Controversy: PPARγ as a target for treatment of colorectal cancer

Rajnish A. Gupta, Raymond N. Dubois


Colorectal cancer (CRC) represents a significant cause of morbidity and mortality worldwide. Recently, ligands for the nuclear hormone receptor peroxisome proliferator-activated receptor γ (PPARγ) have exhibited promise in the treatment of CRC. For example, activation of PPARγ reduces the proliferation of cultured CRC cells grown in vitro or in vivo using the nude mouse xenograft model of tumor growth. Furthermore, agonists of the receptor also reduce the development of preneoplastic lesions in a model of carcinogen-induced CRC in rats. However, ligands for the receptor paradoxically enhance intestinal adenoma formation in another murine model of intestinal polyposis, the APCMinmice. These disparate results may be due to the inherent limitations of the APCMin mouse as a model for humans with CRC. Finally, genetic studies identifying loss of function mutations of PPARγ in human CRC specimens strongly suggest a tumor suppressive role for the receptor during the development of CRC.

  • nuclear hormone receptor
  • intestinal epithelial cell differentiation

colorectal cancer (CRC) represents a significant cause of morbidity and mortality worldwide. Despite its prevalence, current therapies for the disease remain unsatisfactory and largely ineffectual, particularly in patients with advanced disease. Recently, ligands for the nuclear hormone receptor peroxisome proliferator-activated receptor γ (PPARγ) have proven effective in preclincal models of CRC. These findings have direct clinical relevance, because synthetic activators of PPARγ are currently being used for the treatment of non-insulin-dependent diabetes mellitus and do not appear to be associated with significant toxicity. However, the excitement of these findings is tempered by the fact that ligands for PPARγ are reported to enhance intestinal adenoma formation in a mouse model of intestinal polyposis. Thus recent studies on PPARγ in the colon have produced a controversial question that has yet to be completely answered: does activation of PPARγ prevent or enhance CRC growth?

PPARγ forms functional heterodimers with members of the retinoid X receptor family of nuclear receptors (for review, see Ref.27). Putative endogenous ligands for the receptor include both polyunsaturated fatty acids and the eicosanoids 15-deoxyΔ12,14-PGJ2 , 13-hydroxyoctadecadienoic acid, and 15-hydroxyeicosatetraenoic acid, but their respective roles in PPARγ signaling in vivo remain unclear. High-affinity synthetic ligands that selectively activate PPARγ include the thiazolidinedione (TZD) family of drugs. Early studies on PPARγ established its role as a dominant regulator of adipocyte differentiation (for review, see Ref. 15). Part of the cellular response of preadipogenic cell lines to PPARγ activators includes growth arrest characterized by cell cycle withdrawal (1). Not surprisingly, activation of PPARγ was also shown to inhibit the growth of liposarcoma cell lines in vitro (25), a result that has now been confirmed in clinical trials in patients with the disease (6). Because many cell types express at least low levels of PPARγ, the ability of the receptor to regulate differentiation and cell growth pathways in nonadipogenic cell lineages has been examined. Activating ligands of PPARγ have been found to inhibit the growth of virtually all cell types tested, including epithelial cells derived from the breast (12), prostate (13), stomach (23), and lung (4).

In the colon, levels of PPARγ mRNA are nearly equivalent to that found in adipocytes (7), with the highest levels of receptor expression observed in the postmitotic, differentiated epithelial cells facing the lumen (11). Consistent with this expression pattern, exposure of cultured human CRC cells to PPARγ agonists induces growth inhibition associated with a delay in the G1 phase of the cell cycle and an increase in several markers of cellular differentiation (3, 8, 17). However, whether these antineoplastic, prodifferentiation effects of PPARγ ligands in the colon operate in vivo is not clear. Three animal models that have been used to test this hypothesis, including adenomatous polyposis coliMin [(APC)Min] mice, azoxymethane (AOM)-treated rats, and nude mice with tumor cell xenografts. The antineoplastic effects of PPARγ have been confirmed in the latter two models. Rats treated with the chemical carcinogen AOM develop preneoplastic colonic lesions termed aberrant crypt foci (ACF), which later progress into carcinomas. Tanaka et al. (24) demonstrated that administration of the PPARγ ligand troglitazone significantly reduces the number of ACF lesions. Importantly, it will be necessary to carry out these studies further to determine whether these effects are extended to a reduction in the number of adenocarcinomas in the colon. Additionally, administration of troglitazone also significantly reduces the tumor volume of a CRC cell xenografted onto the flanks of athymic mice (17).

In contrast to the above findings, PPARγ ligands have been reported to enhance polyp number in the APCMin mouse model of CRC. These mice harbor a nonsense mutation in the tumor suppressor geneAPC (20); loss of function mutations in this gene are responsible for the hereditary polyposis syndrome familial adenomatous polyposis (FAP) and are also thought to be one of the major genetic initiating events for a large percentage of sporadic CRCs (14). Two different groups have reported that PPARγ ligands slightly increase colon polyps in APCMin mice (increasing on average from a mean of 1 colonic polyp in control mice to 3 colonic polyps in PPARγ ligand-treated mice) (10,16). No effects on either polyp number or size were seen in the small intestine. It has been proposed that the increase in colonic polyps seen with PPARγ ligand treatment is consistent with the increase in polyp burden observed when these mice are placed on a high-fat diet. Because fatty acids can potentially act as endogenous activators of PPARγ, the receptor may serve as a molecular link between dietary fat and colorectal carcinogenesis. However, there are important differences between these two types of treatment, because, unlike the case with PPARγ ligand treatment, APCMin mice placed on a high-fat diet have significant increases in polyp number in both the small and large intestines (26). It should be emphasized that PPARγ ligands did not induce polyp formation in wild-type mice, implying the potential need for a predisposed genetic susceptibility in order for PPARγ ligands to induce this effect. Additionally, extensive carcinogenic testing of several TZD derivatives that are potent PPARγ ligands have not demonstrated any protumorigenic effects of these compounds (T. M. Willson, personal communication). Finally, both of these studies relied exclusively on the TZD class of PPARγ agonists, and the results observed could be due to effects of TZD compounds unrelated to PPARγ agonism. Novel and potent non-TZD PPARγ agonists have been developed, and one of these, GW7845, was shown to be a potent suppressor of tumorigenesis in a rodent model of carcinogen-induced breast carcinoma (21). It will be important to determine the effects on polyposis in the APCMin mice of such non-TZD PPARγ ligands.

How can these results be reconciled with the antineoplastic effects of the same compounds observed both in cultured CRC cells and other preclinical models of tumor growth? And, just how relevant are these findings to human disease? One explanation may be that treatment in these particular mice with an agent that accelerates colon epithelial cell differentiation results in abnormal and excessive proliferation of the intestinal stem cell compartment; such a chronic stimulus may paradoxically result in the selection of stem cells with loss of normal growth control mechanisms. Additionally, in the APCMinmice, polyp formation is much more pronounced in the small intestine (ranging from 70–100 polyps/mouse) compared with the colon (∼0–2 polyps/mouse). Thus one problematic issue is that PPARγ agonists only affected polyp formation in the colon, where the polyp numbers are so small that limited statistical confidence can be achieved. Finally, the APCMin model is clearly imperfect, because in humans with FAP, the majority of intestinal polyps is found in the large intestine. Furthermore, the adenomatous polyps in these animals rarely progress into invasive carcinomas. Recently, Shibata et al. (19) created a novel mouse model of FAP using a conditional gene-targeting strategy that specifically deleted exon 14 in both alleles of APC in the colorectal epithelium, resulting in the rapid development of colonic, but not small intestinal, adenomas and adenocarcinomas. It will be of major interest to determine whether PPARγ activators also enhance intestinal adenoma formation in this variant model of the APCMin mouse.

It could be argued that the protumorigenic effects of PPARγ ligands are only seen in the background of genetic mutations in theAPC gene. Supporting this notion, CRC lesions in the AOM rat normally contain activating mutations in the β-catenin oncogene rather than loss-of-function mutations in APC (22). On the other hand, PPARγ agonists do inhibit the growth in vitro of a broad spectrum of CRC cell lines, many of which harbor loss-of-function mutations in APC. However, these cell lines are derived from invasive adenocarcinomas that have evolved genetic and epigenetic perturbations in multiple pathways and thus represent imperfect models of early intestinal adenomas. In this regard, cell lines derived from a hybrid of the Immorto and APCMin mice have previously been isolated (5), and it would be of interest to determine whether in these cells PPARγ activation also results in growth inhibition and differentiation. Finally, because all of these studies rely on compounds that may have biochemical targets independent of PPARγ, it will also be of major interest to determine whether any results seen with PPARγ ligands can be replicated in mice genetically deficient in PPARγ [either PPARγ −/+ mice, because homozygous null PPARγ are embryonic lethal (2), or in mice with a colon-targeted genetic ablation of PPARγ].

Recent studies on the genetic status of the PPARγ locus in CRCs appear to support the hypothesis that PPARγ has a tumor suppressive, rather than enhancing, role in the colon. Sarraf et al. (18) reported that 8% of primary colorectal tumors contained a loss-of-function point mutation in one allele of the PPARγ gene. Four unique mutations in PPARγ were identified in the study; one resulted in a truncated protein that lacked the entire ligand binding domain, whereas the other three mutations caused defects in the binding of either synthetic or natural ligands. In addition, a chromosomal translocation was recently identified in a subset of human thyroid follicular carcinomas that produces a fusion protein between PAX8 and PPARγ; this oncoprotein is thought to promote thyroid carcinoma formation in part by acting as a dominant-negative inhibitor of wild-type PPARγ (9). Collectively, these findings argue that PPARγ may be one of many factors that prevent the abnormal growth and differentiation associated with malignancy.

In summary, cumulative evidence suggests that, as is the case with many other cell types, activation of PPARγ reduces the malignant potential of CRC cells. Nevertheless, it remains to be determined whether the protumorigenic effects of PPARγ activation seen in the APCMin mice are simply an idiosyncratic result of one particular model system or rather is a finding that can be generally applied to humans. The controversy will likely be settled by the results of ongoing experiments examining the biological role of PPARγ in the colon and in CRC as well as studies evaluating the incidence of CRC in the cohort of humans chronically taking synthetic PPARγ ligands for diabetes.


We thank the T. J. Martell Foundation and the National Colorectal Cancer Research Alliance for generous support.


  • This work was supported, in part, by the National Institutes of Health Grants RO1DK-47279 (to R. N. DuBois), P030 ES-00267–29 (to R. N. DuBois), and P01CA-77839 (to R. N. DuBois).

  • R. N. DuBois is a recipient of a Veteran's Affairs Research Merit Grant and is the Mina C. Wallace Professor of Cancer Prevention.

  • Address for reprint requests and other correspondence: R. N. DuBois, Dept. of Medicine/GI; MCN C-2104, Vanderbilt University Medical Center, 1161 21st Ave. South, Nashville, TN 37232–2279 (E-mail:raymond.dubois{at}

  • 10.1152/ajpgi.00486.2001


View Abstract