Obesity promotes PhIP-induced small intestinal carcinogenesis in hCYP1A-db/db mice: involvement of mutations and DNA hypermethylation of Apc

Summary

Dietary carcinogen 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) induces small intestinal tumors in obese mice. Most of tumors are caused by the loss of functional APC, resulted from a PhIP-induced truncation mutation in one Apc allele and an obesity-associated DNA hypermethylation-induced silence of the other allele.

Abstract

Obesity is associated with an increased risk of cancer. To study the promotion of dietary carcinogen-induced gastrointestinal cancer by obesity, we employed 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) to induce intestinal tumorigenesis in CYP1A-humanized (hCYP1A) mice, in which mouse Cyp1a1/1a2 was replaced with human CYP1A1/1A2. Obesity was introduced in hCYP1A mice by breeding with Lepr^db/+ mice to establish the genetically induced obese hCYP1A-Lepr^db/db mice or by feeding hCYP1A mice a high-fat diet. PhIP induced the formation of small intestinal tumors at the ages of weeks 28–40 in obese hCYP1A mice, but not in lean hCYP1A mice. No tumors were found in colon and other gastrointestinal organs in the lean or obese mice. Using immunohistochemistry (IHC), we found strong positive staining of NF-κB p65, pSTAT3 and COX2 as well as elevated levels of nuclear β-catenin (Ctnnb1) in small intestinal tumors, but not in normal tissues. By sequencing Apc and Ctnnb1 genes, we found that most PhIP-induced small intestinal tumors in obese mice carried only a single heterozygous mutation in Apc. By bisulfite-sequencing of CpG islands of Apc, we found DNA hypermethylation in a CpG cluster located in its transcription initiation site, which most likely caused the inactivation of the wild-type Apc allele. Our findings demonstrate that PhIP-induced small intestinal carcinogenesis in hCYP1A-db/db mice is promoted by obesity and involves Apc mutation and inactivation by DNA hypermethylation. This experimental result is consistent with the association of obesity and the increased incidence of small intestinal cancer in humans in recent decades.

Introduction

In the digestive tract, cancer is common in the colon and the rectum but rare in small intestine. New cases of small intestine cancer (SIC) in the USA are estimated to compose only ~3% of all digestive system cancers in 2015 (1). However, there is a growing concern for SIC because the incidence of SIC has been increasing steadily in the last 30 years (2–4). In the USA, the number of new SIC cases in 2015 is estimated to be of ~9410, a huge increase from 2200 cases in 1986 (1,5). In contrast, the new cases of colon and rectal cancer (CRC) have decreased in the last 10 years, in part due to the successful application of screening colonoscopy and polypectomy (6). Because there is no effective approach to examine the small intestine for preventive screening, the diagnosis of SIC mainly depends on symptoms such as abdominal pain, nausea, vomiting and weight loss. This results in the delay of diagnosis and a lower survival rate. When first diagnosed, ~88% of SIC are invasive and the 5 year survival rate for patients with invasive adenocarcinoma is ~28% in the USA (2,3). Therefore, it is important to elucidate the risk factors and the molecular mechanisms of tumor formation and progression. Knowledge in the causes of the increase of SIC would help develop means for the prevention and earlier detection of SIC.

Several factors are recognized to increase the risk of SIC from retrospective epidemiologic investigations in multiple countries (7,8). First, SIC is diagnosed more frequently in the individuals with predisposing genetic factors and medical conditions, including familial adenomatous polyposis, hereditary non-polyposis colorectal cancer, Peutz–Jeghers syndromes, inflammatory bowel diseases and celiac disease (reviewed in refs. 7,8). Since these conditions are also well-recognized risk factors of CRC, it suggests that SIC shares many common molecular mechanisms of CRC formation (9). Second, SIC incidence increases with age (2,3). Third, overconsumption of animal fat, red meat and salted or smoked meat or fish increases the risk of SIC (8,10). Data from several studies imply that obesity contributes to the increase of SIC incidence (11–15). For example, in a cohort study on cancer incidence in Sweden, SIC incidence was found to increase with obesity (relative risk = 1.4–2.8; 95% confidence interval = 1.6–4.5) (11). A similar risk increase of SIC was observed in overweight and obese individuals in other studies in Europe (12,13), Asia (14) and the USA (15). Indeed, the increase of SIC in the USA in the past 30 years is correlated with the dramatic increase of obesity since the 1980s (16–18). Obesity is the fastest growing disease worldwide in the last few decades (19). It is well recognized that obesity increases the risk and mortality of cancer in several organs and ~20% of all cancers can be attributed to obesity (20,21). Several obesity-associated conditions have been demonstrated to be positively associated with cancer in the gastrointestinal tract. These include chronic inflammation, changes in adipokine profile, increase of insulin and insulin-like growth factor (IGF) signaling and changes in gut microbiota (22–27).

The promotion of colon carcinogenesis by obesity has been demonstrated in the genetically obese male Zucker rats (28), genetically obese [loss of leptin signaling: Lepr^db/db (db/db) or Lept^ob/ob (ob/ob)] mice (29,30) and high-fat (HF) diet-induced obese mice (31). In the Apc ^min/+ mice, which develop predominantly small intestinal tumors, obesity generated by crossing with db/db or ob/ob or by feeding a HF diet promoted not only small intestinal tumorigenesis but also stomach, cecum and colon tumorigenesis (29,30). The suggested mechanisms for the promotion of tumorigenesis by obesity include activation of tumor necrosis factor-α signaling (32) and elevation of IGFs and IGFR signaling (30) depending on the different experimental models used.

A recent study showed that colonic epithelial tissues in the HF-induced obese mice display a set of epigenomic markers distinct from those of lean control mice; the difference accompanies the promotion of cell proliferation and survival (33). This finding suggests that obesity alters epigenome to prime normal cells for carcinogenesis or to introduce additional events to earlier lesioned cells for further progression. Indeed, the epigenetic regulation is an important mechanism in human CRC development, and almost all major Wnt signaling components can be modulated through epigenetic mechanisms (34). For example, APC promoter 1A hypermethylation was found in 18% of primary sporadic CRCs (35). Since promoter hypermethylation leads to the silence of APC, the hypermethylation of APC promoter 1A acts as a second-hit mechanism to inactivate the wild-type (wt) allele in CRCs only carrying one APC mutation. Such an epigenetic modification provides an alternative mechanism of APC inactivation. However, what causes the hypermethylation of APC promoter 1A remains unknown. It is possible that the obesity-associated conditions can lead to the hypermethylation of Apc, but this needs to be further investigated.

Heterocyclic amines (HCAs), which are found in high-temperature cooked meat and fish, may play important roles in human SIC and CRC development (36). HCAs are listed in the United States Department of Health and Human Services as a chemical ‘reasonably anticipated to be a human carcinogen’. Human exposure to HCAs varies in a wide range (0.1–12 µg/day) (36). 2-Amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) is the most commonly occurring and abundant HCA (36). In humans, PhIP is metabolized by human cytochrome P4501A2 (CYP1A2) through N-hydroxylation to a proximal carcinogen, whereas the corresponding mouse enzyme Cyp1a2 mainly detoxifies PhIP through 4′-hydroxylation (37). We utilized the CYP1A-humanized (hCYP1A) mice, made by replacing the mCyp1a gene with the hCYP1A gene (38), for studying colon carcinogenesis induced by PhIP (39,40). Since the PhIP-induced colon carcinogenesis requires the treatment with dextrin sulfate sodium (DSS), it suggests that an additional event, such as inflammation or damage of tissue hierarchical structure, plays a critical role in promoting neoplastic progression of the cells carrying the PhIP-induced mutations. Therefore, this mouse model offers an opportunity to determine whether additional factors, such as obesity, could promote the PhIP-induced colon and small intestinal carcinogenesis. In this study, we used both genetic (db/db)- and HF-induced obese hCYP1A mice for this purpose. We found that non-obese hCYP1A mice treated with PhIP alone did not develop tumors in the digestive tract. Interestingly, in obese mice, PhIP induced tumorigenesis in the small intestine, but not in the colon. The intestinal tumors are featured with the activation of Wnt signaling and increased inflammation. Further characterization revealed that these tumors lost functional Apc, resulted from a heterozygous truncation mutation in one allele and DNA hypermethylation-associated gene silence of the other allele.

Materials and methods

Chemical

PhIP–HCl (catalogue no. 163-15951; purity >99.0% by high-performance liquid chromatography) was purchased from Wake Chemicals USA (Richmond, VA) and dissolved in milliQ water as described previously (40).

Animals and procedures

hCYP1A mice, carrying human CYP1A gene to replace mouse Cyp1A gene, were obtained from Jackson Laboratories (Bar Harbor, ME) and bred as described previously (39). Heterozygous db/+ mice, BKS.Cg-Dock7^m +/+ Lepr^db/J [congenic C57BLKS/J (BKS) pure background] mice, were purchased from Jackson Laboratories and used as founders to breed with hCYP1A mice to generate hCYP1A(+/+)-mCyp1A(−/−)-Lepr^db/wt (hCYP1A-db/+) mice after being crossbred for three generations. Genotyping for hCYP1A and mCyp1a was described previously (39). The Lepr^db allele genotyping was carried out as the protocol provided by Jackson Laboratories. The established Lepr^db heterozygous (hCYP1A-db/+) mice were used as the breeders to produce the Lepr^db homozygous (hCYP1A-db/db) mice for the study. All animal procedures were in accordance with the animal study protocol (nos. 02-027 and 11-004) approved by the Rutgers University Institutional Animal Care and Use Committee. At the age of week 5, the hCYP1A and hCYP1A-db/db mice were switched from a chew diet to the AIN93M diet or 60 kcal % HF diet (Research Diets, New Brunswick, NJ) (41). Body weights were measured weekly. At the age of week 6, mice were treated with two doses of PhIP (100mg/kg body weight 4 day apart) by oral gavage. Mice were continuously maintained on the AIN93M diet or 60 kcal % HF diet until the termination of the experiment (28–40 weeks after the PhIP treatment). Tumors in the digestive tract can cause body weight loss and bloody feces. Once any of these symptoms were observed in a mouse, this mouse was killed for examination. All mice were killed by CO₂ asphyxiation. Digestive tracts including stomach, small intestine, cecum, colon and rectum were then excised and flushed with saline water. The intestine was cut longitudinally and flattened on a piece of filter paper for examination under dissecting microscope. The number, size and location of visible tumors were recorded. For the mice bearing multiple tumors, one tumor including adjacent intestinal tissues for each mouse was formalin fixed for histopathological analysis (41), whereas the other tumor was frozen in O.C.T. medium at −80°C for cryosection. Tumors from some mice were stored in AllProtect Reagent (Qiagen, Valencia, CA) at −80°C for DNA/RNA extraction. Mucosal layers in adjacent normal intestine were collected in AllProtect Reagent by scraping and used for extracting control DNA/RNA samples.

Histopathological analysis

After fixation in 10% formalin (neutral buffered), mouse intestines were rolled and embedded in paraffin for preparation of serially 4 µm thick sections for H&E staining (41). Histopathological analyses were performed on multiple sections of a sample. The histopathology of intestinal tumors was characterized according to the well-recognized standards (42).

Immunohistochemistry

IHC was performed on 4 μm thick paraffin-embedded sections after antigen retrieval as described previously (41). The antibodies used were anti-p65 (Abcam, Cambridge, MA), anti-pSTAT3 (Cell Signaling Technology, Beverly, MA), anti-COX2 (Abcam, Cambridge, MA) and anti-β-catenin (Santa Cruz Biotechnology, Dallas, TX). The IHC staining was carried out using biotinylated secondary antibody (1:200), followed by streptavidin-biotin peroxidase conjugate (Vector Laboratories, Burlingame, CA), and developed with 3,3′-diaminobenzidine substrate (Vector Laboratories) and counterstained with hematoxylin eosin. For negative controls, the incubation with primary antibody was omitted.

DNA extraction, PCR and DNA sequencing

To extract genomic DNA and total RNA from intestinal tissues and tumors, the samples stored in AllProtect Reagent were homogenized in the lysis buffer of AllPrep DNA/RNA Mini Kit (Qiagen) using Omni Bead Raptor 24 (Omni International, Kennesaw, GA). DNA and RNA were then purified according to the manufacturer’s protocol. The qualities of DNA and RNA were examined by gel electrophoresis.

Fragments covering the N-terminal half of Apc (~6kb including the mutation-cluster region) and the exon 3 of Ctnnb1 were PCR amplified using high fidelity Advantage 2 PCR kit (Clontech) with genomic DNAs as the templates. The PCR primers were described previously (40). The amplified products were separated by electrophoresis and then purified using the QIAquick Gel Extraction Kit (Qiagen). The purified products were sequenced from both directions with PCR primers by Genewiz (South Plainfield, NJ). To validate the identified mutation, additional sequencing experiment was performed using the PCR product targeting the same region with primers, 5′-GGTCCTTCCAGACGTGGATA-3′ (forward primer, at the mRNA coding position 4508) and 5′-GAGGCTTACTGGGCTCTCCT-3′ (reverse primer, at the mRNA coding position 6075).

DNA bisulfite conversion and sequencing

Genomic DNA bisulfite conversion was carried out using EpiTect Plus DNA Bisulfite kit (Qiagen) with 2 µg genomic DNA extracted from intestine tumors and control tissues. A cluster consisting of 95 CpG islands in 763bp region was identified in the promoter by Genome Browser (http://genome.ucsc.edu/) according to mouse genome mmp10 (Supplementary Figure 4, available at Carcinogenesis Online). After optimization, this region (except the first two and last four CpG islands) was covered by two PCR-amplified fragments 53–35 (446bp; amplified by primers 5′-TATCGTAGAGGTAGGGTATAGGTTGTTG-3′ and 5′-CGCCGCCCGAACGCCTCCCCTCC-3′) and 54–33 (441bp; amplified by primers, 5′-GTTGAGGAAGGTGGAGTGAGGAGTGGTTTT-3′ and 5′-ACAAAAAACTACTAATAAAAACGAAACTACCTA-3′), as illustrated in Supplementary Figure 4, available at Carcinogenesis Online, using high fidelity Advantage 2 PCR kit with the bisulfite-converted genomic DNA products as the templates. These PCR primers were designed based on the bisulfite-converted DNA (http://www.zymoresearch.com/tools/bisulfite-primer-seeker) and synthesized by Integrated DNA Technologies (IDT; Coralville, IA). The amplified products were separated by electrophoresis and then purified using the QIAquick Gel Extraction Kit for sequencing. The unmethylated C was converted to T and the methylated C remained to be C in the sequencing results. The C peak accounting >10% in the same position was considered as the methylated C as described (43).

Statistical analysis

One-way analysis of variance followed by Dunnett’s test was used for comparison between treatment groups and the control groups. Student’s t-test was used to determine the difference between two groups. For comparisons of lesions’ incidence, Fisher’s exact test was used. Differences were considered statistically significant when P < 0.05 in two-tailed comparisons.

Results

PhIP induces intestinal carcinogenesis in obese mice

To determine the effect of obesity on PhIP-induced intestinal carcinogenesis, we conducted an experiment consisting of four groups of mice: (i) lean hCYP1A mice (on AIN93M diet) treated with PhIP (200mg/kg body weight), (ii) lean hCYP1A mice (on AIN93M diet) treated with PhIP (400mg/kg); (iii) HF-induced obese hCYP1A mice (on 60% kcal HF diet) treated with PhIP (200mg/kg) and (iv) genetically obese hCYP1A-db/db mice (on AIN93M diet) treated with PhIP (200mg/kg). PhIP was administrated at the age of week 6 when there was no significant difference in body weight between the lean and obese mice. The growth curves of the mice are shown in Supplementary Figure 1, available at Carcinogenesis Online. We observed fecal blood and body weight loss in some mice in the two obese groups, as early as week 27 after the PhIP treatment, presumably due to tumor developed in the intestine. Once either of fecal blood or body weight loss was observed, the mice were killed for analysis. The remaining mice in the two obese groups were killed on week 44 after the PhIP treatment, whereas some lean mice were maintained up to 80 weeks. Upon examining the gastrointestinal tract under a dissecting microscope, we found 33.3% of the HF-induced obese hCYP1A mice and 95% of genetically obese hCYP1A-db/db mice developed tumors in the small intestine (Figure 1 and Table 1). The majority of tumors were located in the duodenum or duodenum–jejunum junction and most tumors were 1–3mm in diameter (Figure 2A and B). No tumors were found in the stomach, colon and rectum. The lean hCYP1A mice treated with PhIP (200 or 400mg/kg) did not have any tumors in the gastrointestinal tract at 40–80 weeks after the PhIP treatment; these lean mice grew normally and had no fecal blood or body weight loss during the experimental period. In the control obese mice (without PhIP treatment), no tumors were found (Table I). However, prostate lesions were found in these older mice as we reported previously (41). Compared with the lean control groups, the HF and db/db groups both had higher frequencies of developing tumors in the small intestine (P = 0.001 and P < 0.001, respectively). This result demonstrated that obesity, induced by either genetic factors or diet, promoted small intestinal tumorigenesis in hCYP1A mice.

Figure 1.

The number of tumor in the lean hCYP1A, HF-induced obese hCYP1A and genetically obese hCYP1A-db/db mice treated with PhIP.

Table 1.

Tumor incidence in small intestine in lean and obese mice treated with PhIP

Mouse genotype	PhIP dose (mg/kg body weight)	Diet	Number	Duration (weeks)	Average total tumor numbers	Tumor incidence (%)
hCYP1A	200	AIN93M	37	45–70	0	0.0
	400	AIN93M	10	40–80	0	0.0
	0	HF (60% kcal)	14	40–45	0	0.0
	200	HF (60% kcal)	15	27–44	0.5±0.7	33.3
hCYP1A-db/db	0	AIN93M	6	35	0	0.0
	200	AIN93M	20	27–44	2.3±1.3	95.0

Figure 2.

The PhIP-induced intestinal tumors in hCYP1A-db/db obese mice. (A) PhIP-induced tumorigenesis occurred in the duodenum near stomach in hCYP1A-db/db mice. (B) There were multiple tumors with a size ranging from ~1 to 3mm diameter in duodenum. (C) A section prepared using a roll of small intestine consisting tumor was stained with H&E, and the tumor was featured with serrated crypts (×20; the bar represents 500 µm). (D) Crypts were consisted of cell with stratified pencil-like nuclei (×400; the bar represents 100 µm).

The tumors induced by PhIP in hCYP1A-db/db mice were characterized as tubular adenomas, showing tubular architecture featured with stratified overlapping ‘pencil-like’ nuclei (Figure 2C and D). Some areas displayed features of high-grade dysplasia. In the HF-induced hCYP1A mice, the tumors also display these tubular adenoma features (Supplementary Figure 2, available at Carcinogenesis Online). In either model, invasive carcinoma was not observed at the time of killing. These tumors will probably progress to adenocarcinoma.

The PhIP-induced intestinal tumors in obese mice are featured with inflammation

Proinflammatory cytokines, such as tumor necrosis factor-α and interleukin-6, were found previously to be increased in the blood plasma of HF- and genetic (db/db and ob/ob)-induced obese C57BL6 mice by us and others (44,45). Insulin and IGFs have also been reported to be increased in obese Apc ^min/+ mice (30). To assess the roles of these factors in intestinal tumorigenesis, we examined the levels of pAKT, NF-κB and pSTAT3, which are the key downstream targets of insulin/IGF signal and inflammatory stimulators in the tumors by IHC staining. We found the increase of nuclear NF-κB p65 and pSTAT3 stainings in tumor cells, but not in normal intestinal epithelial cells (Figure 3A–D). Since nuclear accumulation of p65 and pSTAT3 represents the activation of proinflammatory cytokines, these results suggest that inflammation plays an important role in promoting tumor formation in the PhIP-treated obese mice. We did not find any positive pAKT staining in these tumors (data not shown), suggesting that activation of insulin/IGF signaling (an upstream event of phosphorylation of AKT) is not involved.

Figure 3.

Characterization of PhIP-induced small intestinal tumors in hCYP1A-db/db obese mice by immunohistochemical staining. Normal intestinal crypts (A, C and E) and intestinal tumor (B, D and F) in hCYP1A mice at 38 weeks after PhIP treatment were stained for p65 (A and B), pSTAT3 (C and D) and β-catenin (E and F) (×400; the bar represents 100 µm).

We then used IHC staining to assess the expression levels of COX2 (an inflammatory marker) and found an increased number of COX2 positive cells in tumors, but not in the adjacent normal intestinal tissues or the intestinal tissues of the control mice (Supplementary Figure 3, available at Carcinogenesis Online). The positive staining of COX2 was present in the stromal cells of the tumors, but not in the epithelial cells.

The PhIP-induced intestinal tumors in obese mice carry mutations in Apc

Our previous study found that the PhIP/DSS-induced colon tumorigenesis in hCYP1A mice was driven by a dominant active mutant β-catenin (40). With the expectation that similar type of mutations may occur in small intestinal tumors, we performed IHC staining for β-catenin and found increased nuclear accumulation of β-catenin in small intestinal tumors, but not in the adjacent normal tissues (Figure 3E and F). Next, we sequenced β-catenin and Apc in tumors and normal controls and found mutations in Apc in 11 out of 16 tumors from 6 mice, but did not find β-catenin mutation in any tumors (Table 2). All of these tumors carried a G to T mutation in Apc codon 1635, resulting in a substitution of GAA with a stop codon TAA and the expression of a truncated Apc protein. Since codon 1635 is located in the middle of the ‘SAMP’ repeats, the 20 amino acid residue repeats contain 3 repeats of a Ser-Ala-Met-Pro that constitute the binding site for Axin to form the destructive complex for maintaining the inactive state of Wnt signaling (46). Truncated Apc protein at this position is not expected to form the stable destructive complex with Axin, resulting in the accumulation of β-catenin. The sequencing results showed these tumors also had wt Apc. Such heterozygous mutation suggests that an additional event is required for the loss-of-function of the wt allele and the event is likely associated with obesity.

Table 2.

Apc mutations in the PhIP-induced intestine tumors in hCYP1A-db/db mice

Mouse number (sex)	Total tumor number	DNA ID	Mutations found in Apc
1 (M)	2	Controla	Wt
		Tumor 1-1	1635:stopb
		Tumor 1-2	1635:stop
2 (M)	4	Control	Wt
		Tumor 2-1	1635:stop
		Tumor 2-2	1635:stop
		Tumor 2-3	Wtc
		Tumor 2-4	n.d.d
3 (M)	5	Control	Wt
		Tumor 3-1	1635:stop
		Tumor 3-2	1635:stop
		Tumor 3-3	1635:stop
		Tumor 3-4	Wtc
		Tumor 3-5	1635:stop
4 (M)	1	Control	Wt
		Tumor 4-1	1635:stop
5 (F)	4	Control	Wt
		Tumor 5-1	Wtc
		Tumor 5-2	1635:stop
		Tumor 5-3	1635:stop
6 (F)	1	Control	Wt
		Tumor 6-1	Wt

F, female; M, male; n.d., not determined.

^aControls were extracted from the normal adjacent intestine tissues scraped from epithelium.

^b‘1635:stop’ represents a mutation on codon 1635:GAA (Glu) → TAA (stop).

^cSynonymous mutations were found in these samples.

^dNo data due to insufficient or bad-quality DNA for sequencing.

The PhIP-induced intestinal tumors in hCYP1A-db/db mice are associated with DNA hypermethylation in the CpG cluster located in Apc transcription start region

Since we found only a single heterozygous Apc mutation in tumors, the loss-of-function of the wt allele could be attributed to epigenetic mechanisms, similar to the result from a recent study showing that obesity introduces epigenomic changes (33). In the mouse Apc gene, there is a cluster that consists of 95 CpG islands located in the promoter that could be targeted for DNA hypermethylation according to Genome Browser (http://genome.ucsc.edu/) based on mouse genome mmp10 (Supplementary Figure 4, available at Carcinogenesis Online). To determine whether DNA hypermethylation occurs in the small intestinal tumors, we selected 10 Apc-mutant tumors from 5 mice with matching normal controls for bisulfite conversion followed by PCR to amplify this CpG cluster for sequencing. We found that a motif consisting of 20th–40th CpG in this cluster was methylated at various degrees in all 10 tumors, whereas no methylation was found in the normal controls (Figure 4). All of these CpGs were partially methylated (10–30% methylated C based on the percentage of area of C peak versus the total C and T peaks in the same position); therefore, it is likely that methylation occurs on only one allele. This DNA hypermethylation motif (−4 to +210) located in the transcriptional initiation site is consisted of DNA functional elements identified as the binding sites for RNA polymerase II and transcription factors responsible for the expression of Apc according to the data from the Encyclopedia of DNA Elements (ENCODE) Project (47) (Supplementary Figure 5, available at Carcinogenesis Online). Therefore, DNA hypermethylation in this motif could lead to the silencing of Apc. To assess such a possibility, we performed IHC staining to determine the ‘full-length Apc protein’ level and the ‘total Apc level’ using anti-Apc C-terminus and N-terminus antibodies, respectively. In the five tumor samples with positive nuclear β-catenin staining, we found that ‘the total levels of Apc’ were reduced by ~50% in four tumors, whereas the ‘full-length Apc’ was reduced by >90% in three tumors (Supplementary Figure 6, available at Carcinogenesis Online). This result suggests that the loss of wt Apc protein (by 90%) is resulted from the DNA hypermethylation-induced silencing.

Figure 4.

DNA hypermethylation in the CpG cluster located in Apc transcription initiation region in the PhIP-induced intestinal tumors in hCYP1A-db/db obese mice. DNA samples collected from 10 tumors with Apc mutations were loaded for bisulfite conversion, and the products were then PCR amplified for Apc CpG cluster as described in Method. The PCR products were sequenced, and the identified methylated CpG were marked in solid circle.

Discussion

In this study, we investigated the PhIP-induced intestinal carcinogenesis in both genetic- and HF-induced obese hCYP1A mice and demonstrated that obesity promotes the PhIP-initiated carcinogenesis. This finding provides experimental evidence to support the hypothesis that obesity promotes SIC, which is consistent with the association between obesity and the rapid increase of SIC in the USA in the past 30 years. Our results suggest that PhIP induces Apc mutation on one allele and obesity induces pro-oncogenic conditions which lead to the silence of the wt allele of Apc through DNA hypermethylation. This novel hypothesis provides a molecular mechanism for cancer promotion by obesity.

Mutation in Apc was found in 11 out of 16 tumors from 6 mice, and bisulfate sequencing was then applied to 10 tumors carrying Apc mutation (1 tumor with mutation did not have sufficient DNA for bisulfate conversion). DNA hypermethylation in the CpG cluster located in the transcriptional initiation site of Apc was found in these 10 tumors. Therefore, the molecular mechanisms of these 10 tumors could be attributed to the genetic and epigenetic events in Apc. However, the tumorigenesis mechanisms in the rest of the five tumors remain unknown. Whether they carry mutation or epigenetic modification in the other components of Wnt signaling or in the genes independent of Wnt signaling will be characterized in our future study.

The Apc mutation identified in the tumors was a single-nucleotide substitution on dG, presumably caused by the PhIP-DNA adduct formation on dG. PhIP is metabolized through N-hydroxylation by hCYP1A2 and then conjugated by N-acetyltransferase or sulfotransferase, and the acetoxy or sulfate metabolite is spontaneously converted to arylnitrenium ion (R-NH⁺), which can react with DNA to form adducts at the eight-position carbon of deoxyguanine base, dG-C8-PhIP (36). Our previous study on the PhIP/DSS-induced colon carcinogenesis identified that >93% of colon tumors carry dominant active mutations in β-catenin and all mutations are dG substitutions (40). Although we did not find any mutation on Apc in PhIP/DSS-induced colon tumors in hCYP1A mice (40), the PhIP-induced colon tumors in rats carry mutations on Apc (48). The Apc mutations identified in the small intestinal tumors in this study resulted in a truncation in the middle of the ‘SAMP’ repeats and the expression of truncated protein. Since this domain is the binding site for Axin for forming the destructive complex, these mutations lead to the loss-of-function of Apc and the subsequent activation of Wnt signaling by accumulation of β-catenin.

The loss-of-function of the wt Apc allele is most likely caused by gene silencing induced by DNA hypermethylation, suggesting a critical role of epigenetic mechanisms in small intestinal carcinogenesis. Epigenetic alterations have been demonstrated previously to play important roles in colorectal carcinogenesis (34). For instance, APC promoter 1A hypermethylation was found in 18% of human primary sporadic CRCs (35) and gene silencing caused by hypermethylation of APC promoter 1A was recognized as a second-hit mechanism in CRCs (34). Our result suggests that Apc hypermethylation can be trigged by obesity, in analogous to the obesity-induced DNA methylation in colon epithelial cells (33). In our experiment, PhIP was administrated in two doses (4 days apart) at the age of week 6, and the PhIP should be metabolized and produce mutation on Apc after administration. Except for this genetic mutation, other effects (e.g. oxidative stress) are not expected to last for more than a week; the entire small intestine epithelium is renewed in 4–7 days. In comparison with the rapidly induced colon tumor (in ~6 weeks) by PhIP/DSS (39), the tumor development in small intestine in PhIP-treated obese mice has a longer latency. In PhIP/DSS-induced colon carcinogenesis, a heterozygous dominant active mutation of β-catenin is sufficient to activate Wnt signaling (40). In contrast, a heterozygous loss-of-function mutation of Apc in PhIP-treated obese mice is not sufficient to activate Wnt signaling. An additional event, the hypermethylation in obese mice, is required to inactivate the wt Apc allele. The hypermethylation of Apc to silence the wt Apc allele is a slow process during the progression of obesity, and this may contribute to the long latency. However, DNA hypermethylation is unlikely to specifically target the wt allele. Such an epigenetic modification may take place on either the wt or the mutant allele of Apc. But tumors could only be developed from the cells that completely lost the wt Apc protein resulted from the loss-of-function mutation caused by PhIP and the silence of wt allele by DNA hypermethylation.

Approximately 90% human CRCs are found harboring dominant active mutations in one of the Wnt signaling pathway components; among them, mutations in APC gene account for the majority (~70–80%) and APC is considered a gatekeeper (46). In the mouse model with Apc deficiency such as Apc^min/+, tumors are developed throughout the gastrointestinal tract and the small intestine has the most tumors (49). Our finding that the PhIP-induced small intestinal tumors in obese mice carry loss-of-function mutation of Apc is consistent with the high tumor incidence in small intestine in Apc ^min/+ mice. It is possible that PhIP can induce both Apc and β-catenin mutations in the intestine of the hCYP1A mice, and obesity provides an environment in the small intestine for the cells with Apc mutation to develop into cancer, whereas colitis promotes the cells with β-catenin mutation to develop colon cancer (40).

In summary, our finding in this study provides the first experimental evidence that obesity promotes chemical-induced carcinogenesis in the small intestine. This result is consistent with the association of obesity with the increase of small intestinal cancer incidence in humans in the past 30 years. In more than half of the tumors, we identified not only truncation mutation but also DNA hypermethylation in Apc. The latter suggests that aberrant epigenetic modification plays an important role in PhIP-induced intestinal carcinogenesis in obese mice. However, whether the DNA methylation-induced silencing of Apc in human SIC remains to be clarified in future studies.

Supplementary material

Supplementary Figures 1–6 can be found at http://carcin.oxfordjournals.org/

Funding

National Institutes of Health (RO1CA120915); John L. Colaizzi Chair Endowment Fund; CA72720 and ES05022.

Conflict of Interest Statement: None declared.

Glossary

Abbreviations

CRC

colon and rectal cancer

DSS

dextrin sulfate sodium

HCA

heterocyclic amines

hCYP1A

CYP1A-humanized mice

HF

high fat

IGF

insulin-like growth factor

IHC

immunohistochemistry

PhIP

2-amino-1-methyl-6-phenylimidazo[4,5-b] pyridine

SIC

small intestine cancer

wt

wild type