Genes with Aberrant Expression in Murine Preneoplastic Intestine Show Epigenetic and Expression Changes in Normal Mucosa of Colon Cancer Patients

Published OnlineFirst October 29, 2013; DOI: 10.1158/1940-6207.CAPR-13-0198

Cancer Prevention Research Article Research

Daniel Leclerc1, Nancy Levesque 1, Yuanhang Cao1, Liyuan Deng1, Qing Wu1, Jasmine Powell2, Carmen Sapienza2, and Rima Rozen1

Abstract An understanding of early genetic/epigenetic changes in colorectal cancer would aid in diagnosis and prognosis. To identify these changes in human preneoplastic tissue, we first studied our mouse model in þ þ þ which Mthfr / BALB/c mice fed folate-deficient diets develop intestinal tumors in contrast to Mthfr / BALB/c mice fed control diets. Transcriptome profiling was performed in normal intestine from mice with low or high tumor susceptibility. We identified 12 upregulated and 51 downregulated genes in tumor-prone mice. Affected pathways included retinoid acid synthesis, lipid and glucose metabolism, apoptosis and inflammation. We compared murine candidates from this microarray analysis, and murine candidates from an earlier strain-based comparison, with a set of human genes that we had identified in previous methylome profiling of normal human colonic mucosa, from colorectal cancer patients and controls. From the extensive list of human methylome candidates, our approach uncovered five orthologous genes that had shown changes in murine expression profiles (PDK4, SPRR1A, SPRR2A, NR1H4, and PYCARD). The human orthologs were assayed by bisulfite-pyrosequencing for methylation at 14 CpGs. All CpGs exhibited significant methylation differences in normal mucosa between colorectal cancer patients and controls; expression differences for these genes were also observed. PYCARD and NR1H4 methylation differences showed promise as markers for presence of polyps in controls. We conclude that common pathways are disturbed in preneoplastic intestine in our animal model and morphologically normal mucosa of patients with colorectal cancer, and present an initial version of a DNA methylation-based signature for human preneoplastic colon. Cancer Prev Res; 6(11); 1171–81. 2013 AACR.

Introduction methylation differences between colorectal tumors and Nearly one million people worldwide develop colorectal adjacent tissues (2–4). However, there are limited data on cancer every year (1). Colorectal cancer results from a differential methylation in normal colonic mucosa combination of environmental and genetic factors that between controls and patients with colorectal cancer convert normal epithelium into a malignant tumor through (5, 6 and our own recent report (7)). To identify protu- multiple stages. An understanding of early events in tumor- morigenic changes in normal human colonic mucosa, we igenesis will lead to timely diagnoses and improved began with the identification of candidate genes in our outcomes. mouse model that had previously been shown to develop Epigenetic changes are early events in colorectal cancer intestinal tumors after administration of low folate diets and other neoplasias. There are numerous genes with (8). Since folates generate methyl groups for DNA methylation, we predicted that there would be genetic/epigenetic changes in preneoplastic intestine in our mouse model and that some of these changes might be similar Authors' Affiliations: 1Departments of Human Genetics and Pediatrics, McGill University, Montreal Children's Hospital site of the McGill University to those seen in human colonic mucosa. We therefore Health Centre Research Institute, Montreal, Canada; and 2Fels Institute for compared murine candidates to the unrestricted list of Cancer Research and Molecular Biology, Temple University School of Medicine, Philadelphia, Pennsylvania preliminary human genes that had shown changes fol- lowing methylation profiling of normal colonic mucosa Note: Supplementary data for this article are available at Cancer Prevention Research Online (http://cancerprevres.aacrjournals.org/). in colorectal cancer patients and controls (7 and unpublished data). Corresponding Author: Rima Rozen, Montreal Children's Hospital Research Institute, 4060 Ste-Catherine West, Room 200, Montreal, Our mouse model reflects some of the genetic and nutri- Canada H3Z 2Z3. Phone: 514-412-4400, ext. 23839; Fax: 514-412- tional factors that affect risk for human colorectal cancer. 4331; E-mail: [email protected] Individuals with low folate intake are more susceptible to doi: 10.1158/1940-6207.CAPR-13-0198 colorectal cancer than individuals with adequate intake (9). 2013 American Association for Cancer Research. A polymorphism in methylenetetrahydrofolate reductase

www.aacrjournals.org 1171

Downloaded from cancerpreventionresearch.aacrjournals.org on October 1, 2021. © 2013 American Association for Cancer Research. Published OnlineFirst October 29, 2013; DOI: 10.1158/1940-6207.CAPR-13-0198

Leclerc et al.

(MTHFR), which generates methyl groups for S-adenosyl- Materials and Methods methionine–dependent methylation reactions, also mod- Mice and diets ulates colorectal cancer risk. The association between the Animal experimentation was approved by the Montreal MTHFR 677C!T polymorphism and risk for colorectal Children’s Hospital Animal Care Committee, in accordance cancer has been found in several epidemiologic studies and with Canadian Council on Animal Care guidelines. After þ þ þ the proposed mechanisms involve epigenetic remodeling weaning, BALB/c Mthfr / and Mthfr / mice were fed CD through DNA methylation changes (9–11). (2 mg folate/kg diet) or FD diets (0.3 mg folate/kg diet) for TheimpactoffolateandMTHFRdeficiencyontumor- one year. Incidence of neoplasia was consistent with our igenesis is apparent in our mouse model for spontaneous previous reports (8, 15). intestinal neoplasia in BALB/c mice. BALB/c and C57BL/ 6 mice were fed folate-deficient (FD) or control diets Control subjects (CD) for one year. Tumors were only observed in BALB/c Research was approved by the Temple University Office mice fed FD (8, 12); a single functional copy of the Mthfr for Human Subjects Protections Institutional Review Board, gene increased the number of FD mice with tumors (8). protocol 11910. We collected biologic specimens from Several tumor-predisposing candidate genes involved in subjects undergoing routine screening colonoscopy at Tem- cell-cycle control, cell survival, and DNA repair were ple University Medical Center to serve as the control arm of identified by comparing expression profiles of tumor the study (Supplementary Table S1). We excluded subjects tissue with normal tissue (13). These observations, in with a personal or first-degree family history of any cancer combination with increased DNA damage (14, 15), and subjects with a previous colonoscopic finding of decreased expression of tumor suppressors and increased polyps. Subjects who were not excluded underwent com- retinoid/PPARA pathway activity in BALB/c normal pre- plete colonoscopic evaluation by a board certified gastro- neoplastic intestine (15), could explain the susceptibility enterologist. If the colonoscope could not be passed to the of these mice to intestinal tumorigenesis. We observed appendiceal orifice, the subject was excluded. If the com- differential expression of Bcmo1, Aldh1a1,andSprr2a plete colon was visualized, two cold forceps biopsies of when comparing CD- and FD-fed BALB/c mice. Expres- normal colonic mucosa from the ascending colon (proxi- sion differences were also seen for Bcmo1 between mal to the hepatic flexure) were pooled. þ þ þ Mthfr / and Mthfr / mice. These changes were consistent with the hypothesis that enhanced RXR/PPAR activ- Cancer patients ity would increase oxidative stress/damage and lead to We collected biologic specimens from patients undergo- neoplasia (15). ing colon resection for presumed or biopsy-proven colon Our unique mouse model, without germline mutation or cancers. Patients were considered eligible if they had no carcinogen induction, provides an opportunity to study personal or family history of colon cancer before this early events in intestinal neoplasia. In this report, to identify encounter. Patients with known or clinical features of specific candidate genes in tumorigenesis, we used micro- hereditary cancer syndromes (specifically, hereditary non- þ þ arrays to compare BALB/c Mthfr / CD mice and BALB/c polyposis colorectal cancer, or familial adenomatous poly- þ Mthfr / FD mice, which have relatively lower and higher posis syndrome) were excluded. Patients with any personal intestinal tumor susceptibility, respectively (15). We found history of chemotherapy or radiotherapy were also exclud- significant differences in retinoid/PPARA pathway genes ed. Patients who remained eligible (Supplementary Table between BALB/c mice fed FD and CD. The activation of S2) underwent colon resection at a single National Cancer this pathway is consistent with the findings from our Institute designated Comprehensive Cancer Center (Fox previous report on gene expression profiling between Chase Cancer Center/Temple University, Philadelphia, PA). tumor-susceptible BALB/c and tumor-resistant C57BL/6 Specimens, determined by a board certified pathologist to mice (15). be normal appearing colon mucosa, were obtained well We compared our murine candidates, obtained from away from the lesion (10 cm). both of the aforementioned expression microarrays, to the extensive list of candidate human genes identified DNA and RNA isolation from human normal tissue through methylation profiling (7 and unpublished data). Morphologically normal colon mucosa specimens were Our inter-species comparison led to the identification of 5 obtained from colorectal cancer patients or from controls human genes that showed significant pyrosequencing- undergoing screening colonoscopy (7). Samples were trea- based methylation differences, in 14 CpG dinucleotides, ted as described (7). as well as significant expression differences, in normal human colonic mucosa between patients with colorectal RNA extraction from murine normal preneoplastic cancer and controls. Our results suggest that common intestine tumorigenic mechanisms, reflecting an altered metabolic RNA was extracted as described (15). Eight samples for þ state, are shared by our mouse model and human colo- microarrays were prepared from 4 BALB/c Mthfr / mice fed þ þ rectal cancer. Furthermore, these methylation differences FD and 4 BALB/c Mthfr / mice fed CD. High quality of contribute to an epigenetics signature for diagnosis of RNA was verified (Supplementary Fig. S1). In addition, 16 þ colonic neoplasia. RNA samples were extracted from BALB/c Mthfr / and

1172 Cancer Prev Res; 6(11) November 2013 Cancer Prevention Research

Epigenetics in Mouse/Human Preneoplastic Intestine

þ þ þ þ BALB/c Mthfr / mice fed CD and FD (4 mice per group). pared with CD Mthfr / BALB/c mice (Supplementary These biologic replicates were used to confirm microarray Table S6). results by quantitative real-time RT-PCR (qRT-PCR) and These 63 genes were grouped based on functions using verify effects of genotype and diet on expression. IPA (Supplementary Fig. S3B). The top 3 categories were lipid metabolism, small-molecule biochemistry, and Microarray analysis and quantitative real-time RT-PCR nucleic acid metabolism. Fatty acid metabolism, LPS/IL- Microarray experiments were performed using Affymetrix 1–mediated inhibition of RXR function, and PXR/RXR Mouse Gene 1.0 ST Array Chips, as previously described activation were identified as pathways with the most sig- þ þ (15). We considered BALB/c Mthfr / mice fed CD as the nificant changes (data not shown). þ group with higher tumor resistance and BALB/c Mthfr / FD group as the tumor-susceptible group. Genes with Evidence for involvement of PPARA in tumorigenesis expression fold changes more than 1.4 and a P value less Our previous study compared gene expression between than 0.01 after false discovery rate correction were consid- the C57BL/6 and BALB/c mouse strains with different ered significant. Ingenuity Pathways Analysis (IPA) was sensitivity to intestinal tumorigenesis. In that report, we used to assess biologic processes with the most significant showed that the PPARA-oxidation pathway plays a critical changes. role (15). The present work, based on diet and genotype qRT-PCR was performed as described to confirm micro- comparisons in BALB/c, confirms the involvement of array data (15). Primers were designed (Supplementary Table PPARA. Our gene expression profiling identified several S3) and amplified fragments of expected sizes (data not genes related to PPARA activation and oxidative stress that shown). We used 16 mice in 4 groups (4 mice per group); are affected by diet and/or Mthfr genotype (Supplementary þ þ þ þ þ the groups were Mthfr / CD, Mthfr / FD, Mthfr / CD, Table S7), as well as PPARA-responsive genes (Supplemen- þ and Mthfr / FD. tary Table S8). Responsiveness to PPARG was also reported RNA extraction, cDNA synthesis, and gene-specific Taq- for some of these genes (references listed in Supplementary Man probes (Applied Biosystems) were done as described Table S8). (7), to measure steady-state levels of PDK4, SPRR1A, SPRR2A, NR1H4, and PYCARD in normal colon mucosa Confirmed expression changes for eight genes that may þ from patients with cancer and controls. Primers and probes promote tumorigenesis in Mthfr / FD mice are described in Supplementary Table S4. Expression of eight genes that may influence tumorigenesis was confirmed by qRT-PCR (Supplementary Fig. S4). Quantitative CpG methylation analysis by These genes are involved in regulation of proliferation (Atf3; pyrosequencing ref. 17), apoptosis (Plscr1 and Plscr2; ref. 18), cell survival We used bisulfite pyrosequencing to measure methyla- (Ppme; ref. 19–21), recognition of aberrant unmethylated tion of specific CpGs in the 50 region of human PDK4, DNA (Pdctrem; ref. 22), overexpression or chromosomal SPRR2A, NR1H4, SPRR1A, and PYCARD. CpGs in mouse rearrangement related to cancer (Lhfpl2; refs. 23, 24), or orthologous regions were also assessed, as described (15). reduction of retinaldehyde levels (Rdh18 and Akr1c13; Genes and relevant oligonucleotides are presented in Sup- ref. 25). plementary Table S5. Representative pyrograms are shown in Supplementary Fig. S2A–S2F. Five additional murine genes and their human orthologs show changes in expression or methylation Statistical analysis in mice and in normal intestinal mucosa of colorectal Quantitative data are presented as average value of repli- cancer patients cates SEM. Levene test was performed to assess equality of We hypothesized that some protumorigenic mechan- variance. Two-factor ANOVA was used to evaluate effects of isms in our mice might be shared by human preneoplastic diet and genotype on gene expression; strain and diet were colon. To pinpoint the involved genes, we first looked also compared in some cases. Student t test for independent at genes identified in the aforementioned murine micro- samples was performed for specific comparisons where array (affected by diet or Mthfr genotype, Supplementary indicated. Analyses were done using SPSS for WINDOWS Table S6), that would match human orthologs with software, version 11.0. P values < 0.05 were considered demonstrated methylation changes in our recent significant. genome-wide profiling of DNA methylation of normal colonic mucosa in patients with colorectal cancer and Results controls (ref. 7 and C. Sapienza, 2012, unpublished data). þ þ þ Differences between BALB/c Mthfr / FD and Mthfr / We limited our selection to human genes with methyla- CD gene expression profiles tion changes more than 2% and for which increased/ Our microarray results have been deposited in the Gene decreased methylation could correspond to decreased/ Expression Omnibus database (GEO, ref. 16, GEO acces- increased expression in murine mucosa. We also focused sion no. GSE34011). There were 63 genes with significant on genes that were related to the PPAR/oxidative stress expression changes (51 increased and 12 decreased; Sup- pathway. Only three genes passed this stringent screen: þ plementary Fig. S3A) in FD Mthfr / BALB/c mice com- PDK4, SPRR2A,andNR1H4. We applied the same filters

www.aacrjournals.org Cancer Prev Res; 6(11) November 2013 1173

Leclerc et al.

to the list of mouse genes deduced from our previously The human orthologs of the five mouse genes presented publishedstraincomparison(15);thisscangenerated in Fig. 1 showed altered methylation levels in our genome- only two additional candidates: SPRR1A and PYCARD. wide methylation array (7). We confirmed these changes by Microarray-based methylation changes in PYCARD had bisulfite pyrosequencing-based assays for 6 CpGs in the 5 been published (7); the other four genes had not been genes in 6 controls and 6 patients with colorectal cancer. reported. There was excellent correlation between the two methods Confirmation of microarray data for these five murine when b-values from microarrays were compared with genes was undertaken. Pdk4, a target of PPARA, is a positive %methylation from pyrosequencing data for the 12 indi- regulator of glycolysis (26). It is upregulated by FD in mice viduals: Spearman r ¼ 0.94, a total of 72 points; linear with both Mthfr genotypes (Fig. 1A). Expression of Pdk4 is regression r2 ¼ 0.93; P < 0.001 (Supplementary Fig. S5). We þ þ þ also higher for Mthfr / mice than Mthfr / mice, for both then examined a new cohort of 29 controls and 29 patients diets. We previously reported that Sprr2a is downregulated with colorectal cancer, who had not been tested in the in BALB/c compared with C57BL/6 (15). This ROS scaven- original arrays, by pyrosequencing (Supplementary Fig. ger plays a role in tumorigenic events related to oxidative S6, left panel for each marker). They were compared with stress. FD lowered Sprr2a in both genotypes (Fig. 1B). 12 controls and 24 patients with colorectal cancer that had Nr1h4, also known as Fxr, upregulates cell growth and been tested for these same 6 CpGs by arrays only (Supple- þ induces PPARA (27). FD Mthfr / mice demonstrated mentary Fig. S6, right panel for each marker). There are higher expression of Fxr than wild-type mice. Folate defi- significant differences between controls and patients, for þ ciency also stimulated its expression in Mthfr / mice (Fig. these independent cohorts of 58 and 36 individuals tested 1C). Sprr1a has a similar role to Sprr2a and showed by pyrosequencing and microarrays, respectively. decreased levels in BALB/c mice compared with C57BL/6 We then expanded our pyrosequencing assessment to a (Fig. 1D). Surprisingly, a marked elevation by FD was seen total of 14 CpGs for these 5 genes (for 35 controls and 35 in C57BL/6 (Fig. 1D), whereas this was not observed for patients) as seen in Fig. 2. The additional dinucleotides were Sprr2a (Fig. 1B). PYCARD downregulation is well docu- in the vicinity of the originally interrogated CpGs. All 14 mented in colorectal cancer (28). Pycard showed lower tested CpGs showed significant differences in methylation expression in BALB/c compared with C57BL/6, in both in normal mucosa between patients with colorectal cancer diets (Fig. 1E). and controls (Fig. 2A–E). CpGs showed significantly

Figure 1. Effect of diet, Mthfr genotype, or strain on expression of five genes in murine normal intestine; human orthologs for these genes are examined in Figs. 2–5. Gene names are indicated above the graphs. Bars with black and white backgrounds represent data for C57BL/6 (B6) and BALB/c (C) mice, respectively. Values are means SEM. , P < 0.05 and , P < 0.005, diet effect; #, P < 0.05, genotype effect; and , P < 0.001, strain effect (two-factor ANOVA). #, P < 0.05, genotype effect for FD mice; and , P < 0.05, diet effect in Mthfrþ/ mice (independent t tests). For Sprr1a in (D), two-factor ANOVA indicated a strain diet interaction. Post hoc Tukey comparisons indicated a significant diet effect in C57BL/6 mice (, P < 0.001) and a significant strain effect in FD mice (, P < 0.001).

1174 Cancer Prev Res; 6(11) November 2013 Cancer Prevention Research

Epigenetics in Mouse/Human Preneoplastic Intestine

Figure 2. DNA methylation of 5 genes in normal colonic mucosa discriminates between controls and colorectal cancer (CRC) subjects. DNA methylation was determined for PDK4 (A), SPRR2A (B), NR1H4 (C), SPRR1A (D), and PYCARD (E) genes. In this ﬁgure and in Fig. 3, numbering refers to the NCBI36/hg18 version of the UCSC Genome Browser (http://genome.ucsc.edu/). Individual CpGs assayed in the microarray study (12) are boxed. Controls (35 individuals) are shown as black bars, colorectal cancer patients (35 subjects) as white bars. Values are means SEM. , P < 0.05; , P < 0.01; , P < 0.005; , P < 0.001, independent t tests.

decreased methylation in normal colon of patients with markers in patients with colorectal cancer (Fig. 4A–E). No colorectal cancer for PDK4 (Fig. 2A) and NR1H4 (Fig. 2C). expression differences were seen between controls with Significantly higher methylation was observed for SPRR2A polyps and those without polyps. (Fig. 2B), SPRR1A (Fig. 2D), and PYCARD (Fig. 2E). We Although average methylation of our five candidate genes then questioned whether these differences could be used as was significantly different between patients with colorectal methylation-based biomarkers for presence or absence of cancer and controls (Fig. 2A–E), there is some overlap polyps in controls. Approximately half of the controls had between the two groups, for each of the five markers. We polyps. The average methylation of NR1H4 was lower for hypothesized that the methylation pattern for a collection subjects with polyps, although the difference did not reach of markers could have significant diagnostic power to statistical significance (Fig. 3A). However, we observed an distinguish normal mucosa between colorectal cancer increased average methylation for the five PYCARD CpGs, patients and controls. Indeed, hierarchical clustering for with significance for two CpGs (16:31121937 and classification of 70 individuals led to the identification of 2 16:31121918), and borderline significance for three CpGs major clusters: one cluster comprised predominantly of (16:31121929, 16:31121927 and 16:31121902; Fig. 3B). colorectal cancer patients and a second cluster composed We compared transcript levels of these five genes in exclusively of controls (Fig. 5A). normal mucosa of colorectal cancer patients and controls, We applied the same principle to evaluate the discrimi- and observed significantly increased expression for the five natory ability for classification of 35 controls with (n ¼ 17)

www.aacrjournals.org Cancer Prev Res; 6(11) November 2013 1175

Leclerc et al.

Figure 3. CpG methylation for NR1H4 and PYCARD in normal mucosa of controls without or with polyps. CpG methylation was assessed for controls without (n ¼ 18) or with (n ¼ 17) polyps, for NR1H4 (A) and PYCARD (B). Controls without polyps are represented by black bars, controls with polyps as white bars. Values are means SEM. , P < 0.05; borderline signiﬁcant for PYCARD CpGs 16:31121929, 16:31121927 and 16:31121902 (0.07, 0.09 and 0.08, respectively); independent t tests.

or without (n ¼ 18) polyps. Empiric combinatory compar- generates oxidative stress/damage, and enhances glycolysis, ison using different sets of these five biomarkers (data not setting the stage for tumorigenesis. Tumors have altered shown) revealed that NR1H4 and PYCARD aggregated data energy metabolism, with a preference for aerobic glycolysis provide the best combination for a potential methylation (Warburg effect), instead of the tricarboxylic acid cycle (31). signature (Fig. 5B). Hierarchical clustering yielded one Because of the strong link between high fat diets and group almost exclusively composed of controls with polyps development of colon cancer, we had previously argued (8/9 subjects) and a second group enriched in controls that the epigenetic reprogramming of lipid and carbohy- without polyps (17/26 subjects). drate metabolism probably preceded tumor development and was not programmed by the tumor at distant sites (7). þ Discussion Interestingly, only 25% of BALB/c Mthfr / mice developed þ DNA-based biomarkers in normal colonic mucosa would tumors (15), and the BALB/c Mthfr / mice used for con- be extremely useful because they have the potential to be firming mouse expression microarray data by qRT-PCR did diagnostic of colon cancer in the near term or, upon further not have tumors. These findings directly support the development, may become prognostic indicators of colon hypothesis that gene expression changes and reprogram- cancer risk. Such biomarkers would provide discriminatory ming occur before tumor formation. and quantitative biochemical measures to supplement the The retinoid pathway and metabolism of lipids and current endoscopic screening test that is both invasive and carbohydrates were predominant in murine expression subjective. profiling (Supplementary Fig. S3B) and human genome- Environmental factors, such as diet, may be the most wide methylation assessment (7). Correlation between important influences on colorectal cancer risk. Low dietary vitamin A deficiency and tumor initiation has been con- folate is one such risk factor. The polymorphism in MTHFR firmed in several studies (32 and references therein). Reti- (677C!T) can also increase cancer risk when folate status is noic acid regulates gene expression through the retinoic acid inadequate. Our mouse model, which develops intestinal receptor and retinoid X receptor (RAR/RXR) heterodimer. neoplasia after low dietary folate, is a relevant model for In addition, RXR interacts with other nuclear receptors such human sporadic colorectal cancer because the mice do not as PPARs. Retinoids cannot be synthesized in humans; they have germline mutations and develop tumors over an are converted from dietary carotenoids (33). Beta-carotene extended period of time, without carcinogen induction. is the major provitamin A carotenoid. Retinaldehyde, the The use of Mthfr-deficient mice allows us to examine product of BCDO1, prevents formation of the RXR/PPAR gene-nutrient interactions that have also been observed in heterodimer. Downregulation of Bcmo1, as observed in þ human colorectal cancer. DNA methylation is altered by Mthfr / and FD mice, would result in lower retinaldehyde MTHFR 677C!T genotype and folate levels, with folate- levels and increased PPARA activity. Retinaldehyde can be deficient TT individuals showing the lowest global DNA converted to retinoic acid or retinol by aldehyde dehydro- methylation and the highest prevalence of cancer history genases or aldo-keto reductases, respectively. We observed (29). higher Aldh1a1 expression in FD-fed BALB/c mice (Supple- Another dietary risk factor for colorectal cancer is high fat mentary Table S7 and Fig. 4B of ref. 15) and higher expres- þ (30). Expression profiling in mice revealed significant dif- sion of Akr1c13 in BALB/c Mthfr / FD mice (Supplemen- ferences for genes downstream of PPARA, a major regulator tary Fig. S4); these changes would also contribute to low- of lipid and glucose metabolism. We hypothesize that a ering retinaldehyde. There was increased expression of þ disturbance in folate metabolism can result in activation of Rdh18 in BALB/c Mthfr / FD mice; retinol dehydrogenases the RXR/PPARA pathway that increases fatty acid oxidation, can also metabolize retinaldehyde, although it is unclear

1176 Cancer Prev Res; 6(11) November 2013 Cancer Prevention Research

Epigenetics in Mouse/Human Preneoplastic Intestine

Figure 4. Real-time RT-PCR analysis of transcript levels in normal colon mucosa of individual controls and patients with colorectal cancer for PDK4 (A; 23 control, 22 cancer), SPRR2A (B; 13 control, 20 cancer), NR1H4 (C; 23 control, 20 cancer), SPRR1A (D; 10 control, 20 cancer), and PYCARD (E; 23 control, 19 cancer). , P < 0.05; , P < 0.005; , P < 0.001, independent t tests.

whether the Rdh18 transcript encodes a functional protein different methods and sites of biopsy could result in sam- (GenBank accession# AY053573). We have recently shown ples with different cell types, our analysis of methylation at that inhibiting BCMO1 expression increases invasion and more than 27,000 CpGs revealed differences for only 909 migration in human colorectal cancer cells, and that b-car- (3.3%) between cancer and control specimens at our least otene, the BCMO1 substrate, upregulates the gene and stringent statistical threshold (7). To address the issue of reverses these effects (34). tumor sidedness, we examined normal colon from colorec- To identify early human colorectal cancer events, we tal cancer patients with left- or right-sided tumors (n ¼ 15 in compared our list of murine candidate genes (Supplemen- each group). When we compared the 14 CpGs between tary Table S6 in the present report and Supplementary Table groups, there was only one marker (one of the three CpGs in S6 in ref. 15) with genes identified in methylation profiling SPRR2A) that showed a significant difference in methyla- of normal human colon (7 and C. Sapienza, 2012, Unpub- tion (data not shown). Aging can also influence DNA lished data) because disturbances in folate metabolism methylation and modify colorectal cancer risk (36). How- perturb methylation in both species (12, 35). This approach ever, since the age ranges in controls and patients were identified five mouse and human orthologous genes that similar and there was no difference between the mean ages showed, respectively, changes in expression in murine of these groups in our original study (7), from which the five microarrays and changes in human methylation profiling. candidates were selected for validation in this study, it is Bisulfite-pyrosequencing of human DNA in independent unlikely that age was a confounder. samples of normal mucosa of patients with cancer and Some of the controls had polyps, and one of the above controls confirmed significant methylation differences in genes, PYCARD, showed significant methylation differences a total of 14 CpGs (Fig. 2). These genes were: NR1H4 which between controls without polyps and controls with polyps can activate PPARA (27); PDK4, a target of PPARA that (Fig. 3). Interestingly, controls included two cases of HPP enhances glycolysis (26); PYCARD, a proapoptotic gene (hyperplastic polyps), thought not to give rise to colon (28); and two different members of the SPRR family, tumors, and they are at the very low end of the normal involved in protection against oxidative damage. Although PYCARD methylation distribution while other controls

www.aacrjournals.org Cancer Prev Res; 6(11) November 2013 1177

Leclerc et al.

Figure 5. Toward the establishment of epigenetic signatures of cancer or polyps based on methylation of specific genes in normal colonic mucosa. A, DNA methylation in normal intestine may establish a signature for presence of tumors. Unsupervised hierarchical cluster matrix of PDK4, SPRR2A, NR1H4, SPRR1A, and PYCARD according to their respective levels of DNA methylation in their 50 regulatory regions. The epigenetic profile of 35 patients with tumors (blue boxes on the right) and 35 controls (orange boxes on the right) was assessed by bisulfite pyrosequencing of DNA extracted from normal intestine mucosa. Data from the 11 CpGs with significance at P < 0.01 in Fig. 2 were used for this analysis. The blue and orange dashed lines define the limits of the two major sample clusters, with almost exclusive segregation of patients with colorectal cancer or control samples, respectively. B, NR1H4 and PYCARD CpG methylation was analyzed as in (A), but using only the 35 controls, to distinguish between controls with and without polyps. Controls with polyps are shown on the right as yellow boxes, and controls without polyps are shown in red. The asterisks indicate the two controls with hyperplastic polyps. Yellow and red dashed lines depict two large clusters, comprised mainly of individuals with or without polyps, respectively. Dendograms are shown on the left of heatmaps. In the heatmaps, dark boxes indicate low levels of CpG methylation, bright boxes represent highly methylated CpGs. Cluster 3.0 and Java TreeView v1.1.5r2 software were used to perform hierarchical clustering and to visualize results.

with polyps are all in the upper part of the normal PYCARD involving binding efficiency of transcriptional repressor(s) distribution (Supplementary Table S1). Furthermore, in the to methylated regions for example, may also result in heatmap in Fig. 5B, these 2 subjects cluster with the controls changes in expression (39). This mechanism may be rele- without polyps. vant here, since the assessed CpGs are all in the 50 region of A major outcome of this study is the potential for using genes. the methylation differences per se as molecular biomarkers The increased expression of PDK4 and NR1H4 in patients for diagnosis. Although some methylation differences were with colorectal cancer (Fig. 4A and C) is consistent with the þ relatively small, they were observed in two different cohorts, higher expression of their orthologs in BALB/c FD / mice using two different methodologies (Supplementary Figs. S5 (Fig. 1A and C). These results suggest that similar molecular and S6), and are therefore reliable. These five genes also changes exist in human and murine preneoplastic intestine. exhibited significant expression differences in normal Higher expression of SPRR1A, SPRR2A, and PYCARD in human mucosa between controls and patients with colo- normal tissue of patients with colorectal cancer may repre- rectal cancer, with increased expression in patients (Fig. 4). sent compensatory effects in response to the tumorigenic Increased expression was associated with both decreased environment. Only 20% of mice show tumors at one year and increased methylation. While it is true that there is a and they are quite small (1–2 mm); patients with colorectal general (but not absolute) inverse correlation between DNA cancer have well-developed tumors and their normal muco- methylation in the promoter region and transcript levels, sa may have had sufficient time for compensatory expres- our finding is not unusual since methylation within the sion changes. Mice lacking Pycard demonstrate polyp for- body of genes and distal to genes is often positively corre- mation (40); PYCARD is an inflammasome-associated mol- lated with transcript levels (37, 38). Specific mechanisms ecule with influences on diabetes, obesity, and cancer (41).

1178 Cancer Prev Res; 6(11) November 2013 Cancer Prevention Research

Epigenetics in Mouse/Human Preneoplastic Intestine

SPRR genes are often induced in response to stress and Candidate genes were intersected with expression profil- extensively upregulated in various types of cancer (42). ing data from our mouse model. This original, two-filter SPRR2A is overexpressed in early stages of prostate tumor- approach, resulting in gene identification within com- igenesis (43), and is modulated by 5-aza-20-deoxycytidine mon pathways, provides a solid basis for an epigenetics despite the absence of CpG islands (44). signature for normal intestine in colon cancer. As colo- Increased expression of PDK4 and NR1H4 in normal rectal cancer development proceeds through multiple þ mucosa of colorectal cancer patients and in Mthfr / mice stages, it would be useful to develop diagnostic tests for fed FD may be highly tumorigenic. As mentioned, the shift early intervention. Although many studies have reported away from mitochondrial respiration is a hallmark of tumor methylation differences between normal colon and metabolism (31, 45). Inhibition of pyruvate dehydroge- tumors, there are very few genes that have been con- nase (PDH) through phosphorylation, by pyruvate dehy- firmed by quantitative methods to exhibit methylation drogenase kinases (PDK), results in decreased respiration differences between controls and patients with colorectal in tumors (45). PDKs are a family of four kinases in cancer in normal mucosa (5, 7). Our approach has humans (26); siRNA-based knockdown of PDK1 reverses identified 5 genes, at 14 CpG sites, using the highly PDH inhibition and the Warburg effect, and can inhibit quantitative pyrosequencing method which could be tumor growth (45). PDK4 expression is increased by adapted into a clinical setting (47). Methylation changes PPARA, by consumption of high fat diets and in diabetic accumulate over years and could serve as sentinel markers states (26); its role in transformation has not been well before the appearance of polyps. Shedding of colon- studied. Decreased methylation of PDK4 in colorectal derived DNA into stool could allow noninvasive testing cancer mucosa is consistent with increased expression. (48). If differences are systemic, measurements of meth- We also observed significantly decreased methylation in ylation changes in peripheral blood or saliva would also the 50 region of Pdk4 in Mthfr-deficient mice, for both be extremely useful. Additional studies using our two- diets (data not shown). pronged approach may lead to identification of other NR1H4 encodes the farnesoid-X-receptor (FXR). Bile biomarkers for the establishmentofabiochemicalmea- acids, natural ligands for this receptor, can induce PPARA sure of cancer risk that may be more objective than through a FXR response element in the human PPARA routine endoscopy. promoter (27). NR1H4 activation increases PDK4 expres- NR1H4 sion (46). Decreased methylation of in human Disclosure of Potential Conflicts of Interest colorectal cancer mucosa is consistent with increased No potential conflicts of interest were disclosed. expression. In mice, we observed increased Nr1h4 methyl- Mthfrþ/ Mthfrþ/þ ation for FD mice compared with FD Authors' Contributions mice, and a trend for increased methylation at two CpGs in Conception and design: D. Leclerc, C. Sapienza, R. Rozen þ þ the 50 region, in Mthfr / FD mice compared with Mthfr / Development of methodology: D. Leclerc, N. Levesque, Y. Cao, L. Deng, C. Sapienza CD mice (data not shown). Acquisition of data (provided animals, acquired and managed patients, Because tumorigenesis is a complex process, there are provided facilities, etc.): D. Leclerc, N. Levesque, Y. Cao, L. Deng, Q. Wu, certainly other genes identified through microarrays in mice J. Powell, C. Sapienza, R. Rozen Analysis and interpretation of data (e.g., statistical analysis, biosta- or humans that could contribute. For example, in Supple- tistics, computational analysis): D. Leclerc, N. Levesque, Y. Cao, J. Powell, mentary Fig. S4, we show 6 genes, in addition to Rdh18 and C. Sapienza, R. Rozen Akr1c13 Writing, review and/or revision of the manuscript: D. Leclerc, N. (already mentioned), with confirmed expression Levesque, Y. Cao, L. Deng, Q. Wu, J. Powell, C. Sapienza, R. Rozen changes due to folate or Mthfr deficiency. Administrative, technical, or material support (i.e., reporting or orga- Our cluster analysis of bisulfite pyrosequencing-based nizing data, constructing databases): C. Sapienza, R. Rozen Study supervision: C. Sapienza, R. Rozen DNA methylation data provides an initial epigenetic signature for distinguishing normal mucosa from controls versus normal mucosa in patients with colorectal Acknowledgments The authors thank Drs. Elin Sigurdson and Andrew Godwin (Chase cancer(Fig.5A).Additionalmethylationmarkerscould Cancer Center) for assistance with colon biopsies and tissue acquisition, improve the power of this diagnostic assay, and some of and Nuala O’Leary (NCBI) for discussions on the Rdh18 gene. the genes discussed above, as well as other genes in our microarrays, could potentially improve the discriminato- Grant Support ry power. However, we cannot exclude the possibility that This research was supported by the Canadian Institutes of Health Research some misclassification in Fig. 5 may have been due to a (grant no. MOP 77596 to R. Rozen) and NIH (T32 CA 103652-05) and the Fels Institute for Cancer Research (to C. Sapienza). This work was also false negative finding during colonoscopy. Clustering of supported by a McGill University Health Centre Research Institute Fellow- controls without polyps versus those with polyps (Fig. ship (to N. Levesque). The Research Institute is supported by a Centres grant 5B) is also of interest, but requires additional markers. from the Fonds de Recherche du Quebec - Sante. The costs of publication of this article were defrayed in part by the We used a systematic human genome-wide methylation payment of page charges. This article must therefore be hereby marked marker discovery study with patients that were not advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate screened for a specific cause of colorectal cancer, this fact. although none of the subjects had a history of familial Received May 30, 2013; revised August 23, 2013; accepted September 3, cancer, colon polyps, or inflammatory bowel disease (7). 2013; published OnlineFirst October 29, 2013.

www.aacrjournals.org Cancer Prev Res; 6(11) November 2013 1179

Leclerc et al.

References 1. Moghaddam AA, Woodward M, Huxley R. Obesity and risk of colo- pathway activity from protein phosphatase 2A-mediated inactivation rectal cancer: a meta-analysis of 31 studies with 70,000 events. in human malignant glioma. Cancer Res 2009;69:2870–7. Cancer Epidemiol Biomarkers Prev 2007;16:2533–47. 21. Westermarck J, Hahn WC. Multiple pathways regulated by the 2. Goel A, Boland CR. Epigenetics of colorectal cancer. Gastroenterol- tumor suppressor PP2A in transformation. Trends Mol Med 2008; ogy 2012;143:1442–60. 14:152–60. 3. Al-Sohaily S, Biankin A, Leong R, Kohonen-Corish M, Warusavitarne J. 22. Takeshita F, Ishii KJ. Intracellular DNA sensors in immunity. Curr Opin Molecular pathways in colorectal cancer. J Gastroenterol Hepatol Immunol 2008;20:383–8. 2012;27:1423–31. 23. Rizzatti EG, Falcao~ RP, Panepucci RA, Proto-Siqueira R, Anselmo- 4. LiX,YaoX,WangY,HuF,WangF,JiangL,etal.MLH1Promoter Lima WT, Okamoto OK, et al. Gene expression profiling of mantle cell methylation frequency in colorectal cancer patients and related lymphoma cells reveals aberrant expression of genes from the PI3K- clinicopathological and molecular features. PLoS ONE 2013;8: AKT, WNT and TGFbeta signalling pathways. Br J Haematol e59064. 2005;130:516–26. 5. Shen L, Kondo Y, Rosner GL, Xiao L, Hernandez NS, Vilaythong J, et al. 24. García-Escudero R, Paramio JM. Gene expression profiling as a tool MGMT promoter methylation and field defect in sporadic colorectal for basic analysis and clinical application of human cancer. Mol cancer. J Natl Cancer Inst 2005;97:1330–8. Carcinog 2008;47:573–9. 6. Milicic A, Harrison LA, Goodlad RA,HardyRG,NicholsonAM, 25. Porte S, Xavier Ruiz F, Gimenez J, Molist I, Alvarez S, Domínguez M, Presz M, et al. Ectopic expression of P-cadherin correlates with et al. Aldo-keto reductases in retinoid metabolism: search for substrate promoter hypomethylation early in colorectal carcinogenesis and specificity and inhibitor selectivity. Chem Biol Interact 2013;202: enhanced intestinal crypt fission in vivo. Cancer Res 2008;68: 186–94. 7760–8. 26. Jeong JY, Jeoung NH, Park KG, Lee IK. Transcriptional regulation of 7. Silviera ML, Smith BP, Powell J, Sapienza C. Epigenetic differences in pyruvate dehydrogenase kinase. Diabetes Metab J 2012;36:328–35. normal colon mucosa of cancer patients suggest altered dietary 27. Goto T, Kim YI, Funakoshi K, Teraminami A, Uemura T, Hirai S, et al. metabolic pathways. Cancer Prev Res 2012;5:374–84. Farnesol, an isoprenoid, improves metabolic abnormalities in mice via 8. Knock E, Deng L, Wu Q, Leclerc D, Wang XL, Rozen R. Low dietary both PPARa-dependent and -independent pathways. Am J Physiol folate initiates Intestinal Tumors in Mice, with altered expression of G2- Endocrinol Metab 2011;301:E1022–32. M Checkpoint Regulators Polo-Like Kinase 1 and cell division cycle 28. Riojas MA, Guo M, Glockner€ SC, Machida EO, Baylin SB, Ahuja N. 25c. Cancer Res 2006;66:10349–56. Methylation-induced silencing of ASC/TMS1, a pro-apoptotic gene, is 9. Ma J, Stampfer MJ, Giovannucci E, Artigas C, Hunter DJ, Fuchs C, a late-stage event in colorectal cancer. Cancer Biol Ther 2007;6:1710– et al. Methylenetetrahydrofolate reductase polymorphism, dietary 6. interactions, and risk of colorectal cancer. Cancer Res 1997;57: 29. Friso S, Udali S, Guarini P, Pellegrini C, Pattini P, Moruzzi S, et al. 1098–102. Global DNA hypomethylation in peripheral blood mononuclear cells as 10. Ulvik A, Vollset SE, Hansen S, Gislefoss R, Jellum E, Ueland PM. a biomarker of cancer risk. Cancer Epidemiol Biomarkers Prev Colorectal cancer and the methylenetetrahydrofolate reductase 2013;22:348–55. 677C ! T and methionine synthase 2756A ! G polymorphisms: a 30. Sung MK, Yeon JY, Park SY, Park JH, Choi MS. Obesity-induced study of 2,168 case-control pairs from the JANUS cohort. Cancer metabolic stresses in breast and colon cancer. Ann N Y Acad Sci Epidemiol Biomarkers Prev 2004;13:2175–80. 2011;1229:61–8. 11. Teng Z, Wang L, Cai S, Yu P, Wang J, Gong J. The 677C>T (rs1801133) 31. Menendez JA, Joven J, Cufí S, Corominas-Faja B, Oliveras-Ferraros C, polymorphism in the MTHFR gene contributes to colorectal cancer Cuyas E, et al. The Warburg effect 2.0: metabolic reprogramming of risk: a meta-analysis based on 71 research studies. PLoS ONE 2013;8: cancer stem cells. Cell Cycle 2013;12:1166–79. e55332. 32. Altucci L, Gronemeyer H. The promise of retinoids to fight against 12. Knock E, Deng L, Wu Q, Lawrance AK, Wang XL, Rozen R. Strain cancer. Nat Rev Cancer 2001;3:181–93. differences in mice highlight the role of DNA damage in Neoplasia 33. D'Ambrosio DN, Clugston RD, Blaner WS. Vitamin A metabolism: an induced by low dietary folate. J Nutr 2008;138:653–8. update. Nutrients 2011;3:63–103. 13. Garcia-Crespo D, Knock E, Jabado N, Rozen R. Intestinal neoplasia 34. Pham DNT, Leclerc D, Levesque N, Deng L, Rozen R. b-Carotene induced by low dietary folate is associated with altered tumor expres- 15,150-monooxygenase 1 and its substrate b-carotene modulate sion profiles and decreased apoptosis in mouse normal intestine. migration and invasion in colorectal carcinoma cells. Am J Clin Nutr J Nutr 2009;139:488–94. 2013;98:413–22. 14. Knock E, Deng L, Krupenko N, Mohan RD, Wu Q, Leclerc D, et al. 35. Bottiglieri T, Arning E, Wasek B, Nunbhakdi-Craig V, Sontag JM, Susceptibility to intestinal tumorigenesis in folate-deficient mice may Sontag E. Acute administration of L-DOPA induces changes in meth- be influenced by variation in one-carbon metabolism and DNA repair. ylation metabolites, reduced protein phosphatase 2A methylation, and J Nutr Biochem 2011;11:1022–9. hyperphosphorylation of Tau protein in mouse brain. J Neurosci 15. Leclerc D, Cao Y, Deng L, Mikael LG, Wu Q, Rozen R. Differential gene 2012;32:9173–81. expression and methylation in the retinoid/PPARA pathway and of 36. Wallace K, Grau MV, Levine AJ, Shen L, Hamdan R, Chen X, et al. tumor suppressors may modify intestinal tumorigenesis induced by Association between folate levels and CpG Island hypermethylation in low folate in mice. Mol Nutr Food Res 2013;57:686–97. normal colorectal mucosa. Cancer Prev Res 2010;3:1552–64. 16. Barrett T, Wilhite SE, Ledoux P, Evangelista C, Kim IF, Tomashevsky 37. Hahn MA, Wu X, Li AX, Hahn T, Pfeifer GP. Relationship between gene M, et al. NCBI GEO: archive for functional genomics data sets–update. body DNA methylation and intragenic H3K9me3 and H3K36me3 Nucleic Acids Res 2013;41:D991–5. chromatin marks. PLoS ONE 2011;19:e18844. 17. Bottone FG Jr, Moon Y, Kim JS, Alston-Mills B, Ishibashi M, Eling TE. 38. Jjingo D, Conley AB, Yi SV, Lunyak VV, Jordan IK. On the presence The anti-invasive activity of cyclooxygenase inhibitors is regulated by and role of human gene body DNA methylation. Oncotarget 2012;3: the transcription factor ATF3 (activating transcription factor 3). Mol 462–74. Cancer Ther 2005;5:693–703. 39. Pipaon C, Real PJ, Fernandez-Luna JL. Defective binding of transcrip- 18. Huang Y, Zhao Q, Chen GQ. Phospholipid scramblase 1. Sheng Li Xue tional repressor ZEB via DNA methylation contributes to increased Bao 2006;58:501–10. constitutive levels of p73 in Fanconi anemia cells. FEBS Lett 19. Chen W, Possemato R, Campbell KT, Plattner CA, Pallas DC, Hahn 2005;579:4610–4. WC. Identification of specific PP2A complexes involved in human cell 40. Allen IC, TeKippe EM, Woodford RM, Uronis JM, Holl EK, Rogers AB, transformation. Cancer Cell 2004;5:127–36. et al. The NLRP3 inflammasome functions as a negative regulator of 20. Puustinen P, Junttila MR, Vanhatupa S, Sablina AA, Hector ME, tumorigenesis during colitis-associated cancer. J Exp Med 2010;207: Teittinen K, et al.. PME-1 protects extracellular signal-regulated kinase 1045–56.

1180 Cancer Prev Res; 6(11) November 2013 Cancer Prevention Research

Epigenetics in Mouse/Human Preneoplastic Intestine

41. Wen H, Ting JP, O'Neill LA. A role for the NLRP3 inflammasome in 45. Fujiwara S, Kawano Y, Yuki H, Okuno Y, Nosaka K, Mitsuya H, et al. metabolic diseases-did Warburg miss inflammation? Nat Immunol PDK1 inhibition is a novel therapeutic target in multiple myeloma. Br J 2012;13:352–7. Cancer 2013;108:170–8. 42. Hong SH, Lee JE, Jeong JJ, Hwang SJ, Bae SN, Choi JY, et al. Small 46. Mencarelli A, Cipriani S, Renga B, D'Amore C, Palladino G, Distrutti E, proline-rich protein 2 family is a cluster of genes induced by estrogenic et al. FXR activation improves myocardial fatty acid metabolism in a compounds through nuclear estrogen receptors in the mouse uterus. rodent model of obesity-driven cardiotoxicity. Nutr Metab Cardiovasc Reprod Toxicol 2010;30:469–76. Dis 2013;23:94–101. 43. Song H, Zhang B, Watson MA, Humphrey PA, Lim H, Milbrandt J. Loss 47. Mikeska T, Bock C, El-Maarri O, Hubner€ A, Ehrentraut D, Schramm J, of Nkx3.1 leads to the activation of discrete downstream target genes et al. Optimization of quantitative MGMT promoter methylation anal- during prostate tumorigenesis. Oncogene 2009;28:3307–19. ysis using pyrosequencing and combined bisulfite restriction analysis. 44. Yamashita S, Tsujino Y, Moriguchi K, Tatematsu M, Ushijima T. J Mol Diagn 2007;9:368–80. Chemical genomic screening for methylation-silenced genes in gastric 48. Jain S, Wojdacz TK, Su YH. Challenges for the application of DNA cancer cell lines using 5-aza-20-deoxycytidine treatment and oligonu- methylation biomarkers in molecular diagnostic testing for cancer. cleotide microarray. Cancer Sci 2006;97:64–71. Expert Rev Mol Diagn 2013;13:283–94.

www.aacrjournals.org Cancer Prev Res; 6(11) November 2013 1181

Genes with Aberrant Expression in Murine Preneoplastic Intestine Show Epigenetic and Expression Changes in Normal Mucosa of Colon Cancer Patients

Daniel Leclerc, Nancy Lévesque, Yuanhang Cao, et al.

Cancer Prev Res 2013;6:1171-1181. Published OnlineFirst October 29, 2013.

Updated version Access the most recent version of this article at: doi:10.1158/1940-6207.CAPR-13-0198

Supplementary Access the most recent supplemental material at: Material http://cancerpreventionresearch.aacrjournals.org/content/suppl/2014/02/14/1940-6207.CAPR-13-0198.DC 1

Cited articles This article cites 48 articles, 14 of which you can access for free at: http://cancerpreventionresearch.aacrjournals.org/content/6/11/1171.full#ref-list-1

Citing articles This article has been cited by 1 HighWire-hosted articles. Access the articles at: http://cancerpreventionresearch.aacrjournals.org/content/6/11/1171.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerpreventionresearch.aacrjournals.org/content/6/11/1171. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.