wy-14,643, suggesting that like SRC-1, CBP interacts with the activation function helix of PPAR (137 ). However, the LBD of the receptor alone (amino acids 281–468) is not sufficient for a stable ligand-dependent interaction with CBP, which also requires the participation of the T box in the D domain (amino acids 166–179). It is noteworthy that CBP can make functional interactions with SRC-1 through leucine-rich motifs in SRC-1 different from those required for interaction with PPAR. Therefore, it becomes apparent that these motifs serve several functions that are likely to control receptor- and ligand-specific coactivator recruitment as well as the assembly of extended complexes required for the transcriptional induction of receptor target genes.

Schulman et al. (139 ) explored a complex mammalian hybrid system involving fusion proteins to decipher the molecular mechanism of the PPAR:RXR activation by RXR agonists through interaction with CBP. This study shows that the activation of the heterodimer through RXR ligand is independent of the RXR AF2 activation domain but rather involves a conformational change of RXR, which is propagated to the unliganded PPAR moiety and leads to a PPAR ligand-independent interaction of the PPAR AF2 domain with the cofactor CBP. However, the system used does not take into account the role of DNA as triggering some specific constraints to the PPAR:RXR complex. Using the same approach, this "phantom ligand effect" was also previously described for RAR:RXR (153 ). In contrast to PPAR:RXR, RAR:RXR cannot be made transcriptionally active by a RXR ligand alone. Moreover, the RXR AF2 domain is crucial for retinoid signaling since a mouse that bears a deletion mutation of the RXR{alpha} AF2 domain dies around birth with most of the symptoms corresponding to vitamin A deficiency (154 ). Thus, while the findings described above are of great interest with regard to the PPAR:RXR mechanism of action, more experiments are required to integrate them into a physiological context.

Other PPAR interacting proteins, such as PPAR{gamma} coactivator 1 (PGC1), PGC2, PPAR-binding protein (PBP)/thyroid hormone receptor-associated protein 220 (TRAP220), receptor-interacting protein 140 (RIP140), and androgen receptor-associated protein 70, have recently been cloned (Table 2Go). PGC1 is a factor that can interact with PPAR (in a ligand-independent manner), TR and ER (in a ligand-independent manner, but with further reinforcement by thyroid hormone and estrogen, respectively), and RAR (in a ligand-dependent manner) (155 ). This factor is of particular interest because of the strong induction of its expression in muscle and brown fat upon cold exposure. Cotransfection assays indicate a major role of PGC1 in activating a brown fat-specific uncoupling protein 1 (UCP1)-reporter gene in the presence of PPAR{gamma}, RXR{alpha}, and a cocktail of ligands (troglitazone, 8-bromo-cAMP, and 9-cis-RA) (155 ). The ubiquitously expressed PPAR binding protein (PBP) (156 ), also cloned as the TR-associated protein TRAP220 (157 ) and VDR interacting protein DRIP230 (158 ), was identified in a double hybrid screen using the PPAR{gamma} LBD as bait. In a cotransfection assay, its overexpression slightly increases PPAR{gamma} activation while a truncated form has a potent dominant negative effect (156 ). This suggests that PPAR might also use the large multisubunit coactivator complex (DRIPs or TRAPs), which was shown to be anchored through the ligand-dependent interaction of DRIP230/TRAP220/PBP with VDR (158 ). Interaction of the PPAR LBD with RIP140, initially identified in a breast cancer cell line, is ligand independent (159 ). Interaction with the DNA-bound PPAR:RXR only occurs in the presence of 9-cis-RA and not in the presence of a PPAR ligand (160 ), suggesting that RIP140 interacts with the RXR moiety of the heterodimer rather than with PPAR. In transfection assays, the SRC1-mediated enhancement of PPAR{gamma} activity is down-regulated by increasing doses of RIP140 expressing plasmid, suggesting that RIP140 acts as competitor and inhibits SRC1 binding to PPAR:RXR (160 ).


View this table:
[in this window]
[in a new window]
 
Table 2. Cofactors interacting with PPARs

 
In addition to cofactors, PPARs have been shown to functionally interact with at least one other transcription factor. For regulation of the acyl-CoA oxidase promoter, PPAR:RXR exerts its effect through two PPREs in synergy with the transcription factor Sp1 via five Sp1-binding sites (161 ). These and the above observations underscore that PPAR action at any particular time in the cell will depend on the availability of several transcription factors and cofactors as well as on stimuli that can affect the levels or activity of these transcriptional components.

In response to these interactions, coactivators and corepressors alter target promoter activities by a mechanism that associates chromatin modification–via their intrinsic histone acetyltransferase or deacetylase activity, respectively, as demonstrated for CBP/p300 (162 163 164 )–and physical contact with the transcription initiation machinery (152 ) (Fig. 3Go). The fact that coactivators are shared by different nuclear receptors and other transcription factors may indicate that coactivators serve an important role in cell growth and differentiation as contributors to the selection of specific nuclear signaling pathways in a spatio-temporal manner. Assessing which and how a signal or biological situation would trigger preferential interaction of one or the other nuclear factor with PPAR is critical for understanding the integration of PPAR-mediated events in the complexity of the nuclear regulatory network.


    III. Physiological Aspects
 Top
 Abstract
 I. Introduction
 II. Molecular Aspects
 III. Physiological Aspects
 IV. Conclusions
 References
 
A. Differential expression of PPAR mRNAs
Information on PPAR expression patterns is a first step in understanding the biological significance of the existence of different PPAR isotypes and isoforms. PPAR isotypes are often coexpressed in tissues that are of ectodermal, mesodermal, or endodermal embryonic origin, with relative levels that vary from one cell type to the other. For the sake of clarity, the expression patterns will be presented separately for each of the three PPAR isotypes in Xenopus, rodents, and human.

1. PPAR{alpha} expression and regulation. In Xenopus, PPAR{alpha} is expressed at moderate levels during oogenesis. The maternal transcripts persist in the early embryo up to gastrula stages (15 ). They are then replaced by zygotic transcripts at tail bud stage. In the adult, PPAR{alpha} is expressed in all tissues that have been tested, i.e., liver, kidney, muscle, testes, and fat body. In mouse and rat, PPAR{alpha} appears relatively late in development (E13.5) in the tissues where it will be expressed in adulthood. In addition, there is a transient expression of PPAR{alpha} in the developping central nervous system and during skin maturation (8 165 166 ). In the adult rat, relatively high levels of PPAR{alpha} mRNA are detected in brown fat, liver, kidney, heart, and the mucosa of stomach and duodenum. Retina, adrenal gland, skeletal muscle, and pancreatic islets also express significant amounts of PPAR{alpha} mRNA (167 168 ). In the human, its levels in the liver appear lower than in the rodent liver (169 ). In addition, a splice variant lacking exon 6 is found, in addition to the full-length mRNA, in all human liver samples examined (169 ). Because of a shift in the open reading frame, the resulting protein is truncated shortly after the DBD, but no functional analyses of this short form is presently available. Other data available in the human indicate that PPAR{alpha} is well expressed in heart, kidney, skeletal muscle, and large intestine (170 171 ). In summary, and regardless of the species, the expression of PPAR{alpha} correlates with high mitochondrial and peroxisomal ß-oxidation activities, as exemplified by its high levels in cardiomyocytes and cells of the kidney proximal tubules, which primarily use fatty acids as an energy source. Another example is the enterocytes at the top of the intestinal villi, which carry the main burden of fatty acid absorption and have a very active peroxisomal ß-oxidation.

In rat liver, PPAR{alpha} expression is subject to negative and positive regulation by insulin and glucocorticoids, respectively (172 173 ). Accordingly, PPAR{alpha} mRNA and protein levels cycle in parallel with the circadian rhythm of circulating glucocorticoids. Stress situations or fasting, which induce the levels of plasma glucocorticoids, also result in increased synthesis of PPAR{alpha} in hepatocytes (174 175 ). In contrast, exposure of primary culture of rat hepatocytes to GH for several days decreases PPAR{alpha} mRNA levels by 50%. This suppression of PPAR expression may participate in the inhibition of peroxisome proliferator-induced peroxisomal ß-oxidation by GH (176 ). A down-regulation of PPAR{alpha} gene expression was also observed in chronic alcoholic liver disease in the rat (177 ). Finally, an up-regulation of PPAR{alpha} by its own ligands, fibrates, or FFA has been found in the FaO rat hepatoma cell line and in rat pancreatic islets, but whether the regulation is transcriptional or posttranscriptional remains to be clarified (178 179 ).

2. PPARß expression and regulation. In Xenopus, PPARß also accumulates early during oogenesis and is expressed in oocytes even at higher levels than PPAR{alpha} (15 180 ). These maternal transcripts slowly disappear in the early embryo up to gastrula stages to be replaced by zygotic transcripts at the neurula stage. In the adult Xenopus, PPARß expression is ubiquitous, with varying levels in different organs. During rat development, it is already present at relatively high levels in embryonic ectoderm and visceral and parietal endoderm at stage E8.5. The expression shows an important peak in the neural tube between E13.5 and E18.5, and then remains ubiquitous at a lower level throughout the end of development (165 166 ). At the adult stage, PPARß is also abundantly and ubiquitously expressed, often at higher levels than PPAR{alpha} and PPAR{gamma}. It also remains the most expressed isotype in the adult nervous system (165 166 181 ). It is only weakly expressed in liver, as compared with other tissues such as lung and kidney (4 8 167 ). Although it is abundant in skeletal and cardiac muscle, PPARß cannot be detected by in situ hybridization in smooth muscle cells of the digestive tract. In testis, its expression is very high in Sertoli cells (167 ). Interestingly, its expression is markedly induced in the uterus at the time of blastocyte implantation and remains abundantly expressed in the decidua at the postimplantation stage (182 ). In humans, PPARß is present at moderate levels in all tissues tested, with a higher expression in the placenta and the large intestine (170 171 ). So far, very little is known about the regulation of the PPARß gene. In the ob1771, 3T3-L1, and 3T3-F442A adipose cell lines as well as in the myoblast cell line C2C12, levels of PPARß transcripts appear to be low in proliferating cells and are induced upon differentiation (9 ).

3. PPAR{gamma} expression and regulation. In contrast to xPPAR{alpha} and xPPARß, xPPAR{gamma} mRNA is not detected during oogenesis in Xenopus except for a short transcript that does not encode a full-length receptor (15 180 ). At adult stage, xPPAR{gamma} has a relatively restricted expression, with the highest levels found in the fat body and moderate levels found in kidney and liver. Similarly, PPAR{gamma} has a restricted pattern of expression in adult rodents, white and brown adipose tissues being the major sites of expression (183 ). The intestinal mucosa also express high levels of PPAR{gamma} in the colon and cecum but less in the small intestine (184 185 186 ). Strikingly, PPAR{gamma} is abundant in lymphoid tissues such as the spleen (red and white pulp) and Peyer’s patches in the digestive tract (167 168 ). Finally, PPAR{gamma} is also expressed at low levels in the retina and skeletal muscle. In humans, both PPAR{gamma}1 and {gamma}2 are abundant in adipose tissue and are present at low levels in skeletal muscle. In addition, hPPAR{gamma}1 is also found in liver and heart (170 187 ). In contrast to rod