I. PPARγ in the gastrointestinal tract: gain or pain?

Johan Auwerx


The peroxisome proliferator-activated receptor gamma (PPARγ) has recently been implicated in the pathogenesis of inflammatory bowel disease (IBD) and colon cancer. The observation that PPARγ agonists, through immune modulation, protect against inflammatory processes in the intestine justified their expedient evaluation in the clinical management of IBD. PPARγ agonists are reported to have both tumor-promoting and -inhibiting effects in models of colon cancer. These differences can, in part, be explained by PPARγ-independent effects of PPARγ agonists and by differences in the models used. Because it is still unclear how PPARγ impacts on colon cancer, careful monitoring of patients receiving PPARγ agonists and additional basic research is indicated before recommendations on the use of PPARγ ligands in colon cancer can be made.

  • cancer
  • inflammatory bowel disease
  • hepatitis
  • nuclear receptors
  • thiazolidinediones

the peroxisome proliferator-activated receptor (PPAR)γ is a nuclear receptor that is activated by fatty acids and arachidonic acid metabolites. Thiazolidinediones (TZDs), certain nonsteroidal anti-inflammatory drugs, l-tyrosine-based compounds, and FMOC-l-leucine are the main classes of synthetic PPARγ ligands. PPARγ forms a heterodimer with the retinoic X receptor (RXR), and this heterodimer is also permissive for activation by RXR ligands. Ligand binding changes the conformation of the receptor, enabling the recruitment of coactivators, and transcriptional activation. PPARγ has classically been characterized for its implications in adipocyte differentiation and metabolism. In addition to adipose tissue, high levels of PPARγ are also found in the colon, whereas stomach, small intestine, liver, and pancreas express lower but still significant levels. PPARγ is mainly present in epithelial cells, stellate cells, monocytes/macrophages, Kupffer cells, dendritic cells, and B and T cells. In view of its expression and its involvement in immune response and cell proliferation, PPARγ has become a hot research topic in gastroenterology. This review provides a balanced opinion on the role of PPARγ in two important gastrointestinal disorders, i.e., inflammatory bowel disease (IBD) and colon cancer.


PPARγ not only controls the expression of genes involved in differentiation but also negatively regulates the cell cycle. PPARγ activation reduces S phase entry by inhibiting E2F/DP DNA binding, by inhibiting phosphorylation of the retinoblastoma protein, by inducing the cyclin-dependent kinase inibitors p18 and p21, and by decreasing cyclin D1 expression (reviewed in Ref. 4). TZDs induce the tumor suppressor gene PTEN, which also contributes to their antiproliferative activity. PPARγ activation inhibits the proliferation of malignant cells, including those derived from liposarcoma, breast adenocarcinoma, prostate carcinoma, colorectal carcinoma, nonsmall cell lung carcinoma, pancreatic carcinoma, bladder cancer, and gastric carcinoma (reviewed in Ref. 4). This growth inhibition is accompanied by changes in expression of genes linked to growth regulation and cell maturation. In addition, in adipocytes, macrophages, breast, prostate, and nonsmall cell lung cancer cells, TZDs induce apoptosis (reviewed in Ref. 4).

Firm genetic evidence supporting an association or linkage between the various polymorphisms/mutations in the PPARγ gene and the occurrence of cancer is at present lacking. Three studies tested whether somatic mutations in the PPARγ gene were more frequent in cancers. In a first study, four somatic mutations in the PPARγ gene were described in 55 sporadic colon cancers (26). Each of these mutations was reported to reduce PPARγ function (26). In another study, five of eight follicular thyroid carcinomas showed a fusion of the DNA binding domains of the thyroid transcription factor PAX8 to PPARγ (15). The PAX8-PPARγ fusion protein inhibited PPARγ in a dominant negative manner (15). The relevance of these last two studies is, however, still unclear, because another larger study failed to find any mutations in the PPARγ gene in 397 clinical cancer specimens of different origin (including colon, prostate, breast, lung cancers, and leukemias) (13).

In vivo evidence to support an antitumorigenic role of PPARγ is also conflicting. PPARγ activation inhibited tumor growth and progression in a xenograft model of prostate cancer (16), and attenuated breast cancer induced either by nitrosomethylurea (29) or by 7,12-dimethylbenz[a]anthracene (20). In vivo studies in colon cancer are particularly relevant because prostaglandin products, which are generated by cycloxygenase 2 (COX-2), are implicated in colon tumorigenesis and PPARγ might simply function as a downstream mediator. Troglitazone inhibited tumor growth in a xenograft model of colon cancer (25) and reduced the formation of aberrant crypt foci secondary to azoxymethane treatment (30). In sharp contrast, two other studies demonstrated that activation of PPARγ promotes the development of colon tumors in C57BL/6J-APCMin/+ mice (18,24), a clinically relevant model for both human familial adenomatous polyposis and sporadic colon cancer. A similar increase in the frequency of colon tumors was also reported in mice on a high-fat diet (32), suggesting that PPARγ could mediate the effects of high-fat diet on colon cancer.

Also, clinical studies have not provided a conclusive answer on the question whether PPARγ activity is favoring or inhibiting cancer formation and progression, because their outcome was largely deceiving and the clinical benefits were rather limited. The best results were obtained in three patients with liposarcoma, in which lineage-appropriate differentiation was induced (5). Troglitazone also stabilized prostate-specific antigen (PSA) levels in 41 patients with advanced prostate cancer (21). This effect on PSA levels was, in part, mediated by the inhibition of the androgen receptor (12), hence the question of whether such beneficial effects can be translated to other cancers that are independent of androgen receptor activity.

Interestingly, a recent study also showed the involvement of PPARβ in the development of colorectal cancer (11). PPARβ, like PPARγ, is expressed in the colon and can be activated by fatty acids. PPARβ was shown to be a target gene for the β-catenin/Tcf-4 transcription complex, which is formed when the adenomatous polyposis coli (APC) tumor suppressor protein is mutated. In view of the coexpression of PPARβ and COX-2 (10), it was proposed that PPARβ can mediate the protumorigenic effects of prostaglandins in the colon, whereas nonsteroidal anti-inflammatory drugs were suggested to inhibit tumorigenesis because they inhibit PPARβ activity (11). The lack of tumorigenicity of PPARβ −/− human colorectal cells in nude mice supported a protumorigenic role of PPARβ (23). At present, it is unclear how these observations on PPARβ articulate with the involvement of PPARγ in cell proliferation and colon cancer. Finally, it will be of interest to evaluate whether mutations or modulation in expression of cofactors for PPAR could influence PPAR-dependent tumor formation. A precedent for such a role of cofactors was highlighted in estrogen receptor-dependent breast cancers in which the coactivator, amplified in breast cancer-1, a member of the steroid receptor coactivator-1 family, or the nuclear corepressor N-CoR (see Ref. 17) were mutated or downregulated (1,17).

At present, not enough evidence is available to definitely establish whether PPARγ has pro- or antitumorigenic activities, and the field remains confusing. A similar reflection needs to be made concerning PPARβ. Some of this confusion is, however, secondary to the differences in experimental design. First, the differentiation state of the cells/tumors might affect the outcome. PPARγ activity is influenced by numerous other factors (cofactors, mutations in genes such as APC, differentiation status, etc.). For instance, PPARγ activation stimulated polyp formation (more well-differentiated tissue), but implanted tumors were inhibited (less well-differentiated tissue). A second difference, when considering animal models, stems from the different nature of the models. In fact, the C57BL/6J-APCMin/+ mice are more adequate to study the spontaneous development of colon cancer, whereas the xenograft model is better suited to analyze the behavior of a clonal cancerous cell population. A final point is that the concentration of PPARγ agonist used is important. In breast cancer cell lines, for instance, low concentrations of PPARγ agonists induced cell proliferation, whereas higher concentrations of the same agonists correlate with cell cycle arrest and apoptosis (3). In addition, high concentrations of PPARγ agonists elicit biological effects that are independent of PPARγ activation. With the use of PPARγ −/− mouse embryonic stem cells, it was recently shown that inhibition of cell proliferation by TZDs is independent of PPARγ and mainly caused by blocking G1/S transition through inhibiting translation initiation (inactivation of eIF2) (22). This questions whether the antitumorigenic properties of TZDs can really be attributed to activation of the receptor itself. A further point worth stressing is the fact that the majority of antitumorigenic effects is reported with troglitazone, a compound which has significant antioxidant properties.

From all of the above, it is clear that PPARγ's influence on cell cycle proliferation, differentiation, and apoptosis is complex. These effects depend on the concentrations of agonist, the cell type, and/or the mutational events that predispose to cancer development. In the absence of a full understanding of these mechanisms, careful monitoring of Type 2 diabetes patients chronically treated with PPAR agonists in postmarketing studies is indicated. Unfortunately, it will take several years and thousands of patients before an eventual beneficial or detrimental effect of TZDs on colon cancer formation will be unveiled in such a clinical setting. In addition, these conflicting data dictate the need of additional laboratory studies to address the role of PPARγ in tumorigenesis.


The molecular mechanisms mediating the anti-inflammatory action of the PPARγ/RXR heterodimer are, at present, not fully understood. Several studies have shown that activation of PPARγ may interfere with several signaling pathways regulating the expression of proinflammatory genes, such as those controlled by the stress kinases, nuclear factor-κB/Rel (NF-κB), signal transducers, and activators of transcription (STATs), activating protein 1, and the nuclear factor of activated T-cells (reviewed in Ref. 4). Consequent to the inhibition of these signaling pathways, PPARγ activators modulate the production of inflammatory cytokines, chemokines, and cell-adhesion molecules, thereby limiting the recruitment of inflammatory cells (reviewed in Ref. 4).

These anti-inflammatory effects, associated with the activation of the RXR/PPARγ heterodimer, incited several groups to explore their involvement in gastrointestinal inflammatory disorders. Treatment with TZDs has been shown to attenuate colitis induced either by oral administration of dextran sodium sulfate (28, 30) or by intrarectal administration of 2,4,6-trinitrobenzene sulfonic acid (TNBS) (6). This beneficial effect was directly attributed to the RXR/PPARγ heterodimer, because it was reproduced by activation of RXR with specific rexinoids (6). Further evidence in support of the implication of the RXR/PPARγ heterodimer came from the enhanced susceptibility to TNBS-induced colon inflammation of PPARγ +/− and RXRα +/− mice (6). The high expression of PPARγ in epithelial cells suggests that these cells constitute the main target of the RXR/PPARγ activators, a hypothesis reinforced by the persistence of inflammation in deeper layers of the colon in animals treated with PPARγ and/or RXR agonists (6).

Despite evidence for anti-inflammatory actions of the RXR/PPARγ heterodimer in the colon in animal models, the role of PPARγ in ulcerative colitis (UC) and Crohn's disease (CD), the two main forms of IBD in humans is little explored. In patients with UC, we recently observed an impaired expression of PPARγ at the mRNA and protein levels (P. Desreumaux and J. Auwerx, unpublished data). As peripheral mononuclear cells of UC patients expressed normal levels of PPARγ, it is likely that factors within the intestinal lumen may contribute to the observed decreased expression of PPARγ in epithelial cells. CD, on the other hand, is characterized by a localized hypertrophy of mesenteric adipose tissue, resulting in the so-called fat wrapping of the intestine (7). This hypertrophic mesenteric adipose tissue in CD is a rich source of tumor necrosis factor α (TNFα), which sustains local inflammatory responses (7). On the basis of the above information, several clinical trials have been initiated to evaluate the therapeutic efficacy of PPARγ agonists in IBD.

In addition, recent studies hint to eventual potential roles of PPARγ in other gastrointestinal disorders. In gastric epithelial cells, PPARγ activation was shown to be involved in the suppression of NF-κB-mediated apoptosis induced byHelicobacter pylori, suggesting eventual medical applications in gastroduodenal pathologies (9). Furthermore, PPARγ is expressed in hepatoma cell lines (14), Kupffer cells (31), and hepatic stellate cells (HSC) (8, 19), indicating a potential role in the liver. Most literature concerning PPARγ in the liver, however, relates to the hepatotoxicity of troglitazone, a problem that is much rarer with other TZDs (27). The two liver pathologies in which some preliminary evidence exists for the involvement of PPARγ are fibrosis and nonalcoholic steatohepatitis (NASH). In rat Kupffer cells, both PPARγ and RXR agonists inhibit lipopolysaccharide-induced nitric oxide and TNFα production (31). Likewise, activation of the RXR/PPARγ pathway has anti-inflammatory effects on HSC, in which they are reported to block proliferation (8), migration, and production of the chemokine MCP-1 (19). Because HSC are the main collagen-producing cells in the liver and because MCP-1 expression is directly related to recruitment of inflammatory cells to the liver, these data suggest that RXR/PPARγ may represent a therapeutic target for liver inflammation and fibrosis. PPARγ has also been suggested to be involved in the pathophysiology of fatty liver diseases, because it is highly expressed in the liver of genetically obese animals. NASH, a complex multifactorial disease often associated with obesity, Type 2 diabetes, and hypertriglyceridemia, recapitulates certain aspects of the liver pathology seen in these animal models. Although often silent, NASH progresses to cirrhosis in ± 5–15% of patients. Administration of troglitazone has been reported to induce normalization of liver enzymes with histological improvement in 7 of 10 patients enrolled in a small pilot study (2).

In conclusion, PPARγ evolved quickly from a nuclear receptor controlling metabolism to a pleiotropic regulatory factor, which affects numerous processes ranging from carcinogenesis to inflammation. At present, it is not yet clear whether PPARγ activation favors, or rather inhibits, colon tumorigenesis. Pharmacological studies using PPARγ ligands are compromised of PPARγ-independent effects of these compounds, confounding the interpretation of results. Human genetic studies and studies using gain and/or loss of function mouse models are not plagued by these confounding factors and will be required to unequivocally establish a role of PPARγ in colon neoplasia. Until these issues are adequately addressed, careful clinical monitoring of patients treated with TZDs will be required to detect potential effects on colon cancer incidence. Furthermore, it is premature to advocate the use of PPARγ agonist, antagonists, or modulators in neoplastic syndromes.

Current scientific evidence, derived from a combination of pharmacological and genetic studies, provides solid evidence for the involvement of the RXR/PPARγ in the control of gastrointestinal inflammation. This seems to justify the thorough clinical evaluation of RXR and PPARγ activators, either alone or in combination, as new anti-inflammatory drugs. Potential clinical applications could extend beyond the two common forms of IBD (UC and CD) to include inflammatory syndromes in the upper gastrointestinal tract and liver. In the future, it is expected that the better understanding of how RXR/PPARγ affects transcriptional control will guide the development of selective modulators of this heterodimer, which are specifically targeted to the digestive tract and would avoid side effects in other organs.


Work in the laboratory of the author is supported by grants of Centre National de la Recherche Scientifique, Institut Nationale de la Sauté et de la Recherche Medicale, Hopitaux Universitaires de Strasbourg, Association pour la Recherche center le Cancer, the European Community, the National Institutes of Health, the Human Frontier Science Program, and Association Régionale pour l'Euseiguement et la Recherche Scientifique.


  • Due to the limited number of references allowed for themes articles not all work related to PPARγ and the gastrointestional tract is cited. Interested readers can obtain a complete list from the author.

  • Address for reprint requests and other correspondence: J. Auwerx, Institut de Génétique et de Biologie Moléculaire et Cellulaire, 1 Rue Laurent Fries, 67404 Illkirch, France (E-mail:auwerx{at}igbmc.u-strasbg.fr).

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

  • 10.1152/ajpgi.00508.2001


View Abstract