Dietary Methyl Donor Depletion Protects Against Intestinal Tumorigenesis in Apc^Min/+ Mice

Epidemiologic studies have suggested that low folate levels are associated with an increased risk of colorectal cancer (1–5) and other health risks such as congenital neural tube defects. To reduce the incidence of neural tube defects, the U.S. Food and Drug Administration instituted a mandatory folate fortification program in 1998, leading to near doubling of median folate levels. Because folate is an essential cofactor in DNA methylation and nucleotide synthesis, there is ongoing concern that folate fortification may lead to increased risk for some forms of cancer, particularly among subjects harboring neoplastic foci. Indeed, early research in the 1940s showed that administration of folic acid to patients with malignancy actually results in rapid progression of the disease (6). One recent clinical study showed acceleration in the development of premalignant adenomas (7), and 2 other clinical trials observed an increase in cancer incidence (8, 9) among subjects randomly assigned to receive folic acid supplements. Nevertheless, 2 contemporary postfortification studies showed that compared with individuals on relatively low folate diet, individuals on adequate folate diets had reduced risk of colorectal cancer (10, 11), suggesting that adverse effects of folate might be limited to populations consuming large amounts of folic acid.

The folate controversy stems in part from the relative lack of mechanistic data about specific cellular pathways through which either low or high folate intake modulates the risk of colorectal cancer. We recently showed in human subjects that changing folate availability by dietary depletion or supplementation alters the expression of a number of genes involved in the inflammatory and immune response in the colonic mucosa (12). Another human study showed that the administration of 1.2 mg folic acid for 12 weeks increases the levels of serum proteins involved in the regulation and activation of immune function and the complement cascade (13). In addition, in the Aspirin/Folate Prevention of Large Bowel Polyps trial, folic acid administration was shown to antagonize the suppressive effects of aspirin on circulating inflammatory markers (14). These folate-associated molecular changes may provide clues as to how folate status might modulate the risk of colorectal carcinogenesis and neoplastic progression, but additional studies examining tumor outcomes are needed.

The aim of this study was to determine the effects of folate/methyl donor depletion on intestinal tumorigenesis in Apc^Min/+ mice. We hypothesized that folate/methyl donor depletion would attenuate the formation of intestinal tumors and, based on our recently published human data (12), reduce the expression of a panel of immune-related genes within the colonic mucosa. The use of antibiotics was avoided, thus preventing the severe methyl donor deficiency associated with suppression of the gut microflora (15, 16). Overall, our data show that folate/methyl donor deficiency markedly suppresses intestinal tumorigenesis in Apc^Min/+ mice. The effect is not solely dependent on reduced caloric intake, however, as further shown by a pair-feeding study with MDS diet. Importantly, the protection afforded to the intestine is highly dependent upon the timing and duration of the dietary intervention.

Animal experiments were conducted with approval from the Center for Laboratory Animal Care Committee (CLACC), The University of Connecticut Health Center. Apc^Min/+ mice, originally purchased from The Jackson Laboratory, were maintained on a C57BL/6J background. Genotyping was carried out by tail biopsy using PCR-based assays to determine the Apc gene status. As outlined in Fig. 1A, in study 1, a total of 39 Apc^Min/+ mice were randomized at 5 weeks of age and placed into 4 experimental groups. In groups 1 and 2, experimental diets were initiated at 5 weeks of age. In group 1, 12 Apc^Min/+ mice were fed a methyl donor sufficient (MDS) diet (TD.99366; Harlan Laboratories) for a total of 11 weeks. In group 2, 14 Apc^Min/+ mice were fed a methyl donor deficient (MDD) diet (TD.00605), devoid of folate, choline, methionine, and vitamin B12 (complete details of the diets are available in Supplementary Table S1) for a total of 11 weeks. In group 3, a total of 5 Apc^Min/+ mice received the MDS beginning at 10 weeks of age, whereas in group 4, 8 Apc^Min/+ mice received the MDD beginning at 10 weeks of age. In study 2, as outlined in Fig. 1B, a total of 37 Apc^Min/+ mice were randomized at 5 weeks of age and placed into 3 experimental groups. In each group, experimental diets were initiated in mice at 5 weeks of age for a total of 11 weeks. In group 1, Apc^Min/+ mice (n = 13) were fed the MDS ad libitum. In group 2, Apc^Min/+ mice (n = 12) were fed a MDD diet ad libitum. Finally, in group 3, to prevent weight differences between the experimental and control diets, Apc^Min/+ mice (n = 12) were pair-fed with MDS diet to the same amount consumed by the group 2 animals. All mice in each group were sacrificed at 16 weeks of age. Body weights of mice were recorded weekly and reported as shown in Fig. 2, Supplementary Fig. S1, and Fig. 3.

At the time of sacrifice, blood was collected from the posterior vena cava and centrifuged at 10,000 × g for 10 minutes at 4°C. Serum was collected and stored at −80°C until analysis. Serum folate concentrations were measured by a microtiter plate assay using Lactobacillus casei, as described previously (17).

Mice were sacrificed at 16 weeks of age. The small intestine and colon were harvested, flushed immediately with ice-cold PBS, and slit open longitudinally. Specimens were fixed flat in 10% buffered formalin and stored in 70% ethanol. For tumor quantitation, fixed tissues were stained with 0.2% methylene blue and examined under a dissecting microscope.

Sections of excised colons with no macroscopic evidence of tumors from the study 1 MDS group 1 (n = 3) and MDD group 2 (n = 5) were maintained in liquid nitrogen until analysis. Total RNA was extracted using the Trizol Reagent (Invitrogen). All RNA samples were assessed for quality on an Agilent 2100 Bioanalyzer (Agilent Technologies). The Microarray Core at Children's Hospital (Boston, MA) carried out array processing and cRNA hybridization. Briefly, total RNA was converted into cRNA labeled with biotin and hybridized onto the Mouse-Ref8 v1.1 Sentrix BeadChip Array (Illumina Corp.) for approximately 18 hours. The BeadChip Array was then washed and stained with Streptavidin-Cy3 and scanned by the Illumina BeadArray Reader (Illumina Corp.). Each sample was hybridized to 2 separate BeadChips. Images were quantified using the Illumina BeadStudio software and values were reported as average signal for each gene.

Expression data were analyzed by Genespring software (Agilent Technologies, Inc.,) after normalization. Before analysis, the arrays were normalized to the 50th percentile per chip (array) and median per gene. Quality control was carried out by analyzing gene expression correlation coefficients, and samples with coefficient less than 0.97 were excluded (one control and one intervention animal). Biologic duplicate samples signals were averaged. The differences in gene expression levels were determined using t tests, adjusted by the Benjamini–Hochberg method at a false discovery rate (FDR) of less than 0.1. The significant gene list was subjected to hierarchical clustering using squared Euclidian distance and a centroid linkage rule (18). This microarray data set has been deposited in the NCBI Gene Expression Omnibus with the accession number GSE37010. Subsequently, Gene Set Enrichment Analysis (GSEA) was used. GSEA is a computational method that determines whether an a priori defined set of genes show statistically significant differences between 2 phenotypes. We ranked gene expression differences between MDS and MDD groups to identify gene sets that were significantly enriched after methyl donor depletion (19, 20). FDR Q was used to rank the enrichment results.

For reverse transcriptase PCR (RT-PCR), duplicate 1-μg samples of total RNA were used as a template for cDNA synthesis using Superscript III First Strand kit (Invitrogen). cDNA was diluted in RT buffer and amount corresponding to 100 ng of original RNA was used for gene expression by quantitative RT-PCR (qRT-PCR). All qRT-PCR reactions used TaqMan Gene Expression Assay probes and primers (Applied Biosystems) and the ABI Prism 7500 RT-PCR system at the Liver Center (Yale University, New Haven, CT). Three genes that are related to inflammation were chosen from among the top 10 most downregulated genes (TNF-α, CXCL1, and CD28) and were measured by qRT-PCR to validate the expression array data. Quantification of expression changes used the delta method, and 18S mRNA was used as endogenous controls. Two independent RT-PCR reactions were used for each sample to calculate results.

Statistical analysis of the gene expression arrays is described above. For the comparison of serum folate levels, the number of intestinal polyps, and qRT-PCR data, statistical analyses were done using an unpaired t test, and P values are indicated in the figures. In study 2, one-way ANOVA with Bonferroni posttest was used to compare the number of intestinal polyps between groups. The difference in colon tumor size in response to methyl donor depletion was tested using χ² test (size cutoff of 2 mm).

In study 1, Apc^Min/+ mice were maintained on either the MDS or MDD diet, beginning at either 5 or 10 weeks of age (Fig. 1A). Regardless of timing, mice lost weight on the MDD diet, although the weight loss leveled off after 8 weeks on the experimental diet (Fig. 2). At the end of the study period, there was almost a 40% body weight difference between the MDS and MDD groups (Fig. 2), consistent with previous folate depletion studies in rodents (15, 21–23). Although the MDD diet caused a significant reduction in body weight gain, there were no premature deaths recorded. To confirm the effects of dietary folate depletion, the serum concentrations were also measured. As shown in Fig. 4A, the mean serum folate concentration in the MDD group was reduced by approximately 80% compared with the MDS dietary group (P < 0.0001) when dietary intervention with the MDD diet was initiated at 5 weeks of age.

To determine the effects of methyl donor depletion on Apc^Min/+-induced tumorigenesis, the number and size of small intestinal and colonic tumors were examined in all 4 dietary treatment groups. As shown in Fig. 4B, analysis of methylene blue–stained whole mounts of the small intestine revealed a significant reduction (80%, P < 0.0001) in tumor numbers in mice maintained for 11 weeks on the MDD diet compared with the MDS diet. Despite this marked suppression in tumor numbers, methyl donor depletion did not affect the size of tumors in the small intestine (data not shown). Consistent with previous studies in Apc^Min/+ mice (24–26), tumors were mainly confined to the small intestine, and their distribution was not affected by dietary methyl donor status. In the colon, although the total number of tumors was not affected by methyl donor depletion (Fig. 4C), there was a trend toward a reduction in the size of colon tumors in the MDD diet group (P = 0.063). As shown in Fig. 4D, there were no medium-sized (between 2 and 4 mm) or large-sized (≥4 mm) colon tumors in the MDD group, suggesting that methyl donor deficiency may prevent the progression of colon tumors in Apc^Min/+ mice.

When the start of the MDD diet was delayed until 10 weeks of age, serum folate was reduced by less than 25% in the MDD group (P < 0.08) as shown in Fig. 5A. However, cancer suppression was largely lost in the small intestine and colon (Fig. 5B and C). The inhibitory effect on colon tumor growth was also lost (Fig. 5D). The lack of protection, however, may be related to the mild reduction in serum folate levels. These results indicated that the timing and duration of methyl donor depletion is a critical factor in suppression of intestinal tumor formation in the Apc^Min/+ mouse model.

Because we observed extensive weight loss with the MDD diet, we conducted a second study (study 2) using a pair-feeding design (Fig. 1B) to adjust for the potential effect of weight loss on intestinal tumorigenesis. As shown in Fig. 3, pair-fed animals showed the same reduction in body weight gain when compared with the MDD group. However, even after adjusting for reduced body weight gain, the MDD diet induced a striking and significant further tumor reduction (∼96%, P < 0.0001), as shown in Fig. 6A. Also, as anticipated, the calorie-restricted animals (i.e., MDS pair-fed group) showed a significant reduction (∼54%, P < 0.0001) in tumor burden (Fig. 6A). In the colon, although the total number of tumors was not affected by methyl donor depletion (Fig. 6B), there was a nonsignificant (P = 0.285) reduction in medium-sized (between 2 and 4 mm) tumors and an absence of large-sized (≥4 mm) tumors in the MDD group (Fig. 6C). These data suggested that methyl donor deficiency may prevent the progression of colon tumors in Apc^Min/+ mice, a conclusion that is consistent with our findings in study 1.

RNA was isolated from normal-appearing colonic mucosa from study 1 mice maintained on the MDS and MDD diets for 11 weeks. A total of 1,009 genes were differentially expressed between the control and MDD groups at FDR less than 0.1 (refer to Supplementary Data for complete results); 516 genes were upregulated and 493 genes were downregulated. Hierarchical clustering of the most significantly altered genes represented as individual animals is shown in Fig. 7A. Among the significantly differentially regulated genes, many were downregulated and showed distinct clustering patterns (Fig. 7B). Within this set were a number of downregulated genes that are involved in various aspects of immune cell function and inflammation, as well as cell-cycle control.

To provide further validation of the microarray data, we selected 3 differentially regulated genes from the most significantly altered genes identified earlier by t tests. Included in this analysis are TNF-α, CXCL1, and CD28. As shown in Fig. 8, there was a marked downregulation in the expression of TNF-α, CXCL1, and CD28 (P = 0.0139, P = 0.0159, and P = 0.0005, respectively). These data confirmed the gene expression changes identified in Fig. 7B.

To gain additional insight into the molecular alterations that may be associated with methyl donor depletion and tumor suppression, we carried out a GSEA for Gene Ontology categories within the 2 dietary groups. Again, Gene Ontology showed that among the most downregulated categories were those related to immune cell function and inflammatory response. Other downregulated functional categories include cell turnover, comprising the chemokine and cytokine activity, cell-cycle progression, and mitosis. The most upregulated categories were related to extracellular structure and ribosome organization. GSEA for canonical pathway analysis showed that the most downregulated pathways included Toll-like receptor and T-cell receptor signaling, cytokine–cytokine receptor, cell adhesion molecules, and p53 signaling.

During the course of these studies, we began to observe a remarkable overlap in gene expression profiles within the colons of methyl donor–depleted Apc^Min/+ mice and that of a folate depletion study that was recently carried out in human subjects (12). On the basis of these similarities, we compared mouse GSEA data to our recently published human folate depletion study (12). Although we observed significant differences in gene expression patterns in the colons of mice maintained on the MDS and MDD diets, we cannot exclude the possibility that these changes may be due in part to the marked suppression in body weight gain observed in the MDD group. Nevertheless, a direct comparison of GSEA for both Gene Ontology categories and canonical pathways across the 2 species showed that folate and/or methyl donor depletion was associated with a significant downregulation of gene ontology categories and pathways involved in the inflammatory response and mucosal immunity, including components of the innate immune response (Fig. 9).

This study shows that dietary restriction of methyl donors, including folate, suppresses intestinal tumorigenesis in Apc^Min/+ mice. Although the tumor suppression is most dramatic in the small intestine, the primary site of tumors in this model, some trend toward a reduction in the size of tumors was also observed in the colon. Importantly, the protective effect of methyl donor depletion is strongly dependent upon the timing and duration of the dietary treatment. Gene expression profiling within the normal colonic mucosa identified a striking downregulation by dietary methyl donor depletion of a large number of genes involved in the immune and inflammatory response. These observed gene expression changes in mice overlapped extensively with expression changes observed within the human colorectum obtained from subjects maintained on low dietary folate (12), validating the use of this mouse model.

We observed that animals in the methyl-depletion group exhibited a reduced body weight gain during the experimental period. Previous nutritional studies in rodents have also shown that folate deficiency results in a reduction in body weight gain (15, 21–23). There is an extensive body of literature describing the effects of caloric restriction on protection from experimentally induced carcinogenesis, beginning with the seminal studies in the 1940s by Tannenbaum and coworkers (ref. 27; reviewed by Kritchevsky in ref. 28). Thus, we speculated that the tumor-suppressive effects observed in the Apc^Min/+ mice maintained on the MDD diet might be an indirect result of caloric restriction, especially as body weight gain was reduced by almost 40% during the course of treatment in this study. For example, in a recent study in primates, caloric restriction was found to significantly reduce the incidence of age-related diseases, including a suppression of gastrointestinal adenocarcinomas, the most common cancer found in rhesus monkeys, by approximately 50% compared with a control diet (29). In fact, caloric restriction was shown to reduce the frequency of intestinal polyps in Apc^Min/+ mice by almost 60%, an effect that was mediated, in part, through reduced levels of IGF-1 and leptin (30). Caloric restriction may also afford protection against colon carcinogenesis in human populations. In a study of colonic proliferation in obese subjects, caloric restriction significantly reduced markers of colonic proliferation (31). Importantly, these cancer preventive effects are also highly dependent on the extent and duration of caloric restriction, as well as the age at which the restriction was initiated in animal models (32–34). However, in a follow-up pairfeeding study in which Apc^Min/+ mice on MDS diet showed comparable weight loss to the MDD dietary group, we clearly showed that the cancer-suppressive effects of the MDD diet extend beyond the moderate protection observed in the calorie-restricted group.

Therefore, the tumor-suppressive effects observed in this study are unlikely to be due entirely to reduced food intake and the associated reduction in body weight gain. Furthermore, when methyl donor deficiency was delayed until 10 weeks of age, mice continued to lose weight (up to 40% at the end of the study) as shown in Supplementary Fig. S1, yet the cancer protection was essentially lost. In addition, low serum folate levels were comparable in both study groups. Moreover, the colons of methyl deficient animals exhibited gene expression profiles resembling published human data, yet human subjects were under strict dietary control and did not lose weight during the 3-month study (12). Thus these data indicate that the tumor-suppressive effect is most likely related to the timing and extent of methyl donor deficiency rather than simply caloric restriction.

In previous studies, the timing of folate intervention was also found to influence the number of intestinal tumors in Apc^Min/+ mice. When moderate folate supplementation was started before the formation of neoplastic foci in 3-week-old mice, there was a significant decrease in the number of intestinal and colonic tumors (35). However, when folate levels were moderately reduced in 6-week-old mice after neoplastic foci have developed, only small intestinal adenomas were reduced (35). The same group also reported that dietary folate supplementation at weaning suppresses both ileal and colonic tumor formation (24). This study shows that methyl donor depletion suppresses tumor formation when the dietary intervention was initiated at 5 weeks of age, at just about the time when neoplastic foci are established in this model (35). In contrast to a previous study (35), however, the suppressive effect was lost when the MDD intervention was initiated at 10 weeks of age after establishment of neoplastic foci. In addition, Lindzon and colleagues (36) tested the effect of folate supplementation in an azoxymethane-treated rat model. Supplementation with folic acid beginning 6 weeks after carcinogen treatment was found to promote the growth of preneoplastic aberrant crypt foci and tumors (36).

In this study, we describe significant alterations occurring within a number of signaling networks associated with the colonic mucosa in Apc^Min/+ mice that result from methyl donor deficiency. Using GSEA, we show that several of the most significantly enriched gene categories are related to immune and inflammatory response. Within these categories of inflammatory genes showing significant downregulation are a large number of cytokines and chemokines (Fig. 7B). Canonical pathway analyses showed that chemokine receptor signaling and innate and mucosal immune responses were also downregulated, in remarkable agreement with our recent human study in which dietary folate levels were reduced for up to 8 weeks (Fig. 9; ref. 12). In the human study, folate depletion also resulted in reduced expression of a number of immune response genes, including, most notably, cytokines and chemokines (12). Interestingly, folate supplementation in the same human study increased the expression of inflammatory and immune response genes (12). In a related human study, the administration of 1.2 mg folic acid for 12 weeks increased the levels of serum proteins involved in the regulation and activation of immune function such as acute phase reactants and complement fixation (13). The analysis of circulating inflammatory markers obtained from the human Aspirin/Folate Polyp Prevention trial suggested that folic acid administration antagonizes the colon cancer–suppressive effects of aspirin (14). Taken together, these studies suggested that folate/methyl donor status may modulate the intestinal inflammatory response.

Although we found that methyl donor deficiency reduced the expression of a panel of genes associated with the immune response, the impact of these expression changes on the inflammatory response within the colonic mucosa is not known at the present time. There is abundant evidence supporting the association of chronic inflammation and colon cancer risk (37), and numerous inflammatory signals are important modulators of colon cancer risk. In fact, the suppression of intestinal inflammation in Apc^Min/+ mice has already been shown to reduce tumor burden (26, 38–40). In addition, the use of anti-inflammatory agents such as NSAIDs can also reduce this risk, but the significant potential for serious adverse effects limit the long-term use of these agents for chemoprevention (41–43). Our mouse data clearly show that folate/methyl donor depletion does in fact markedly reduce tumor incidence in a high-risk animal model, but further studies are necessary to more definitely establish the implications of this dietary intervention on the human colonic mucosa.

In summary, this study addresses the influence of folate/methyl donor status on intestinal tumorigenesis. We show that this dietary restriction suppresses intestinal tumorigenesis in Apc^Min/+ mice. Although the tumor suppression is most dramatic in the small intestine, the primary site of tumor formation in this model, a trend toward tumor protection was also observed in the colon. Importantly, the protective effect of methyl donor depletion is strongly dependent upon the timing and duration of the dietary intervention. Gene expression profiling within the normal colonic mucosa identified a striking downregulation by dietary methyl donor depletion of a large number of genes involved in the immune and inflammatory response. These gene expression changes in mice generally resemble the results of our human folate depletion study (12), validating the use of this mouse model to study the interplay between dietary interventions and colon cancer risk.

E.J. Greenspan has held the title of program manager in NIH/National Cancer Institute.

Conception and design: K. Kadaveru, P. Protiva, E.J. Greenspan, Y.-I. Kim, D.W. Rosenberg

Development of methodology: E.J. Greenspan, D.W. Rosenberg

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Kadaveru, D.W. Rosenberg

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Kadaveru, P. Protiva, E.J. Greenspan, Y.-I. Kim, D.W. Rosenberg

Writing, review, and/or revision of the manuscript: K. Kadaveru, P. Protiva, E.J. Greenspan, Y.-I. Kim, D.W. Rosenberg

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P. Protiva, Y.-I. Kim, D.W. Rosenberg

Study supervision: D.W. Rosenberg

Additional microarray supporting material could be obtained on request from Petr Protiva (E-mail: [email protected]).

This work was supported in part by NIH grant 1R01CA125691, a generous gift from the Garre Family to the University of Connecticut Foundation and the Neag Comprehensive Cancer Center.

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