Peroxisome proliferator-activated receptor γ agonist troglitazone induces colon tumors in normal C57BL/6J mice and enhances colonic carcinogenesis in Apc1638 N/+Mlh1+/− double mutant mice


  • The authors declare that they have no competing financial interests.


The role of the nuclear peroxisome proliferator-activated receptor-γ (PPAR-γ) in colon tumorigenesis remains controversial. Notwithstanding evidence that PPAR-γ ligands impede murine colorectal carcinogenesis, PPAR-γ agonists have been shown to enhance in vivo tumor formation in mouse models of human colon cancer. Our study was designed to determine whether troglitazone (TGZ) induces colonic tumor formation in normal C57BL/6J mice and enhances colorectal carcinogenesis in double mutant Apc1638N/+Mlh1+/− mice fed a standard AIN-76A diet. We report herein that not only does TGZ enhance carcinogenesis in the large intestine of mutant mice predisposed to intestinal carcinogenesis but TGZ also induces colonic tumors in normal mice without gene targeting or carcinogen administration. This observation indicates that preexisting mutational events are not necessary for induction of colonic tumors by activated PPAR-γ in vivo. © 2005 Wiley-Liss, Inc.

Peroxisome proliferator-activated receptor, a member of the nuclear receptor superfamily is a ligand-activated transcription factor that heterodimerizes with the retinoid X receptor and binds to PPAR response elements in the promoter regions of target genes, controlling adipocyte differentiation, systemic glucose levels and lipid homeostasis.1

Studies of PPAR-γ function have been facilitated by the synthesis of high affinity ligands, i.e., the thiazolidinedione class of antidiabetic drugs that includes troglitazone (TGZ), pioglitazone and rosiglitazone.2 Ligand activation and signaling via PPAR-γ causes growth arrest and induces a more differentiated phenotype in a large variety of cultured human cancer cells, including colon cancer cells (reviewed in references 3–6). In vivo studies have shown that PPAR-γ activation inhibits tumor growth in a xenograft model of prostate cancer7 and attenuates breast cancer chemically induced in rodents.8

In contrast, the effects of PPAR-γ agonists on colon tumorigenesis in vivo remain highly controversial.5, 6, 9, 10, 11 Thus, treatment with PPAR-γ agonists inhibits the growth of human colonic cancer cells implanted in immunodeficient mice12 and decreases the formation of colonic aberrant crypt foci (ACF) and tumors in azoxymethane (AOM)-treated mice.13 Furthermore, heterozygous PPAR-γ knockout mice treated with AOM exhibit an increased number of colonic polyps compared with wild-type littermates,14 suggesting an anticancer action of wild-type PPAR-γ. However, PPAR-γ ligands including TGZ when supplemented to a standard diet have promoted colonic tumorigenesis in ApcMin/+ mice.15, 16, 17

We report herein that not only does TGZ enhance carcinogenesis in the large intestine of mice with both Apc and Mlh1 mutations, but TGZ also induces colonic tumors in normal C57BL/6J mice without carcinogen administration or gene targeting.

Material and methods

Animals and diets

Two mouse strains were used in this study. Wild-type C57BL/6J mice were purchased from The Jackson Laboratory (Bar Harbor, ME) at 4 weeks of age and Apc1638N/+Mlh1+/− mice were produced by gene targeting as previously described.18 The PPAR-γ agonist troglitazone (Rezulin) was provided by the Parke Davis Pharmaceutical Research (Warner-Lambert Co., Ann Arbor, MI). Standard control AIN-76A diet and AIN-76A supplemented with 0.2% TGZ (AIN+TGZ) were used. The TGZ dose selected for our study was used previously to induce colonic tumors in C57BL/6J-Apcmin/+ mice.15, 16, 17 All diets were prepared by Research Diets, Inc. (New Brunswick, NJ) and stored at −20°C until use.

Animal treatments

On arrival, mice were housed in the Laboratory Animal Research Center (LARC) of The Rockefeller University, an American Association for Accreditation of Laboratory Animal Care facility complying with USDA regulations and NIH guidelines for the Care and Use of Laboratory Animals. The mice were maintained on a 12 hr light-dark cycle with water ad libitum and fed AIN-76A diet during the acclimatization period. At 6 weeks of age, all mice were randomly sorted into 1 of 2 dietary groups and fed one of 2 diets for 6 months. The number of mice allotted for dietary treatments and the ratio of male-to-female mice are shown in Table I. The mice were sacrificed by neck dislocation at completion of dietary treatments.

Table I. Tumor Incidence in the Gastrointestinal Tract of Normal C57BL/6J Mice and Apc1638N/+Mlh1+/− Mutant Mice
Mouse groupnGender (M:F)2Number (%) of mice with tumorsNumber (%) of mice with tumors
Small intestineLarge intestine
  • 1n: Number of mice, TGZ: Troglitazone.

  • 2

    Ratio of male(M) to female (F) mice.

  • 3

    p < 0.001; Comparisons were made by Fisher exact probability test between AIN-76A+TGZ (AIN-76A diet with 0.2% TGZ) and control diet (unmodified AIN-76A).

C57BL/6J mice 
AIN-76A100:100 (0)0 (0)0 (0)
AIN-76A+TGZ90:99 (100)30 (0)9 (100)3
AIN-76A121:1.410 (83)9 (75)2 (17)
AIN-76A+TGZ101:0.410 (100)8 (80)10 (100)3

Tumor identification

The gastrointestinal tract was removed, opened longitudinally and fixed in 10% buffered formalin. Gross specimens were examined under a dissecting microscope, and tumor incidence, multiplicity and location in the small and large intestine were determined. Tumor volume (in mm3 units) was measured using a microscopic eyepiece grid. All tumors of the large intestine and representative tumors from the small intestinal tract were taken for processing and embedded in paraffin. Sections (4 μm) of paraffin-embedded tumor tissues were examined histologically after staining with hematoxylin and eosin. Diagnoses were based on the WHO classification.19 No tumors were found in other organs, which included lungs, liver, spleen, pancreas, bladder, uterus, ovaries, mammary glands and skin.


Fisher exact probability, Mann-Whitney test and binomial exact calculations were used to compare data from different diet groups. Differences were considered statistically significant when p < 0.05.


No changes in food consumption, overall body weight, activity and behavior were observed in C57Bl/6J and Apc1638N/+Mlh1+/− mice fed AIN-76A containing TGZ compared with control mice maintained on unmodified standard diet.

Normal C57BL/6J, mice fed a standard AIN-76A diet supplemented with 0.2% TGZ for 6 months developed benign and malignant tumors in the large intestine (p<0.001) compared with normal controls not fed TGZ (Table I). All tumors were located in the cecum (Fig. 1). Tumor multiplicity and volume also markedly increased (p<0.001, Table II) in normal mice treated with TGZ. No tumors were found in the small intestine.

Figure 1.

Tumor incidence in different regions of the gastrointestinal tract of normal C57/6J mice and Apc1638N/+Mlh1+/− mice fed for 6 months a standard AIN-76A diet or AIN-76A diet supplemented with 0.2% troglitazone.

Table II. Multiplicity and Volume of Intestinal Tumors in Normal C57BL/6J Mice and Apc1638N/+Mlh1+/− Mice Fed Troglitazone
Mouse groupnGender (M:F)1TotalSmall intestineLarge intestine
  • 1

    Ratio of male (M) to female (F) mice. n: Number of mice, TGZ: Troglitazone.

  • 2

    Tumor number per mouse, mean ± SEM.

  • 3

    p<0.001; comparisons were made between AIN-76A + TGZ (AIN-76A supplemented with 0.2% troglitazone) and corresponding control diet (unmodified AIN-76A) using Mann-Whitney and exact binomial calculation.

  • 4

    Tumor volume (mm3), mean ± SEM.

Tumor multiplicity2
C57BL/6J Mice
AIN-76A100:100.00 ± 0.000.00 ± 0.000.00 ± 0.00
AIN-76A+TGZ90:91.10 ± 0.1130.00 ± 0.001.10 ± 0.113
AIN-76A121:1.43.83 ± 1.283.67 ± 1.300.16 ± 0.11
AIN-76A+TGZ101:0.42.60 ± 0.431.30 ± 0.401.30 ± 0.213
Tumor volume4
C57BL/6J Mice
AIN-76A100:100.00 ± 0.000.00 ± 0.000.00 ± 0.00
AIN-76A+TGZ90:93.05 ± 0.4930.00 ± 0.003.05 ± 0.493
AIN-76A121:1.415.26 ± 5.5815.05 ± 5.880.21 ± 0.21
AIN-76A+TGZ101:0.410.45 ± 4.296.40 ± 4.674.05 ± 0.833

Apc1638N/+Mlh1+/− mice fed with TGZ (0.2% in AIN-76A diet) for 6 months had increased tumor incidence in the large intestine (4.88-fold, p<0.001, Table I) compared to double mutant control mice fed unmodified AIN-76A diet. The tumors were located in the cecal region and distal colon (Fig. 1). TGZ administration also increased colonic tumor multiplicity (7.1-fold, p<0.001, Table II) and enhanced tumor volume (18.3-fold, p<0.001, Table II) compared to Apc1638N/+Mlh1+/− mice fed unmodified AIN-76A. The feeding of AIN-76A diet supplemented with TGZ resulted in fewer small intestinal tumors of smaller size, but differences were not statistically significant compared to unmodified standard diet (Table II, Fig. 1).

The histopathological types of colonic tumors induced in mice fed TGZ were determined. Of 10 tumors examined in normal C57BL/6J mice, 3 were early invasive carcinomas, 6 were flat tubular adenomas and 1 was a microadenoma. All 12 colonic tumors examined in TGZ-fed Apc1638N/+Mlh1+/− mice were flat tubular adenomas. Figure 2a shows an early invasive carcinoma in the cecum of a normal C57BL/6J mouse 6 months after TGZ treatment. Higher magnification of the same tumor shows invasion of submucosa with an inflammatory, fibroblastic reaction (Fig. 2b,c).

Figure 2.

Early invasive carcinoma in the cecum of a normal C57BL/6J mouse after 6 months of TGZ feeding. (a) Tumor at the junction of the cecum and proximal colon with an indented center (arrow; ×1.75 original magnification). (b) Histological features of the same tumor. The tumor infiltrated into the submucosa with ulceration in the central part (×40 original magnification). (c) Higher magnification of the same tumor showing features of carcinoma cells and an inflammatory fibroblastic reaction (×100 original magnification).


The main observation of our study is that dietary administration of TGZ for 6 months resulted in colonic tumor development with early invasive carcinomas in normal C57BL/6J mice. The same dietary treatment enhanced colonic tumorigenesis in C57BL/6J genetically-altered mice bearing mutations in both Apc and Mlh-1 genes. Apc1638N/+ mice and mice with the Mlh1+/− genotype alone or in combination with Apc1638N/+ rarely develop colonic tumors18, 20 (present results). Apc1638N/+ mice develop polyps from 3 months of age and each mouse develops relatively few tumors in the small intestine;18, 20Mlh1+/− mice have also few intestinal tumors.18 The double mutant Apc1638N/+Mlh1+/− mice have a comparable low number of intestinal tumors.18 The carcinogenic effect of TGZ in the colon of Apc1638N/+Mlh1+/− mice is of interest since these double mutant mice are widely used as a model for hereditary non polyposis colon cancer in humans.21, 22

We have also observed that TGZ-treated Apc1638N/+Mlh1+/− mice had fewer tumors in the small intestine. Although the difference was not statistically significant when compared with double mutant mice fed unmodified standard diet, it is consistent with the recurring observation that the small and large intestines respond differently to drug treatment (reviewed in reference 23). Thus, piroxicam markedly decreases the number of polyps in the small intestine of ApcMin/+ mice; however, piroxicam increased the number of colonic tumors in the mutant mice (reviewed in reference 23). In addition, we have recently observed an opposite regional response to tumorigenesis after sulindac in the small intestine and colon of ApcMin/+ mice. These mice developed fewer tumors in the small intestine compared to untreated controls; in contrast, sulindac significantly increased the incidence, multiplicity and volume of tumors in the large intestine.24

All normal C57BL/6J mice used in our study were females. Double mutant mice consisted of both males and females because of supply limitations. It is noteworthy that TGZ enhanced colonic tumor formation in both male and female genders.

While current findings on the activity of the above drugs in humans and rodents do not indicate precise mechanisms for their site-specific effects in the intestinal tract, it has been suggested that functional discrepancies between the small and the large intestine arise, at least partly, from differences in the expression/activities of enzymatic pathways involved in responsiveness to drugs (reviewed in reference 23).

The procarcinogenic activity of dietary TGZ in normal mice was restricted to the large intestine in the region of the cecum. This observation is consistent with previous findings showing that dietary TGZ and other PPAR-γ ligands promote tumorigenesis in ApcMin/+ mice only in the colon.15, 16, 17 Gupta et al.25 have recently reported that treatment of ApcMin/+ mice with a selective agonist of PPAR-δ led to a marked increase in polyp number and size in the small intestine with no changes in the colon.

Mouse strains, doses and duration of TGZ administration may have an important role in tumor development in rodent models of human intestinal carcinogenesis. Of note, normal C57BL/6J mice used in our study are considered to be relatively resistant to carcinogen-induced neoplasia. Other investigators did not observe colonic tumorigenesis in normal BALB/c mice fed TGZ13 when a lower dose of TGZ and a shorter 5-week duration of dietary treatment were used compared to the 6 months treatment in our study. Time dependence of colon neoplasia development in normal mice was also seen in a previous dietary study, i.e., normal mice fed a high-fat Western-style diet developed tumors mostly in the colon only after 18 months of treatment.26 However, no colonic tumors were observed after long-term feeding of normal mice with the PPAR-γ agonist pioglitazone.13

In our study, TGZ promoted formation of flat colonic adenomas believed to be more aggressive precursors of carcinomas in humans with increased propensity to high-grade dysplasia relative to their small size.27 Mutational analysis using the protein truncation test has revealed multiple APC mutations in flat adenomas and aberrant expression of key cell cycle components in a significant subset of these adenomas.28, 29

Our findings not only support previous observations15, 16, 17 that activated PPAR-γ enhanced colonic carcinogenesis in genetically altered mice but also show for the first time that normal mice not exposed to a carcinogen or gene targeting respond to TGZ administration with formation of tumors in the large intestine, clearly indicating that preexisting mutational events are not necessary for in vivo cancer-inducing action of activated PPAR-γ.

The molecular basis of this carcinogenic effect remains to be elucidated. In our study, we have added a PPAR-γ-activating ligand to a standard diet. Since PPAR-γ activation in vivo may provide a molecular link between a high fat diet and increased risk of colorectal cancer, an attractive hypothesis is that the promoting effects of Western diets on intestinal neoplasia are mediated through procarcinogenic fatty acids that are able to act as endogenous PPAR-γ agonists.1 Saez et al.30 have recently presented findings suggesting that the tumor-promoting action of PPAR-γ is associated with upregulation of the Wnt signaling pathway in mammalian cells, an interesting possibility consistent with the recurring observation that altered activation of the Wnt signaling cascade is commonly observed in human tumors, including colonic tumors.31

The possibility must be also considered that some biological effects of TGZ at the dietary concentration used in our study were associated with PPAR-γ-independent pathways. Several hypotheses, recently reviewed,11 have been proposed for actions of PPAR-γ ligands, such as TGZ, not related to PPAR-γ activation.

Although TGZ is presently not used for treatment of type 2 diabetes in humans, additional PPAR-γ synthetic agonists such as rosiglitazone and pioglitazone are now administered as hypoglycemic compounds to humans with type 2 diabetes,2 a disease associated with increased risk for colorectal cancer.32, 33, 34 The available results of studies in mice that were treated with these thiazolinedione derivatives remain controversial. Thus, while rosiglitazone promoted colonic tumors in ApcMin/+ mice,16, 17 the same PPAR-γ synthetic agonist inhibited AOM-induced ACF formation in mouse colon.13 Pioglitazone was recently shown to promote tumor formation in ApcMin/+ mice;17 in contrast, this PPAR-γ agonist inhibited intestinal polyp formation in mice having truncation mutations in the Apc gene35, 36 and impeded ACF formation in AOM-treated mice.13 Significantly, recent results indicate that the controversy pertaining to cancer-promoting action of PPAR-γ in vivo is not restricted to the large intestine. Thus, while heterozygous PPARγ+/− mice acquired increased susceptibility to DMBA-induced mammary, ovarian and skin carcinogenesis,37 suggesting a protective anticancer action of wild-type PPAR-γ in these tissues, breeding of transgenic mice that express a constitutively active form of PPAR-γ to a transgenic mouse strain prone to breast cancer enhanced mammary gland tumor development.30

In summary, our findings show for the first time that TGZ induces colon cancer in normal mice. Key questions to be addressed pertain to the mechanisms through which activated PPAR-γ modulates tumor growth in mice in vivo and their possible relationship to humans treated with PPAR-γ agonists.