A high-fat diet generates alterations in nuclear receptor expression: Prevention by vitamin A and links with cyclooxygenase-2 and β-catenin


  • Barbara Delage,

    Corresponding author
    1. Laboratory of Food and Colon Carcinogenesis, Unit of Nutrition and Cellular Signalling, University Bordeaux 1, Talence, France
    • Laboratory of Food and Colon Carcinogenesis, Unit of Nutrition and Cellular Signalling, University Bordeaux 1, Avenue des Facultés, 33405 Talence cedex, France
    Search for more papers by this author
    • Fax: +33-0-5-40-00-27-76 or +33-0-5-56-37-03-36

  • Céline Bairras,

    1. Laboratory of Food and Colon Carcinogenesis, Unit of Nutrition and Cellular Signalling, University Bordeaux 1, Talence, France
    Search for more papers by this author
  • Benjamin Buaud,

    1. Laboratory of Food and Colon Carcinogenesis, Unit of Nutrition and Cellular Signalling, University Bordeaux 1, Talence, France
    Search for more papers by this author
  • Véronique Pallet,

    1. Laboratory of Food and Colon Carcinogenesis, Unit of Nutrition and Cellular Signalling, University Bordeaux 1, Talence, France
    Search for more papers by this author
  • Pierrette Cassand

    1. Laboratory of Food and Colon Carcinogenesis, Unit of Nutrition and Cellular Signalling, University Bordeaux 1, Talence, France
    Search for more papers by this author


Epidemiologic studies suggest that intake of high energy from fat, inducing overweight, increases the risk of cancer development and promotes colon carcinogenesis. It is therefore important to understand which parameters are affected early on by a high-fat diet in order to devise and improve protective nutritional strategies. We investigated the effect of high energy/fat intake on colon mucosa of male Wistar rats induced by a single 1,2-dimethylhydrazine (DMH) injection. Aberrant crypt foci (ACF) were numbered and modifications in cyclooxygenase-2 (COX-2) and β-catenin levels assessed. Peroxisome proliferator– and retinoic acid–activated receptors (PPAR and RAR, RXR) are key transcription factors regulating gene expression in response to nutrient-activated signals. A short-term study was designed to evaluate whether alterations in mRNA expression of nuclear receptors can be detected at the beginning of the weight gain phase induced by an appetizing hyperlipidic diet (HLD). HLD consumption induced early downregulation of PPARγ (−33.1%) and RARβ (−53.1%) mRNA expression concomitant with an increase in levels of COX-2 (+45.5%) and β-catenin (+84.56%) and in the number of ACF (191.56 ± 88.60 vs. 21.14 ± 11.64, p < 0.05). These findings suggest that HLD increases ACF occurrence, possibly through alterations in the mRNA expression profile of nuclear receptors. Moreover, the use HLD rich in retinyl esters or supplemented with all-trans retinoic acid led to a reduction in the number of ACF. Vitamin A also prevented HLD-induced alterations and the increase in levels of COX-2 and β-catenin. The present observations show a protective role for vitamin A against disturbances associated with HLD exposure in induced colon carcinogenesis. © 2005 Wiley-Liss, Inc.

The incidence of obesity and colon cancer is increasing in all industrialized countries. Colon carcinogenesis is a multifactorial disease, and diet is strongly involved in its etiology. A prospective cohort study has reported that increased body weight was associated with increased death rates form cancer.1 Thus, the high prevalence of obesity, explained by a lifestyle characterized by high energy/fat intake, led to increased risk of colon cancer development. These conclusions are reinforced by data suggesting that the number of preneoplastic colonic lesions2, 3, 4 induced by a high-fat diet is reduced by restriction of energy intake5 or by fiber- and vitamin-rich diets.6, 7 These data illustrate the potential of nutritional methods for reducing the deleterious consequences of high fat/energy intake on the risk of colon cancer.

It is likely that diet affects cancer occurrence at the level of the cell signaling pathways rather than at the level of mutations. Nuclear receptors have a central role in the regulation of gene expression in response to diet. Thus, growing evidence points to the involvement of nuclear receptors in colon tumorigenesis. Indeed, they function as ligand-activated transcription factors and regulate a wide range of target genes that affect almost all biologic processes. Nuclear receptors belonging to the class II superfamily heterodimerize with the obligate partner RXR and bind to specific DNA sequences. Among numerous free fatty acids, PUFAs bind PPARs with a higher affinity than SFAs and MUFAs. Thus, PPARs represent key elements mediating, at least in part, the effects of a high-fat diet on gene regulation. However, the wide range of PPAR activity remains poorly understood. Various gain-of-function experiments argue for a protective role of PPARγ, while others support the opposite idea.8 Vitamin A and its most potent natural form, RA, also regulate a wide spectrum of processes, including proliferation, differentiation and development. Vitamin A is now recognized as a promising agent for chemoprevention and chemotherapy for a variety of human cancers, including epithelial cancers.9 This nutrient acts mainly by activating RAR and RXR. Among the different nuclear receptor isotypes, the involvement of PPARγ and RARβ is the most studied in the context of colon cancer development.10, 11 Although less investigated, other isoforms, such as PPARγ, RARα and RXRα, could also participate in the modulation of cancer risk.12, 13, 14

In addition to the direct regulation of target genes, nuclear receptor expression may affect 2 signaling pathways related to colon carcinogenesis involving COX-2 and β-catenin. COX-2, which catalyzes prostaglandin biosynthesis, is induced by dietary fatty acids and overexpressed in tumors. Alterations in the Wnt pathway lead to nuclear accumulation of β-catenin, resulting in abnormal gene expression and tumoral promotion. Both PPARγ and vitamin A–activated nuclear receptors are involved in the regulation of these 2 proteins.15, 16, 17, 18 Although some investigations have examined the use of vitamin A and its natural and synthetic derivatives in the treatment and prevention of many cancers,19 very few have explored the potential efficiency of retinoids in modulating the risk of colon cancer.

The present study is focused on the hypothesis that overfeeding a high fat/energy diet affects nuclear receptor expression in colonic mucosa and therefore may facilitate the occurrence of ACF. Because the model of Min mice might not be suitable for studying nuclear receptor expression,16, 20 we used DMH-induced Wistar rats. The occurrence of ACF, nuclear receptor expression and levels of COX-2 and β-catenin were analyzed in these animals to investigate the connection between diet, receptor expression and abnormal proliferation.


AC, aberrant crypt; ACF, aberrant crypt foci; ATRA, all-trans retinoic acid; COX-2, cyclooxygenase-2; Cp, crossing point; DMH, 1,2-dimethylhydrazine; FAM, 6-carboxyfluorescein; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HLD, hyperlipidic diet; Min, multiple intestinal neoplasia (mouse); MUFA, monounsaturated fatty acid; PPAR, peroxisome proliferator–activated receptor; PUFA, polyunsaturated fatty acid; PVDF, polyvinylidene difluoride; RA, retinoic acid; RAR, retinoic acid–activated receptor; RXR, retinoid X receptor; SFA, saturated fatty acid; TAMRA, 6-carboxytetramethylrhodamine; VAHLD, vitamin A–enriched hyperlipidic diet.

Material and methods

Animals and diets

Ninety male Wistar rats (7 weeks old) were purchased from Harlan (Gannat, France). They were randomly divided into 6 experimental groups and housed singly with a 12:12 hr light–dark cycle at 50% humidity and 21 ± 1°C. Each rat was weighed 3 times weekly. All animals were fed and given water ad libitum. Food intake was recorded daily.

After 7 days of acclimatization to the housing conditions, each group of rats was fed one of 3 diets (Table I): the standard diet or one of 2 high-fat diets, HLD and VAHLD. The first was a standard rodent diet (A04-type pellets; Usine d'Alimentation Rationelle, Villemoisson-sur-Orge, France), while the high-fat diets consisted of a selection of palatable human foods that induce voluntary and spontaneous hyperphagia in rats. The high-fat diets were highly comparable to a diet-induced obesity model called “cafeteria diet”, previously described by Berraondo et al.21 Animals receiving the high-fat diets were presented daily with a fresh mix of the following items: pâté, bacon, chocolate, potato chips, biscuits and pelleted chow in a proportion of 2:1:1:1:1:1. In the high-fat diets, energy supplied as lipids represented 59% of the total energy intake and consisted of 43% SFA, 39% MUFA and 18% PUFA. Lipid composition was determined by gas chromatography.

Table I. Composition of Experimental Diets
 Standard dietHigh-fat diets
  • 1

    Values are in percent of total energy provided.

  • 2

    Lipids were extracted from food according to the method of Folch et al.,58 transmethylated and submitted to gas chromatography.

  • 3

    Values are in IU/g of diet (1 μg retinol equivalents = 3.33 IU of retinol or retinyl esters).

  • 4

    Vitamin A levels were determined by normal-phase HPLC according to Norme Francaise European Norm 12823-1 (Institut des Corps Gras, Pessac, France). Data are provitaminA activities in the different diets and do not take account of the ATRA administrated by gavage.

Provitamin A activity347.57.527.3

The quantity of vitamin A in the high-fat diets was different, depending on the type of pâté used: the HLD, prepared with a pâté rich in ham, contained 7.5 UI/g of chow and the VAHLD, prepared with a pâté rich in liver, contained 27.3 UI/g of chow. The quantity found in the standard diet was 7.5 UI/g of chow. Vitamin A quantity expressed as provitamin A activity was determined by HPLC.

Experimental protocol

Rats were fed one of the 3 diets during a 2-month period. After 1 month of diet, animals were given one i.p. injection of 15 mg/kg of DMH (Fluka, St. Quentin Fallavier, France). Once every 2 days, in each group, half of the animals received intragastric administration of ATRA (Sigma, St. Quentin Fallavier, France) in 500 μl of coconut oil at a concentration of 1 mg/kg body weight (Fig. 1). ATRA suspensions were prepared fresh daily under dim illumination and used not more than 45 min after preparation. Control rats were gavaged with 500 μl of the coconut oil vehicle.

Figure 1.

Experimental design. ATRA was administered intragastrically, 1 mg/kg body weight once every 2 days.

Rats were killed 2 months after the beginning of the experiment. Colons were rapidly removed, slit open longitudinally and cleaned with ice-cold 0.9% NaCl solution. Except for ACF analysis, mucosal scrapings were quickly frozen in liquid nitrogen and stored at −80°C until analysis.

Assessment of ACF in the colon

The full length of the colon was examined for ACs. Tissue was washed with physiologic saline, cut open longitudinally, pinned out flat and fixed in 10% buffered formalin. Colon samples were stained with 0.5% methylene blue for 5 min, rinsed, placed on a glass slide and examined microscopically using ×40 magnification for assessment of the number of ACs and ACF following a procedure previously described.22 The criteria for the identification of ACs were (i) increased size, (ii) thicker epithelial cell lining and (iii) increased pericryptal zone relative to normal crypts. Crypt multiplicity was defined as the number of crypts in each focus and scored blindly by a single observer. The efficacy end point was inhibition of the overall occurrence of ACs and ACF.

RNA extraction and reverse transcription

Total RNA was isolated from colonic mucosa using an extraction kit (RNAgents Total RNA Isolation System; Promega, Charbonnières, France) according to the manufacturer's protocol. RNA samples were reversed-transcribed as follows: 2 μg of total RNA were mixed with RNasin (0.5 U, Promega) and DNAse I (0.5 U, Roche, Meylan, France) and incubated for 15 min at 37°C. Reverse primers (0.75 μM of each) were added and incubated for 10 min at 70°C. ImProm-II 5X reaction buffer (1×, Promega), MgCl2 (10 mM, Promega), dNTP (0.5 mM each, Roche) and ImProm-II Reverse Transcriptase (10 U, Promega) were added for 1 hr at 42°C. Total volume was 20 μl, and each target mRNA was co-reverse-transcribed with β2-microglobulin mRNA except for RARβ, which was co-reverse-transcribed with GAPDH mRNA. Parallel reactions for each RNA sample were run in the absence of ImProm-II Reverse Transcriptase to assess the degree of contaminating genomic DNA. Moreover, RT-PCRs without any RNA sample were also assessed, to verify the absence of other contamination.

Analysis of gene expresion using real-time PCR

Real-time quantitative RT-PCR involving LightCycler technology (Roche, Mannheim, Germany), using SYBR green detection, was performed according to the protocol recommended by the manufacturer and previously described.23 PCR products were analyzed by electrophoresis on a 1.5% agarose gel (Sigma). The identity and specificity of amplified products were assessed by sequencing with the Dye Terminator Reaction Cycle Kit on an ABI PRISM 377 automated DNA sequencer (Perkin-Elmer, Norwalk, CT).

Forward and reverse primer sequences for PPARγ and RXRα were similar to those used by Groubet et al.24 The nucleotide sequences for the other primer pairs were as follows: β2-microglobulin sense 5′-GCCCAACTTCCTCAACTGCTACG-3′, antisense 5′-GCATATACATCGGTCTCGGTGGG-3′; PPARδ sense 5′-CGCAACAAGTGTCAGTACTG-3′, antisense 5′-CCAAAGCGGATAGCGTTGTG-3′; RARα sense 5′-GCCTCGAATCTACAAGCCTTGC-3′, antisense 5′-GGATACTGCGTCGGAAGAAGC-3′; COX-2 sense 5′-GCAAAGGCCTCCATTGACCAGAG-3′, antisense 5′-CGGGATACAGTTCCATGGCATCG-3′.

Quantification data were analyzed using the LightCycler analysis software, version 3.5. In this analysis, the end point used in the real-time PCR quantification, Cp, was defined as the PCR cycle number that crosses an arbitrarily placed signal threshold. The standard curve was a plot of the Cp vs. the amount of initial cDNA used for amplification. The Cp was always the same for a given dilution of β2-microglobulin or GAPDH cDNA, whatever the dietary conditions, demonstrating that reference mRNA expression was not altered by diets and could be used as a normalizer for data from target mRNA. The relationship between the Cp and the initial amount of cDNA was linear. The correlation coefficient (r) was 1, and PCR amplification efficiencies of the target and the housekeeping gene were similar and close to 100%. Standard curves were used to estimate the concentration of both the target and the reference gene in each sample. Then, the results were normalized by the ratio of the relative concentration of target to that of β2-microglobulin in the same sample.

Quantification of RARβ mRNA by TaqMan PCR

The TaqMan PCR technique measures an accumulating PCR product in real time using a dual-labeled TaqMan fluorogenic probe. The experiment was conducted as previously described23 using the same primers and fluorogenic probes purchased from Proligo (Paris, France). Each probe was synthesized with the fluorescent reporter dye FAM attached to the 5′ end and the quencher dye TAMRA attached to the 3′ end.

Quantification data were analyzed using the LightCycler Relative Quantification software (Roche, Mannheim, Germany) because RARβ and reference (GAPDH) cDNA amplification did not have the same efficiencies (Fig. 2a,b). This software provided calibrator-normalized relative quantification, including PCR efficiency correction. The calibrator was chosen among the rats fed standard diet. RNA was prepared and reverse-transcribed as described above. cDNA was used as the calibrator in all experiments for RARβ quantification. Results are expressed as the target/reference ratio divided by the target/reference ratio of the calibrator.

Figure 2.

(a) Fluorescence detection vs. cycle number of amplification. The figure reveals distinct target and reference PCR efficiencies. (b) Standard curves for target (RARβ) and reference (GAPDH) are created by a dilution series of a nucleic acid. The figure shows the initial cDNA log concentration plotted vs. Cp for the detection of significant fluorescence and reveals nonlinear standard curves. LightCycler Relative Quantification software takes account of the differences in PCR efficiencies according to the concentration of cDNA.


About 0.2 g of colonic mucosa was homogenized with an Ultra-Turrax T25 homogenizer in 3 ml of extraction buffer (Igepal CA-630 1%, sodium deoxycholate 0.5%, SDS 0.1%, EDTANa2 13% and a cocktail of protease inhibitor in PBS, pH 7.4). After centrifugation at 11,000g for 15 min at 4°C, the supernatant was collected and the protein concentration measured at 595 nm using a bicinchoninic acid protein assay reagent (Pierce, Rockford, IL) with BSA as a standard. Proteins were subjected to 10% SDS-PAGE and then transferred to PVDF membranes (Bio-Rad, Yvry-sur-Seine, France) in semidry medium at 90 mA for 2 hr. Transfer of proteins to PVDF membranes was confirmed by staining gels with Coomassie brilliant blue R-250 0.1% in methanol 40%, acetic acid 10%. Membranes were incubated with anti-β-catenin (developed in rabbit, Sigma) or anti-COX-2 (developed in rabbit; Santa Cruz Biotechnology, Tebu-bio, Le Perray en Yvelines, France) or anti-β-actin antibodies (developed in mouse, Sigma). Immunoreactive polypeptide bands were visualized enzymatically in a secondary antibody reaction using peroxidase-conjugated antirabbit or antimouse IgG (Sigma). Peroxidase substrates were added with Western Lightning Chemiluminescence Reagent Plus (Perkin-Elmer Life Sciences, Courtaboeuf, France). The chemiluminescence reaction was visualized by autoradiography (Autoradiography films biomax light 1, Perkin-Elmer Life Sciences) and evaluated with an image analysis software (Bio 1D; Vilbert Lourmat, Marne La Vallée, France). The software provides data which are the sum of the intensities of the pixels inside the volume (i.e., the signal obtained by the light emission captured on film) × the area of a single pixel (in mm2). A range of protein concentrations was loaded onto the gels, and different times of film exposure were used to ensure linear responses. β-Actin was revealed to assess the quality of loadings and transfer.

β-Catenin immunohistochemistry

To determine whether diets and/or ATRA administration affect localization of β-catenin, 7 μm sections of paraffin-embedded colon tissues from 6 rats of each dietary group were mounted on slides, deparaffinized, rehydrated and washed in PBS (pH 7.2). Antigen retrieval was achieved using microwave heating at 800 W for 8 min in citrate buffer (10 mM, pH 6). Endogenous peroxidase activity was quenched by incubation with 3% hydrogen peroxide in PBS for 30 min, and then sections were washed and blocked with 10% horse serum in PBS for 20 min. Sections were incubated overnight in a moist chamber at 4°C with the primary antibody, anti-β-catenin (developed in rabbit, Sigma) at a dilution of 1:100. Slides were washed in PBS and then incubated with a biotinylated goat antirabbit secondary antibody (Jackson Immunoresearch Laboratories, West Grove, PA) for 1 hr at room temperature at 1:500. Antigen–antibody complexes were detected with the streptavidin-biotin peroxidase (Amersham, Orsay, France) method using 3-amino-9-ethylcarbazole as a chromogenic substrate (AEC kit; Vector, Société Abcys, Paris, France). After washing in tap water, some slides were counterstained with hematoxylin for 1 min. One section on each slide had the primary antibody step eliminated. Immunostained sections were examined by light microscopy.

Statistical procedure

Experimental data were analyzed by one-way ANOVA followed by Student's t-test using the Statgraphics Plus software (Sigma-Plus, Paris, France). Experimental data are expressed as means ± SEM, and differences at p < 0.05 were considered statistically significant.


General observations

As expected, within 2 months, the body weights of animals fed the high-fat diets (HLD and VAHLD) were significantly higher than in the standard diet group. However, weights started to differ significantly only in the second period of the experimental procedure, i.e., at the end of the second 4-week period (p < 0.05) (Table II). Energy provided by high-fat diets was greater than that of the standard diet (2,939 kJ/100 g vs. 1,868 kJ/100 g). Moreover, animals fed high-fat diets consumed a greater amount of food than those controls. Because our aim was to study the early molecular alterations linked to diet, the experimental procedure was ended 1 month after chemical initiation and after 2 months of diet consumption, at the beginning of significant weight gain. Intragastric administration of ATRA did not modify the weight gain of animals in each diet group. No difference in weight gain was observed in the HLD and VAHLD groups, which received different amounts of vitamin A. Concomitantly with these observations, the high-fat diets induced an increase in fat mass for both visceral and s.c. adipose tissues compared to the standard diet (data not shown).

Table II. Dietary Characteristics and Body Weights of Animals Fed The Different Experimental Diets1
Diet groupsMean food intake (g/day)Mean calorie intake (kJ/day)Mean body weight (g) on experimental diets at
Week 0, initial weightWeek 4Week 8, final weight
  • 1

    Values are means ± SEM of measures performed on 15 rats, Student's t-test or ANOVA.

  • For each column, different letters (a, b) indicate statistical differences between values obtained in each dietary group (p < 0.05).

Control diets
 Standard diet23.84 ± 2.05a445.33 ± 38.29a196.33 ± 3.65a320.50 ± 15.44a390.50 ± 23.17a
 Standard diet + ATRA22.82 ± 0.88a446.28 ± 16.44a193.67 ± 3.80a314.92 ± 8.60a392.64 ± 10.13a
High-fat diets
 HLD32.11 ± 2.58b943.71 ± 75.83b197.00 ± 6.99a331.13 ± 15.59a421.38 ± 20.10b
 HLD + ATRA30.69 ± 2.30b901.98 ± 67.60b196.00 ± 5.42a321.20 ± 10.08a421.43 ± 13.13b
 VAHLD34.94 ± 1.94b1,026.89 ± 57.02b200.70 ± 2.67a327.60 ± 9.48a423.28 ± 23.22b
 VAHLD + ATRA32.29 ± 2.29b949.00 ± 67.30b199.00 ± 2.30a316.67 ± 7.51a413.53 ± 11.66b

On the basis of food consumption and body weight gain, the present data show that the weight gain of rats fed the high-fat diets was due to the fact that they ate more food rich in energetic components.

Quantification of ACs and ACF

To limit potential disturbances of molecular events by the chemical inducer, animals were treated with a single low DMH dose (15 mg/kg body weight) sufficient to initiate abnormal proliferation.25 ACF were numerated in 9 rats per group. The data are reported in Table III. ACF were present throughout the length of the colons except at the level of Peyer's patches. Rats fed HLD showed a significantly greater number of total ACF/colon compared to those fed standard diet (191.56 ± 88.60 vs. 21.14 ± 11.64, p < 0.05). Colon mucosa from the VAHLD group presented an intermediate number of total ACF (99 ± 28.43), both significantly different from the standard and HLD groups (p < 0.05). The occurrence of ACF induced by DMH in rats fed the standard diet was unchanged by ATRA administration. No significant effect was observed in ATRA-treated HLD-fed rats in spite of a decrease in the mean number of total ACF (167.50) and ACF with 1, 2 or 3 ACs compared to the results observed in HLD-fed rats. A significant effect of ATRA administration on ACF occurrence was observed in the VAHLD + ATRA group compared to the VAHLD alone group (46.2 ± 19.41 vs. 99.00 ± 28.43, p < 0.05). Under these dietary conditions (VAHLD + ATRA), the occurrence of ACF with 1–3 ACs appeared comparable to that observed in controls.

Table III. Effect of Experimental Diets on The Occurrence of Preneoplastic Lesions in The Colon of Rats Administrated DMH
Diet groupsACF/colonAC/ACF
1 AC2 AC3 AC or more
  1. Rats were fed the diets for 4 weeks, injected once with a low dose of DMH (15 mg/kg body weight) and then fed the diets for an additional 4 weeks. Rats received intragastric administration of ATRA (1 mg/kg body weight) or vehicle once every 2 days during the experimental period. Values are means ± SEM of measures performed on 9 rats, Student's t-test or ANOVA. For each column, different letters (a–d) indicate statistical differences between values obtained in each dietary group (p < 0.05).

Control diets
 Standard diet21.14 ± 11.64a12.86 ± 8.19a6.57 ± 4.23a1.71 ± 2.25a
 Standard diet + ATRA22.89 ± 24.65a,d16.22 ± 20.02a5.33 ± 6.29a1.33 ± 2.03a
High-fat diets
 HLD191.56 ± 88.60b109.44 ± 56.36b58.30 ± 32.67b17.33 ± 8.76b
 HLD + ATRA167.50 ± 128.30b,c121.83 ± 102.13b37.50 ± 28.72b8.00 ± 10.28a,b,c
 VAHLD99.00 ± 28.43c65.25 ± 21.20b26.50 ± 9.19b6.50 ± 4.07c,d
 VAHLD + ATRA46.2 ± 19.41d28.80 ± 13.29a12.80 ± 6.98a4.10 ± 4.01a,c

mRNA expression of nuclear receptors

We then investigated whether the modulation of ACF occurrence observed after a 2-month period of high-fat diet was associated with modifications in the expression of peroxisome proliferator and retinoid receptors. Relative quantification of the mRNA expression of these nuclear receptors was performed on the colonic epithelium of 9 rats per group (Table IV). In contrast to the effects observed on AC and ACF occurrence, administration of ATRA did not modify the mRNA expression of nuclear receptors induced by each kind of diet. Expression of PPARδ mRNA was not modified by the different diets, while downregulation of PPARγ was induced by HLD. Surprisingly, VAHLD, with an identical composition in fatty acids as HLD, provoked the opposite effect, with overexpression of PPARγ mRNA compared to the standard diet (p < 0.05). High-fat diet in DMH-treated rats had no significant effect on the mRNA expression of RARα or on that of the common partner RXRα. Rats fed HLD showed a significantly decreased level of RARβ mRNA expression compared to those fed the standard diet (p < 0.05).

Table IV. Effect of Experimental Diets on Nuclear Receptor mRNA Expression
Diet groupsmRNA relative quantification1
  • Values are means ± SEM of measures performed on 9 rats, Student's t-test or ANOVA.

  • 1

    Target mRNA are expressed in percent of β2-microglobulin mRNA, except for RARβ expressed first in percent of GAPDH mRNA and then submitted to normalization with a calibrator. For each column, different letters (a–c) indicate statistical differences between values obtained in each dietary group (p < 0.05).

Control diets
 Standard diet0.44 ± 0.10a1.69 ± 0.36a1.32 ± 0.28a0.32 ± 0.21a4.63 ± 0.93a
 Standard diet + ATRA0.42 ± 0.07a1.51 ± 0.30a1.42 ± 0.32a0.29 ± 0.34a4.59 ± 0.43a
High-fat diets
 HLD0.48 ± 0.09a1.13 ± 0.24b1.10 ± 0.21a0.15 ± 0.03b4.81 ± 0.58a
 HLD + ATRA0.48 ± 0.08a1.10 ± 0.16b1.55 ± 0.34a0.17 ± 0.03b5.28 ± 0.81a
 VAHLD0.44 ± 0.12a2.58 ± 0.37c1.45 ± 0.72a0.66 ± 0.37a5.18 ± 1.15a
 VAHLD + ATRA0.48 ± 0.06a2.13 ± 0.55c1.15 ± 0.29a0.64 ± 0.35a4.46 ± 0.52a

Expression of COX-2 and β-catenin

In a large majority of colon cancers, the tumorigenic process is characterized by upregulation of COX-2 and loss of β-catenin degradation, leading to its accumulation. Whether the different diets could modify the levels of COX-2 and β-catenin proteins in the colonic mucosa was thus investigated.

Compared to the control group, the amount of COX-2 mRNA transcript and protein was increased in the HLD group (p < 0.05) both with and without ATRA (Table V). A slight, but not statistically significant, increase in the COX-2 mRNA level was also detected in the VAHLD group. However, the amount of COX-2 protein was comparable to that of controls and significantly lower than in the HLD group (p < 0.05).

Table V. Effect of Experimental Diets on COX-2 Expression
Diet groupsCOX-2 expression
  • 1

    Target mRNA are expressed in percent of β2-microglobulin mRNA.

  • 2

    Data for protein levels were obtained from autoradiographic film analyzed by Bio ID software, which provides data representing the sum of the intensities of the pixels inside the volume (signal revealed by the specific antibody) × area of a single pixel (in mm2).

  • Values are means ± SEM of measures performed on 6 rats, Student's t-test or ANOVA. For each column, different letters (a, b) indicate statistical differences between values obtained in each dietary group(p < 0.05).

Control diets
 Standard diet1.03 ± 0.26a7,695.05 ± 1,843.96a
 Standard diet + ATRA1.15 ± 0.16a7,743.50 ± 1,423.70a
High-fat diets
 HLD1.55 ± 0.26b11,194.50 ± 2,190.85b
 HLD + ATRA1.69 ± 0.52b11,790.10 ± 5,664.63b
 VAHLD1.31 ± 0.48a,b7,847.34 ± 1,610.78a
 VAHLD + ATRA1.28 ± 0.12a,b6,782.30 ± 3,941.40a

HLD affected the level of total β-catenin (Fig. 3a). Rats fed HLD exhibited the highest level of β-catenin, whereas no significant modifications were found in the HLD + ATRA and the VAHLD groups (p < 0.05). In contrast, rats fed VAHLD + ATRA had the lowest level of β-catenin. Immunohistochemical analysis on 7 μm sections of colon did not reveal any modification in β-catenin localization in any of the diet groups. In each dietary condition, β-catenin was clearly detectable along the basolateral membrane and in the cytoplasm of epithelial cells facing the lumen (Fig. 3b).

Figure 3.

β-Catenin levels and localization. (a) Effect of high-fat diet administration and ATRA supplementation on total β-catenin level in colon mucosa. Data represent means of measures performed on 6 animals, with SE represented by vertical bars. Different letters (a–c) indicate statistical differences between values obtained in each dietary group (p < 0.05). (b) Qualitative data on β-catenin localization obtained by immunohistochemistry. Intense staining of cells facing the lumen is observed (b1, original magnification ×400) whatever the diet, with or without ATRA. Counterstaining with hematoxylin reveals no nuclear localization of β-catenin (b2, original magnification ×200).


Analysis of the relationship between lifestyle and disease occurrence has enjoyed increasing interest in recent decades. The idea that the risk of developing several malignant disorders is related to dietary fat intake and overweight now emerges as a general consensus. Nutritional factors are designated as the causative agents and estimated to account for approximately one-third of cancers in Western countries.26 Since overweight and obesity participate in enhancing the risk of colon cancer, not only the type but also the amount of food that is consumed must be taken into account in health-maintenance programs. The present study is based on the use of high-fat diets containing various sources of fat commonly consumed in Western countries. These diets contained 60% of lipids, of which a large majority were saturated and possessed a palatable quality, leading to overconsumption and subsequent weight gain.21 Epidemiologic and experimental data suggest that energy in the form of fat and carbohydrate has a stronger influence on the risk of cancer than protein.27 However, when animals were fed high-fat diets, they also ate less protein and carbohydrate than those fed the standard diet. Levels and sources of protein in the diet could have a significant influence on tumorigenesis.28, 29, 30 Carbohydrates as well as proteins may induce a broad range of effects,31, 32 including modulation of cancer risk.33 Although potential effects of proteins and carbohydrates on the occurrence of ACF and associated molecular disturbances cannot be ruled out, energy intake provided by fat appears to be the prevalent factor in the promotion of intestinal cancer. Overfeeding the high-fat diets (HLD and VAHLD) was associated with high energy intake and high fat intake, both of which increase the risk of colon cancer development.34 As expected, early exposure to the high-fat diets induced both weight gain and an increase in the number of DMH-induced colonic preneoplastic lesions (ACF). Previous studies have suggested that there is a high degree of correlation between the number of multicrypt ACF and the outcome of colonic tumors at later stages.35 The present study was conducted to analyze early molecular alterations associated with high energy intake/weight gain in the earliest steps of carcinogenesis.

Colon carcinogenesis is a multistep process involving both progressive loss of growth control mechanisms and accumulation of mutations resulting in an increased level of neoplasia. The multistage process has been described as a “progressive disorder in signal transduction”.36 According to this model, progressive nongenetic disruptions in homeostatic mechanisms controlling proliferation, differentiation and apoptosis can increase the occurrence of cancer.

A number of studies on the colon tumor–promoting effects of dietary fat have focused on PPARγ, first, because this receptor has a central role in nutrient-controlled gene regulation and, second, because of its involvement in colon carcinogenesis. The majority of reports concerning PPARγ are gain-of-function experiments. Colon tumor cell lines respond to PPARγ agonists by reducing growth rate and increasing differentiation or apoptosis.37, 38, 39, 40 Our results show that HLD downregulates the level of PPARγ mRNA. Underexpression of PPARγ in rodent and human mammary gland carcinomas41, 42 and in human prostate adenocarcinomas have been reported previously.43 Moreover, Sarraf et al.44 detected somatic loss-of-function mutations in the gene encoding PPARγ in a few cases of sporadic colorectal carcinoma, suggesting that the wild-type gene compromises the survival of abnormal cells. The mechanism by which HLD downregulates PPARγ mRNA remains unknown. In addition to downregulation of PPARγ mRNA, we found that HLD produced an increase in the number of ACF. This observation is consistent with a clinical report that reduced levels of PPARγ mRNA might be correlated with increased prevalence of colonic polyps.45 In contrast to PPARγ, the δ isoform appears to be involved in tumor formation rather than in ACF occurrence.46 The PPARδ isoform, expressed aberrantly during the development of colorectal cancer, was not affected by HLD in our experiment, though its mRNA levels were reported to be regulated in response to nutritional state in other tissues.47 PPARδ is a direct transcriptional target of β-catenin, and β-catenin levels were increased by HLD. It might thus be surprising that we did not detect an increase in the level of PPARδ mRNA. However, even if the level of total β-catenin increased in the mucosa of rats fed HLD, immunohistochemical analyses did not reveal any abnormal nuclear localization.

Concomitant with the decrease in PPARγ mRNA level, we found that HLD negatively affected the expression of RARβ. Vitamin A is involved in the maintenance of epithelial homeostasis, mainly through its interaction with RARs and RXRs. Loss of vitamin A sensitivity is a common feature of human cancer cells and appears to be a consequence of the silencing of the RARβ gene.48 Various mechanisms have been evoked to explain how dietary fat modifies RARβ expression. High-fat diets may modify cell membrane properties49 and, thus, alter retinol transport across colonocyte membranes. Alternatively, HLD might increase retinol esterification or modulate the rate of retinol metabolism.50, 51 Autoregulation of RARβ gene expression plays an important role in the response of colon cancer cells to RA. Thus, one hypothesis is that HLD leads to a decrease in the cellular concentration of active RAs. A decrease in RA availability should then affect RARβ expression. However, why in our experiment HLD specifically induced downregulation of RARβ without modifying the levels of RARα and RXRα mRNA remains to be elucidated. Our results do not exclude the possibility of alterations of RARα and RXRα at the protein level.

Several reports have indicated that COX-2 is involved in colon carcinogenesis since this protein is overexpressed in human tumors.52, 53 Fatty acids and derivatives are involved in the regulation of COX-2, which appears to be one of the main targets of these nutrients.54 In the present study, feeding rats HLD induced an increase in the level of COX-2 mRNA and protein concomitant with underexpression of both nuclear receptors, PPARγ and RARβ. Interestingly, associations between nuclear receptors and COX-2 regulation are supported by data from the literature. Inhibition of COX-2 expression is one of the mechanisms by which the PPARγ pathway induces apoptosis and inhibits inflammation.15 Badawi and Badr42 also demonstrated a correlation between downregulation of the PPARγ mRNA and the increase of COX-2 in human breast cancer, suggesting that the 2 events are closely related. Other reports also indicate that RARβ inhibits COX-2, to mediate a protective action against carcinogenesis.17, 55 Thus, underexpression of RARβ and PPARγ could participate in the effect of fatty acids on DMH-induced COX-2 overexpression.

The loss of β-catenin degradation is another event occurring early in colon tumorigenesis.56 Even though our results showed an increase in the level of β-catenin in rats fed HLD, there was no change in β-catenin localization compared to rats fed the standard diet. The present results may describe an event prior to β-catenin translocation toward the nucleus. Vitamin A nuclear receptors are implicated in the targeting of β-catenin for degradation.18 Thus, we can postulate that alterations in nuclear receptor expression patterns could modify the level of β-catenin. Therefore, the observed β-catenin accumulation might be a consequence of RARβ downregulation. These results might be relevant to the role of nuclear receptors in the risk of colon cancer.

Vitamin A and its active form, ATRA, are used with promising results in the treatment of many tumors.9 Because overfeeding HLD affected early RARβ expression, another high energy/fat diet containing a higher level of vitamin A (mainly under the form of retinyl esters) was used, to determine whether vitamin A could prevent or reverse the HLD-induced modifications in gene expression. As expected, the level of RARβ mRNA was increased when vitamin A was included. Surprisingly, the level of PPARγ mRNA in rats fed VAHLD also appeared significantly increased compared to that in rats fed HLD. Because this pattern of expression was associated with a reduction in ACF occurrence, our results support the idea that upregulation of both RARβ and PPARγ is associated with an antiproliferative effect The partial prevention of ACF occurrence by VAHLD also revealed the capability of vitamin A to thwart or reverse the effects of high energy/fat intake by restoring or maintaining levels of COX-2 and β-catenin comparable with those obtained with the standard diet. Interestingly, these results demonstrate that the alterations in gene expression pattern induced by fat/energy intake can be, at least in part, antagonized by vitamin A.

The present results also demonstrate that RARβ underexpression is a crucial parameter in disorders induced by HLD. Diet-derived ATRA is the main signaling retinoid in the body, and oral administration of ATRA restores expression of certain nuclear retinoid receptors altered during premalignant or malignant lesion development.57 ATRA administration is useful in treating acute promyelocytic leukemia and dermatologic diseases and is promising in new therapies against cancer.9 Vitamin A under its active form is efficiently taken up by gastrointestinal cells and subsequently delivered to target cells. ATRA is not submitted to hepatic immobilization since in vivo it is synthesized from the vitamin A storage form in an irreversible way. Until now, very few in vivo experiments have analyzed the chemopreventive action of ATRA in colon tumorigenesis. Oral administration of ATRA, performed in each of our dietary groups, tended to reduce ACF occurrence in rats fed HLD or VAHLD. However, ATRA did not prevent the diet-induced alterations of the different molecular biomarkers that we analyzed. The dose of ATRA that we used appears insufficient to modify the expression level of its own receptor, RARβ. Only β-atenin levels appeared to be affected by the presence of ATRA. The mechanism by which ATRA affects β-catenin levels is unknown, but we cannot exclude that the regulation of β-catenin levels relies on the activation of RARβ by ATRA even in the absence of induction of RARβ expression. Our results reveal that maintenance of low levels of β-catenin could participate in the prevention of ACF occurrence. Additional studies are needed to further document the role of ATRA in the prevention of HLD-induced disturbances. There are, however, limitations in the use of ATRA since this compound could be toxic, as observed in acute promyelocytic leukemia therapy.

In conclusion, our study demonstrates that excessive fat/energy intake inhibits RARβ and PPARγ expression at an early stage and participates in DMH-induced ACF occurrence by increasing the level of COX-2 and β-catenin. These molecular alterations might correspond to some of the early events in the modification of the cell program that facilitates the onset of cancer. Although the mechanisms by which fat/energy intake promote ACF occurrence remain to be elucidated, downregulation of RARβ justifies the interest in the mode of action of vitamin A. This nutrient, probably in part through RARβ upregulation, might prevent HLD-induced alterations of COX-2 and β-catenin levels. A detailed understanding of nutrient-induced signaling mechanisms and their consequences in colon cancer and in normal colon physiology will certainly contribute to the development of effective therapies.


We thank Mr. L. Caune for animal care and Ms. K. Mayo and Mr. S.J. Saupe for correcting the English.