PPARγ: a Nuclear Regulator of Metabolism, Differentiation, and Cell Growth

com-prise an important subfamily of the nuclear hormone receptor(NHR) superfamily. These ligand-activated transcription factorshave been intensively studied for more than a decade and havebeen implicated in such diverse pathways as lipid and glucosehomeostasis, control of cellular proliferation, and differentiation.The name PPAR derives from the initial cloning of one isoform asa target of various xenobiotic compounds that were observed toinduce proliferation of peroxisomes in the liver (1). This proteinwas called the peroxisome proliferator-activated receptor, nowknown as PPAR . Within a few years, the group of PPARs wasexpanded to include PPAR and PPAR (also referred to as PPAR ,NUC1, and FAAR) (2–6). This review will focus on PPAR .

which mediate contact between the PPAR-RXR heterodimer, chromatin, and the basal transcriptional machinery and which promote activation and repression of gene expression, respectively. Coactivator proteins, which include members of the p160/CBP/p300 and DRIP/TRAP families, are general coactivators for NHRs and indeed many non-NHR transcription factors. There are no known receptor-specific coactivators or corepressors, although selectivity for one or another NHR has been illustrated in certain cases (11,12). Coactivator proteins either possess or recruit histone acetyltransferase (HAT) activity to the transcription start site. Acetylation of histone proteins is believed to relieve the tightly packed structure of the chromatin, allowing the RNA polymerase II complex to bind and initiate transcription. Coactivators also recruit the chromatin remodeling SWI⅐SNF complex to target promoters (13,14).

What Are the Physiological Roles Played by PPAR␥?
PPAR␥ is the most intensively studied PPAR isoform. Studies have shown that this receptor participates in biological pathways of intense basic and clinical interest, such as differentiation, insulin sensitivity, type 2 diabetes, atherosclerosis, and cancer. PPAR␥ exists in two protein isoforms that are created by alternative promoter usage and alternative splicing at the 5Ј end of the gene; PPAR␥ 2 contains 30 additional amino acids at the N terminus compared with PPAR␥ 1 (6). Whereas many tissues express a low level of PPAR␥ 1 , PPAR␥ 2 is fat-selective and is expressed at very high levels in that tissue.
PPAR␥ Ligands-Because of its involvement in so many critical physiologic and pathologic functions (see below), great effort has been spent in trying to identify an endogenous, high affinity ligand for PPAR␥. A variety of fatty acids and their derivatives have been found to bind to PPAR␥ with relatively low affinity, but most investigators believe that their relevant concentrations in the nuclei of target cells are likely to be too low for them to be bona fide ligands. Certain eicosanoids have been shown to bind and activate PPAR␥ with higher affinity (15,16). 15-Deoxy-⌬12,14-prostaglandin J 2 , for example, binds to PPAR␥ with a k D in the low micromolar range and can activate PPAR␥ target genes at concentrations at or near the k D (17,18). 15-Deoxy-⌬12,14-prostaglandin J 2 , however, has never been definitively proven to exist in vivo, nor are its effects specific to PPAR␥. Many actions of this compound, which have been ascribed to PPAR␥ activation, have actually been shown to be mediated through inhibition of the NF-B pathway (19,20). Other eicosanoids, such as 13-HODE and 15-HETE, have been suggested to act as PPAR␥ ligands (21), a notion supported by the requirement for 12/15-lipoxygenase in some PPAR␥ responses in vitro (22).
Despite the paucity of information on true endogenous ligands, several high affinity synthetic PPAR␥ ligands have been generated. These include the thiazolidinedione (TZD) class of drugs, which are used clinically as insulin sensitizers in patients with type 2 diabetes (23) and were developed without knowledge of their molecular target. Other novel agents, including aryl-tyrosine derivatives, have been developed and are likely to show promise in both the laboratory and the clinic (24).
PPAR␥ and Adipogenesis-PPAR␥ was cloned as a transcription factor important in fat cell differentiation; it was also isolated in screens seeking new members of the PPAR family. In the former case, PPAR␥ was identified as a protein that bound to an enhancer in the 5Ј-flanking region of the aP2 gene, which encodes a fat cell-selective fatty acid-binding protein (6). This discovery was rapidly followed up by experiments showing that ectopic expression of PPAR␥ could dramatically promote adipogenesis in nonadipogenic, fibroblastic cells such as NIH-3T3 cells (25). When combined with an appropriate agonist and the pro-adipogenic protein C/EBP␣, even myoblasts could be "trans-differentiated" to adipocytes (26). PPAR␥ plays a crucial role in the function of many, and perhaps most, fat cell-specific genes. PPAR␥ binding is absolutely required for the function of the fat-selective enhancers for the aP2 and PEPCK genes in cultured fat cells (27). This analysis of the PEPCK gene has recently been extended in vivo, where activation of this promoter in fat was shown to be dependent on a PPAR␥ binding site, whereas expression in other tissues was not (28). The role of PPAR␥ in adipogenesis is also illustrated in studies that have deleted this gene in mice. The homozygous null mutation is lethal relatively early in gestation (embryonic days 10 -10.5) secondary to a defect in placental development (29,30), forcing investigators to use alternative means to investigate whether PPAR␥ is required for fat cell differentiation. Chimeric mice derived from both wild-type ES cells and cells with a homozygous deletion of PPAR␥ showed exclusion of null cells from white adipose tissue, but not several other tissues (31). Another group succeeded in bringing a single PPAR␥ Ϫ/Ϫ mouse to term by making tetraploid chimeric placentas; although the animal died shortly after birth it was found to lack brown adipose stores (30). In vitro, it has also been shown that PPAR␥ is required for the differentiation of adipose cells from ES cells and from embryonic fibroblasts (29,31). The results of these genetic studies have been complemented by experiments using pharmacological inhibitors and dominant negative alleles of PPAR␥ (32,33). These approaches have primarily been used to demonstrate a loss of PPAR␥ agonist-induced adipogenesis in vitro, although one study has shown a reduction in differentiation induced by the usual hormonal stimulants (34).
The CCAAT/enhancer binding proteins C/EBP␣, -␤, and -␦ have also been shown to be important in adipogenic differentiation. A transcriptional cascade exists in which C/EBP␤ and -␦ induce the formation of PPAR␥ and C/EBP␣ almost simultaneously (reviewed in Ref. 35). These latter two proteins then go on to promote the fully differentiated phenotype. In a manner analogous to the situation with PPAR␥, ectopic expression of C/EBP␣ in pre-adipocytes is able to drive adipogenesis to completion. Studies on fibroblasts engineered to lack C/EBP␣ show that they are deficient in PPAR␥ but can still become adipocytes (albeit without full insulin sensitivity) if PPAR␥ is added back (36). Conversely, ES cells or fibroblasts that lack PPAR␥ are deficient in C/EBP␣ (29,31). This raises the possibility that induction of PPAR␥ and C/EBP␣ represent redundant pathways for fat cell development. We have recently obtained data, however, that this is not the case, as fibroblasts that lack PPAR␥ are incompetent to undergo adipogenesis even when functional C/EBP␣ is added back at high levels. 2 The role of C/EBP␣ in adipogenesis, therefore, is ancillary to the role of PPAR␥ (see Fig. 2).
PPAR␥ and Type 2 Diabetes-A role for PPAR␥ in type 2 diabetes is clearly suggested by the efficacy of TZD ligands in ameliorating insulin resistance, an effect used by over a million patients currently taking these drugs (37). Several lines of evidence converge to prove that PPAR␥ is the relevant target of these drugs, including the finding that novel ligands designed to bind the PPAR␥ ligand binding domain with high affinity are very potent insulin sensitizers in vivo (24).
Additionally, mutations have been discovered in a few patients with severe insulin resistance (38). The protein product of these mutated alleles behaves in a dominant negative fashion in vitro, suggesting a role for PPAR␥ in the maintenance of basal insulin sensitivity. Interestingly, animals heterozygous for PPAR␥ exhibit increased insulin sensitivity relative to wild-type controls and also show resistance to diet-induced obesity (29,39). This may result from elevated serum leptin levels and decreased food intake in these mice (29). Regardless, there exists a discrepancy between the human and rodent situations that requires further explication. To make matters more confusing, a common polymorphism in the PPAR␥ gene (P12A) has been associated with protection from type 2 diabetes, despite the fact that this allele generates a weaker PPAR␥ in heterologous transcription assays (40,41).
Despite intensive investigation and years of clinical use of TZDs, much still remains unclear about the mechanisms by which PPAR␥ promotes insulin sensitivity. For example, the specific target tissue(s) of TZDs remain unknown. Adipose tissue is one likely target, and a recent study has shown that "fatless" mice expressing a dominant negative C/EBP allele do not show improvement in insulin sensitivity when treated with TZDs (42). An earlier paper (43) on a milder rodent model of lipodystrophic diabetes did not agree with this result, however. Other candidate sites for TZD action include skeletal muscle, liver, and pancreatic beta cells, and tissuespecific conditional knockouts of PPAR␥ are now being used to address these questions.
Uncertainty also surrounds the key transcriptional events by which PPAR␥ reduces insulin resistance (see Fig. 3). PPAR␥ activation in fat increases levels of Glut4, the insulin-stimulated glucose transporter (44), and may have other direct effects on important genes involved in glucose homeostasis. Unbiased target analysis of PPAR␥ in metabolically important tissues has revealed changes in gene expression that would have the net effect of translocating triglycerides and fatty acids from muscle and liver and promoting their storage in adipose tissue (45). This activity would theoretically improve glucose utilization in muscle and liver, although it must be remembered that similar effects could be equally explained as a consequence of improved insulin signaling in those tissues as well as a cause of insulin sensitization. Repression of genes involved in the promotion of insulin resistance could also explain the effects of TZDs and PPAR␥. In fact, TNF-␣ and IL-6 have been implicated in the development of the insulin resistance associated with obesity; PPAR␥ activation reduces levels of these cytokines in fat (46,47). Recently, a small secreted protein called resistin was discovered to be produced by fat cells and to promote systemic insulin resistance, and there is evidence that TZDs may repress expression of this factor as well (48), although recent data call this point into question (49). Finally, a recently discovered protein secreted by adipocytes, known alternatively as adiponectin, acrp30, adipoQ, and aPM1, has been found to be both a TZD target as well as a humoral mediator of insulin sensitivity (50,51).
PPAR␥ and Atherosclerosis-The discovery that PPAR␥ was expressed at relatively high levels in monocytes and macrophages led to studies showing that PPAR␥ agonists could promote macrophage differentiation and directly induce the scavenger receptor CD36 (52). These findings, coupled with the identification of  As preadipocytes begin to differentiate they express C/EBP␤ and C/EBP␦, which in turn activate both PPAR␥ and C/EBP␣. These two proteins potently induce each other's expression. PPAR␥ is required for differentiation, whereas C/EBP␣ plays a more ancillary role by promoting full insulin sensitivity and specific gene expression (see text).

Minireview: The Nuclear Receptor PPAR␥ 37732
PPAR␥ in "foam cell" macrophages within human atherosclerotic lesions (53,54), led to fears that TZDs could be promoting atherosclerosis in humans taking these drugs. Endogenous ligands of PPAR␥ were identified in atherogenic oxidized low density lipoprotein particles in serum, and it was shown that these particles could induce expression of PPAR␥ itself (21). A pathological cycle was proposed in which these particles induced their own uptake through activation of PPAR␥ and expression of CD36, leading to foam cell formation.
Other evidence, however, suggested that PPAR␥ might be beneficial in atherosclerosis (reviewed in Ref. 55). TZDs, for example, have been shown to reduce blood pressure in several mammalian models. Other atherogenic pathways are also inhibited by TZDs, including proliferation and migration of vascular smooth muscle cells and suppression of proinflammatory signals within macrophages in the vessel wall, such as IL-6, IL-1␤, TNF-␣, gelatinase, and scavenger receptor A (56,57). PPAR␥ also induces the expression of proteins involved in reverse cholesterol transport, presumably leading to a net reduction of cholesterol in atherosclerotic lesions. These transporters, ABCA1 and ABCG1, are actually induced by the orphan NHR LXR␣, which is itself a target of PPAR␥ (58 -60). Reassuringly, TZDs administered to LDL receptor knockout mice reduced atherosclerotic lesion number and size in males and had no adverse effect in females (61).
Interestingly, recent genetic studies show that PPAR␥ is not required for the formation of macrophages from monocytes, although macrophages lacking PPAR␥ have greatly reduced basal expression of CD36 (62,63).
PPAR␥ and Cancer-The activity of PPAR␥ in inhibiting the proliferation of fibroblasts during adipose differentiation first suggested that this receptor might be capable of reducing malignant behavior. This was examined in human liposarcoma, a malignancy of the adipose lineage. Most liposarcomas have been found to express much higher levels of PPAR␥ than other sarcomas, and cells grown from liposarcomas were found to have a dramatic differentiation response to PPAR␥ ligands, including lipid accumulation, cessation of growth, and expression of mRNAs characteristic of fat differentiation (64). A small clinical trial of TZD administration in liposarcoma showed that activation of PPAR␥ caused signs of adipose differentiation including changes in tissue morphology and gene expression, although the ultimate clinical outcome in these patients remains to be determined (65).
PPAR␥ is also expressed in a number of epithelial tissues that are important in human cancer, including breast, prostate, and colon. The colonic mucosa has been of special interest because PPAR␥ is expressed at very high levels here, comparable with the levels of expression in adipose tissue (66). Application of synthetic ligands brings about a marked reduction in cell growth in large numbers of human colon cancer cell lines, and PPAR␥ activation results in alterations in patterns of gene expression favoring a more mature, less malignant phenotype (67). Additionally, ligand administration to nude mice slows the growth of tumors derived from human colon cancer cells. Finally, mutations of PPAR␥ in tumor tissue have been detected in some patients with adenocarcinoma of the colon (68). All mutations were heterozygous, and all involved loss of function of PPAR␥, suggesting that PPAR␥ has tumor suppressor function in the human colon.
Paradoxically, administration of PPAR␥ ligands caused an increase in colon tumor number in Min mice, a mouse model of APC deficiency (69,70). No increases in polyp number were seen in wild-type mice, nor have there been reports of PPAR␥ ligands causing increased tumor formation in humans. Nevertheless, these observations are interesting and suggest that the role of PPAR␥ in the biology of the colon may be complex.
PPAR␥ in the prostate may also play an important role in tumor suppression. Up to 30% of patients with prostate cancer have heterozygous deletions of the 3p25 region containing PPAR␥, although these deletions are rather large and include many genes. In cultured prostate cell lines, TZDs have been shown to halt cell growth and to reduce secretion of the tumor marker PSA (prostatespecific antigen), and an encouraging response has been seen in some men with metastatic prostate cancer taking TZDs (71).
An interesting observation has also placed PPAR␥ in the spotlight in follicular thyroid carcinoma. In some cases of this disease, a fusion oncoprotein is formed by a chromosomal translocation between PAX8, deleted in its C-terminal activation domain, and full-length PPAR␥ 1 (72). The resulting fusion protein, the expression of which in the thyroid is presumably driven by the PAX8 promoter, has an extremely powerful dominant negative activity on the transcriptional activity of wild-type PPAR␥. The addition of ligand does not relieve this dominant negative activity. This translocation is not observed in benign follicular adenomas, suggesting that it is associated with carcinogenesis. Although the contribution of both the PAX8 and PPAR␥ components are likely to be important, the crucial role of PPAR␥ as a tumor suppressor moiety in this oncoprotein is shown by the fact that other cases of this disease have a fusion protein formed between PPAR␥ and as yet unidentified partners.

Conclusions
The last few years have seen an explosion of information about PPAR␥, implicating this NHR in biological processes as diverse as differentiation, regulation of metabolism, control of cellular proliferation, and maintenance of insulin sensitivity. The fact that PPAR␥ is a ligand-activated transcription factor has opened the door for pharmacological manipulation, allowing rapid application of basic discoveries to the clinical arena. One area of intense focus is the development of selective PPAR␥ activators, which could activate the receptor in some tissues but not in others. This will hopefully result in the development of drugs that provide the glucose-lowering benefit of TZDs, for example, without the doselimiting toxicity or the promotion of unwanted adipogenesis. Similarly, agents that exploit the growth-inhibiting effects of PPAR␥ in cancer cells without inducing metabolic sequelae would be useful. The amount and breadth of research effort devoted to these proteins ensures that more discoveries are certain to emerge. FIG. 3. PPAR␥ promotes insulin sensitivity. Noncompeting models of the mechanisms by which PPAR␥ activation by TZD drugs ameliorates insulin resistance are shown. In A, TZDs act on PPAR␥ in adipose tissue to increase the glucose transporter Glut4 and to decrease levels of cytokines that induce insulin resistance in liver and muscle. In B, TZDs act directly on multiple tissues to redistribute fatty acids away from muscle and liver and into fat, resulting in improved glucose utilization in the periphery.