Correlation of B-Catenin Localization with Cyclooxygenase-2 Expression and CpG Island Methylator Phenotype (CIMP) in Colorectal Cancer 1

Transcription

1 RESEARCH ARTICLE Neoplasia. Vol. 9, No. 7, July 2007, pp Correlation of B-Catenin Localization with Cyclooxygenase-2 Expression and CpG Island Methylator Phenotype (CIMP) in Colorectal Cancer 1 Takako Kawasaki*,2, Katsuhiko Nosho y,2, Mutsuko Ohnishi*, Yuko Suemoto*, Gregory J. Kirkner z, Reiko Dehari, Jeffrey A. Meyerhardt*,z,b, Charles S. Fuchs*,z,b and Shuji Ogino*,#, ** *Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA 02115, USA; y Department of Internal Medicine, Sapporo Medical University, Sapporo, Japan; z Department of Medicine, Brigham and Women s Hospital, Boston, MA 02115, USA; Department of Pathology, Kanagawa Cancer Center, Kanagawa, Japan; b Department of Medicine, Harvard Medical School, Boston, MA 02115, USA; # Department of Pathology, Brigham and Women s Hospital, Boston, MA 02115, USA; **Department of Pathology, Harvard Medical School, Boston, MA 02115, USA Abstract The WNT/B-catenin (CTNNB1) pathway is commonly activated in the carcinogenic process. Cross-talks between the WNT and cyclooxygenase-2 (COX-2 or PTGS2)/prostaglandin pathways have been suggested. The relationship between B-catenin activation and microsatellite instability (MSI) in colorectal cancer has been controversial. The CpG island methylator phenotype (CIMP or CIMP-high) with widespread promoter methylation is a distinct epigenetic phenotype in colorectal cancer, which is associated with MSI-high. However, no study has examined the relationship between B-catenin activation and CIMP status. Using 832 population-based colorectal cancer specimens, we assessed B-catenin localization by immunohistochemistry. We quantified DNA methylation in eight CIMP-specific promoters [CACNA1G, CDKN2A (p16), CRABP1, IGF2, MLH1, NEUROG1, RUNX3, and SOCS1] by real-time polymerase chain reaction (MethyLight). MSI-high, CIMP-high, and BRAF mutation were associated inversely with cytoplasmic and nuclear B-catenin expressions (i.e., B-catenin activation) and associated positively with membrane expression. The inverse relation between B-catenin activation and CIMP was independent of MSI. COX-2 overexpression correlated with cytoplasmic B-catenin expression (even after tumors were stratified by CIMP status), but did not correlate significantly with nuclear or membrane expression. In conclusion, B-catenin activation is inversely associated with CIMP-high independent of MSI status. Cytoplasmic B-catenin is associated with COX-2 overexpression, supporting the role of cytoplasmic B-catenin in stabilizing PTGS2 (COX-2) mrna. Neoplasia (2007) 9, Keywords: Colon cancer, CTNNB1, CpG island methylator phenotype, PTGS2, microsatellite instability. Introduction Transcriptional inactivation by cytosine methylation at promoter CpG islands of tumor-suppressor genes is an important mechanism in human carcinogenesis [1]. A number of tumor-suppressor genes have been shown to be silenced by promoter methylation in colorectal cancers [1]. In fact, a subset of colorectal cancers has been shown to exhibit promoter methylation in multiple genes, which is referred to as the CpG island methylator phenotype (CIMP) [2]. CIMP-positive colorectal tumors have a distinct clinical, pathological, and molecular profile, such as associations with proximal tumor location, female gender, poor differentiation, microsatellite instability (MSI), and high BRAF and low TP53 mutation rates [3 6]. Promoter CpG island methylation has been shown to occur early in colorectal carcinogenesis [7]. WNT genes and products (derived from Wingless and INT ) constitute the WNT signaling pathway, which controls virtually every developmental decision making [8]. The central player in the WNT pathway is b-catenin [CTNNB1, the Human Genome Organization approved official gene symbol; catenin (cadherin-associated protein), b1], whose stability is regulated by APC complex [8]. When WNT receptors are inactive, b-catenin localizes with the membrane protein E-cadherin (CDH1), and kinases in the APC complex phosphorylate cytoplasmic b-catenin for its rapid degradation. Abbreviations and HGNC-approved Official Gene Symbols: APC, adenomatous polyposis coli; CACNA1G, calcium channel, voltage-dependent, T type a-1g subunit; CDKN2A, cyclindependent kinase inhibitor 2A, p16/ink4a; CIMP, CpG island methylator phenotype; COX-2, cyclooxygenase-2 (PTGS2, the official symbol); CRABP1, cellular retinoic acid binding protein 1; CTNNB1, catenin (cadherin-associated protein), b1 (b-catenin); IGF2, insulin-like growth factor 2; MSI, microsatellite instability; MSS, microsatellite stable; NEUROG1, neurogenin 1; PMR, percentage of methylated reference (degree of methylation); PTGS2, prostaglandin endoperoxide synthase 2 (cyclooxygenase-2); RUNX3, runt-related transcription factor 3; SOCS1, suppressor of cytokine signaling 1;TGFBR2, transforming growth factor-b receptor, type 2; WNT, wingless-type MMTV integration site family (wingless-int) Address all correspondence to: Shuji Ogino, MD, PhD, Department of Pathology, Brigham and Women s Hospital, Dana-Farber Cancer Institute and Dana-Farber/Harvard Cancer Center, 44 Binney Street, Room JF215C, Boston, MA shuji_ogino@dfci.harvard.edu 1 This work was supported by National Institute of Health grants P01 CA87969 and P01 CA55075 and, in part, by Bennett Family Fund and a grant from the Entertainment Industry Foundation (EIF) through the EIF National Colorectal Cancer Research Alliance. 2 Takako Kawasaki and Katsuhiko Nosho contributed equally to this work. Received 10 April 2007; Revised 21 May 2007; Accepted 21 May Copyright D 2007 Neoplasia Press, Inc. All rights reserved /07/$25.00 DOI /neo.07334

2 570 B-Catenin, COX-2 and CIMP in Colorectal Cancer Kawasaki et al. When WNT receptors are activated, kinases in the APC complex are inhibited, leading to accumulation of cytoplasmic b-catenin and its translocation to the nucleus, where it facilitates the transcription of various target genes [8]. Other pathways implicated in cross-talks with the WNT pathway include the transforming growth factor-b (TGF-b) pathway[9]and the cyclooxygenase-2 [COX-2 or PTGS2 (prostaglandin endoperoxide synthase 2)]/prostaglandin pathway [10 13]. Activation of WNT/b-catenin signaling has been implicated in various human malignancies, and both APC and CTNNB1 have been shown to be targets of genetic/epigenetic alterations in colorectal cancer [14 20]. There are conflicting data regarding correlations of b-catenin activation with MSI status in colorectal cancer [19,21 27]. Moreover, no study has comprehensively examined the relationship between b-catenin and CIMP in colorectal cancer. Molecular correlates with b-catenin activation are important in understanding carcinogenic mechanisms in various molecular subtypes of colorectal cancer. In this study, using quantitative DNA methylation analysis (MethyLight technology) and a large number of populationbased colorectal cancer samples, we examined b-catenin expression in relation to various clinicopathological and molecular features, especially COX-2 expression and combined MSI and CIMP statuses. We have found an inverse correlation between b-catenin activation and CIMP, and a positive correlation between cytoplasmic b-catenin expression and COX-2 overexpression. Our results indicate that one should examine CIMP status when analyzing WNT/b-catenin pathway activation in relation to clinical and/or other molecular variables because CIMP status reflects global epigenomic status in tumor cells and may be a confounding factor. Materials and Methods Study Group We used the databases of two large prospective cohort studies: the Nurses Health Study (n = 121,700 women followed since 1976) [28] and the Health Professional Follow-up Study (n = 51,500 men followed since 1986) [29]. Informed consent was obtained from all participants before inclusion into cohorts. We excluded from analyses all cohort participants with cancer (except for nonmelanoma skin cancer) at the time of study entry. A subset of cohort participants developed colorectal cancers during prospective follow-up. Thus, these colorectal cancers represented population-based relatively unbiased samples (compared to retrospective or single hospital based samples). Previous studies on Nurses Health Study and Health Professional Follow-up Study have described baseline characteristics of cohort participants and incident colorectal cancer cases, and they have confirmed that our colorectal cancer cases were representative as a population-based sample [28,29]. Followup of these cohorts is ongoing, and survival data have not yet been made available. We collected paraffin-embedded tissue blocks from hospitals where cohort participants with colorectal cancers had undergone resections of primary tumors. We excluded cases if adequate paraffin-embedded tumor tissue was not available at the time of the study. As a result, 832 colorectal cancer cases (365 from men s cohort and 467 from women s cohort) were included. Among our cohort studies, there was no significant difference in demographic features between cases with available tissue and those without available tissue [30]. Many of the cases have been previously characterized for CIMP, MSI, KRAS, and BRAF statuses [5,31]. However, no tumor has been examined for b-catenin expression in our previous studies. Tissue collection and analyses were approved by the Institutional Review Boards of the Dana-Farber Cancer Institute and the Brigham and Women s Hospital. Histopathological Evaluation Hematoxylin and eosin stained tissue sections were examined under a light microscope by one of the investigators (S.O.) blinded from clinical and other laboratory data. Tumors were classified into well/moderately differentiated (< 50% solid areas) and poorly differentiated tumors (z 50% solid areas). In addition, the extent and the type of mucinous component in each tumor were evaluated, and tumors were classified into five categories: 1) tumors with no mucinous or signet ring cell component (nonmucinous tumors); 2) tumors with 1% to 49% mucinous component but no signet ring cells; 3) tumors with z 50% mucinous component but no signet ring cells; 4) tumors with 1% to 49% signet ring cell component; and 5) tumors with z 50% signet ring cell component. Genomic DNA Extraction and Whole Genome Amplification (WGA) Genomic DNA was extracted from dissected tumor tissue sections using QIAmp DNA Mini Kit (Qiagen, Valencia, CA), as previously described [32]. Normal DNA was obtained from colonic tissues at resection margins. WGA of genomic DNA was performed by polymerase chain reaction (PCR) using random 15-mer primers [32] for subsequent MSI analysis and KRAS and BRAF sequencing. Previous studies by us and others have shown that WGA did not significantly affect KRAS mutation detection or microsatellite analysis [32,33]. MSI Analysis and Loss of Heterozygosity (LOH) for APC Methods used to analyze MSI status and mutation in TGFBR2 (the TGF-b receptor type 2 gene) have been previously described [34,35]. In addition to the recommended MSI panel consisting of D2S123, D5S346, D17S250, BAT25, and BAT26 [36], we also used BAT40, D18S55, D18S56, D18S67, and D18S487 (i.e., a 10-marker panel) [34]. A high degree of MSI (MSI-high) was defined as the presence of instability in z 30% of markers. A low degree of MSI (MSIlow) was defined as the presence of instability in < 30% of markers, and microsatellite stable (MSS) tumors were defined as tumors without an unstable marker. LOH positivity at the APC (D5S346) locus was defined as a z 40% reduction in one of two allele peaks in two duplicated runs in tumor DNA relative to normal DNA.

3 B-Catenin, COX-2 and CIMP in Colorectal Cancer Kawasaki et al. 571 Sequencing of KRAS and BRAF Methods of PCR and sequencing targeted for KRAS codons 12 and 13 and BRAF codon 600 have been previously described [32,37]. Pyrosequecing was performed using the PSQ96 HS System (Biotage AB and Biosystems, Uppsala, Sweden), according to the manufacturer s instructions. Real-Time PCR (MethyLight) for Quantitative DNA Methylation Analysis Sodium bisulfite treatment on genomic DNA was performed as previously described [38]. Real-time PCR to measure DNA methylation (MethyLight) was performed as previously described [39,40]. Using ABI 7300 (Applied Biosystems, Foster City, CA) for quantitative real-time PCR, we used eight CIMP-specific promoters [CACNA1G (calcium channel, voltage-dependent, T type a-1g subunit); CDKN2A (cyclin-dependent kinase inhibitor 2A, p16/ink4a); CRABP1 (cellular retinoic acid binding protein 1); IGF2 (insulin-like growth factor 2); MLH1; NEUROG1 (neurogenin 1); RUNX3 (runt-related transcription factor); and SOCS1 (suppressor of cytokine signaling 1)] [5,6] as a CIMP diagnostic panel [31]. COL2A1 (the collagen 2A1 gene) was used to normalize for the amount of input bisulfite-converted DNA [38,40]. Primers and probes had been previously described [6]. The percentage of methylated reference (PMR; i.e., degree of methylation) at a specific locus was calculated by dividing the GENE:COL2A1 ratio of template amounts in a sample by the GENE:COL2A1 ratio of template amounts in SssI-treated human genomic DNA (presumably fully methylated) and multiplying this value by 100 [39]. A PMR cutoff value of 4 was based on previously validated data [5,38]. We set a PMR cutoff value of 6 for CRABP1 and IGF2 based on PMR distribution. The precision and performance characteristics of bisulfite conversion and subsequent MethyLight assays have been previously evaluated, and such assays have been validated [38]. CIMP-high was defined as the presence of z 6 of 8 methylated promoters; CIMP-low was defined as the presence of 1 to 5 of 8 methylated promoters; and CIMP-0 was defined as the absence (0 of 8) of methylated promoters, according to previously established criteria [31]. Tissue Microarrays (TMAs) and Immunohistochemistry for COX-2 and p53 TMAs were constructed as previously described [41], using Automated Arrayer (Beecher Instruments, Sun Prairie, WI). We examined two to four tumor tissue cores for each marker. A previous validation study has shown that examining two TMA cores can yield results comparable to those when examining whole tissue sections in > 95% of cases [42]. We examined whole tissue sections for cases in which no tissue block was available for TMA construction or for cases in which results were equivocal in TMAs. Immunohistochemistry for p53 and COX-2 was performed as previously described [30,43]. p53 positivity was defined as z 50% of tumor cells with unequivocal strong nuclear staining. COX-2 overexpression was recorded as positive, weak, or negative compared to normal colonic mucosa. Appropriate positive and negative controls were included in each run of immunohistochemistry. All immunohistochemically stained slides were interpreted by one of the investigators (S.O.) blinded from any other clinical and laboratory data. A random sample of 108 cases was examined for COX-2 expression by a second observer (R.D.), and the concordance between the two observers was 0.92 (j coefficient = 0.62) [30]. Immunohistochemistry for -Catenin For b-catenin immunohistochemistry, antigen retrieval was performed; deparaffinized tissue sections in citrate buffer (BioGenex, San Ramon, CA) were treated with microwave in a pressure cooker for 15 minutes. Tissue sections were incubated with 3% H 2 O 2 (15 minutes) to block endogenous peroxidase, with 10% normal goat serum (Vector Laboratories, Burlingame, CA) in phosphate-buffered saline (10 minutes), and with serum-free protein block (10 minutes; DAKO, Carpinteria, CA). Primary antibody against b-catenin (clone 14; 1:400 dilution; BD Transduction Laboratories, Franklin Lakes, NJ) was applied for 1 hour at room temperature. Secondary antibody (20 minutes; BioGenex) and then streptavidin peroxidase conjugate (20 minutes; BioGenex) were applied. Sections were visualized by diaminobenzidine (2 minutes) and methyl green counterstain. Normal colonic epithelial cells served as internal positive controls with membrane staining (Figure 1). Cytoplasmic, nuclear, and membrane expressions were recorded separately as either no expression, weak expression, or moderate/strong expression. Positivity in each compartment (cytoplasm, nucleus, or membrane) was defined as moderate/strong expression in that compartment. We also calculated b-catenin activation score as the sum of nuclear score (+2 = positive expression; +1 = weak expression;0 = no expression), cytoplasmic score (+2 = positive expression; +1 = weak expression; 0 = no expression), and membrane score (0 = positive membrane expression; +1 = negative membrane expression), as previously described by Jass et al. [26]. Appropriate positive and negative controls were included in each run of immunohistochemistry. All immunohistochemically stained slides were interpreted by one of the investigators (K.N.) blinded from any other clinical and laboratory data. A second observer (S.O.) examined 402 cases, and j coefficients for agreement in these 402 cases were as follows: 0.52 for cytoplasmic positivity, 0.63 for nuclear positivity, 0.57 for membrane positivity, and 0.65 for b-catenin scores z 4 (all P <.0001), indicating overall moderate to substantial agreement. Statistical Analysis In statistical analysis, chi-square test (or Fisher s exact test when the number in any category was < 10) was performed for categorical data, and j coefficients were calculated to determine the degree of agreement between two observers, using SAS program, Version 9.1 (SAS Institute, Cary, NC). All P values were two-sided, and statistical significance was set at P V.05.

4 572 B-Catenin, COX-2 and CIMP in Colorectal Cancer Kawasaki et al. Figure 1. -Catenin expression in colorectal cancer. (A) Normal colorectal epithelial cells with membrane expression (arrows). (B) Colorectal cancer with membrane expression (arrow). (C) Colorectal cancer with strong cytoplasmic expression (empty arrows). (D) Colorectal cancer with nuclear expression (empty arrowheads). Results -Catenin Expression in Colorectal Cancer We examined b-catenin expression in 832 colorectal cancers by immunohistochemistry (Figure 1). Among the 832 tumors, 336 (40%) showed cytoplasmic positivity, 381 (46%) showed nuclear positivity, and 328 (39%) showed membrane positivity. Cytoplasmic and nuclear expressions were correlated positively with each other (P <.0001) and correlated inversely with membrane expression (P <.0001) (detailed data not shown). Table 1 shows the frequencies of b-catenin expression in colorectal cancers according to clinical or pathological features. Carcinomas with mucinous component, right-sided tumors, and poorly differentiated tumors generally showed lower frequencies of cytoplasmic/nuclear b-catenin expression and higher frequencies of membrane expression, compared to nonmucinous tumors, left-sided tumors, and well/ moderately differentiated tumors, respectively. MSI-High, CIMP-High, and BRAF Mutation Are Inversely Associated with -Catenin Activation Using MethyLight technology, we quantified DNA methylation in a CIMP-specific marker panel of eight promoters (CACNA1G, CDKN2A, CRABP1, IGF2, MLH1, NEUROG1, RUNX3, and SOCS1) [5,6,31]. Among 789 tumors with CIMP status determined, there were 108 (14%) CIMP-high tumors (z 6 of 8 methylated promoters), 301 (38%) CIMPlow tumors (1 to 5 of 8 methylated promoters), and 380 (48%) CIMP-0 tumors (0 of 8 methylated promoters). CIMP-high, MSI-high, and BRAF-mutated tumors showed significantly lower frequencies of cytoplasmic and nuclear expressions and higher frequencies of membrane b-catenin expression (Table 2). We also calculated b-catenin activation scores based on the methods of Jass et al. [26]. Briefly, b-catenin score was the sum (0 5) of nuclear (0, +1, or +2), cytoplasmic (0, +1, or +2), and membrane (0 = positive membrane expression; +1 = negative membrane expression) scores, and was considered to reflect the degree of b-catenin activation [26]. CIMP-high, MSI-high, and BRAF-mutated tumors showed low frequencies of b-catenin scores z 4, indicating low frequencies of b-catenin activation in these tumors (Table 3). Because of the possible cross-talk between the WNT and TGF-b pathways, we examined the relationship between b- catenin expression and TGFBR2 mutation (mononucleotide repeat) in MSI-high tumors; however, there was no significant correlation (data not shown). The Inverse Relation between -Catenin Activation and CIMP-High Is Independent of MSI or BRAF Status To examine the effect of CIMP (or MSI) on b-catenin expression independent of MSI status (or CIMP status), we classified tumors into four subtypes according to combined

5 B-Catenin, COX-2 and CIMP in Colorectal Cancer Kawasaki et al. 573 Table 1. Frequency of b-catenin Expression in Colorectal Cancer. Total n Cytoplasmic Nuclear Membrane (+) [n (%)] P (+) [n (%)] P (+) [n (%)] P All cases (40) 381 (46) 328 (39) Men (41) 168 (46) 123 (34) Referent Women (40) 213 (46) 205 (44).003 Differentiation Well/moderate (42) Referent 349 (47) 283 (38) Referent Poor (28) (35) 33 (51).04 Mucinous/signet ring cell features Nonmucinous carcinoma (44) Referent 283 (51) Referent 195 (35) Referent Carcinoma with mucinous component (no signet ring cell) 1 49% mucinous (33) 58 (34) (45).02 z 50% mucinous (31) (39) 34 (55).003 Carcinoma with signet ring cell component 1 49% signet ring cells (34) 11 (29) (53).03 z 50% signet ring cells 10 4 (40) 5 (50) 3 (30) Location Right (29) Referent 70 (31) Referent 138 (61) Referent Left (excluding rectum) (54) < (60) < (39) <.0001 Rectum (57) < (57) < (29) <.0001 Only significant P values are described. MSI and CIMP statuses (Figure 2). The inverse relation between nuclear/cytoplasmic b-catenin (or high b-catenin score) and CIMP-high, as well as the positive correlation between membrane b-catenin and CIMP-high, persisted even after tumors had been stratified by MSI status. However, the relationship between MSI and b-catenin expression did not persist after tumors had been stratified by CIMP status (Figure 2). These data implied that the relationship between b-catenin and CIMP was independent of MSI status and that the relationship between b-catenin and MSI was mediated by CIMP. CIMP and BRAF are tightly correlated, and BRAF and KRAS mutations are inversely correlated. To examine the effect of CIMP (or BRAF mutation) on b-catenin expression independent of BRAF status (or CIMP status), we stratified tumors according to CIMP, KRAS, and BRAF statuses (Figure 3). The relationship between CIMP and b-catenin appeared to be independent of BRAF mutation, whereas the relationship between BRAF mutation and b-catenin was not present in CIMP-high tumors. CIMP-low/CIMP-0 tumors with BRAF mutation appeared to show b-catenin expression profiles similar to those of CIMP-high tumors. This is perhaps because BRAF-mutated CIMP-low/CIMP-0 tumors were, by themselves, uncommon (only n = 31) and might contain rare CIMP-high tumors misdiagnosed as CIMP-low tumors (false negatives). However, the possible effect of BRAF mutation Table 2. Frequency of b-catenin Expression in Colorectal Cancer with Various Molecular Features. Molecular Feature Total n Cytoplasmic Nuclear Membrane (+) [n (%)] P (+) [n (%)] P (+) [n (%)] P MSI status MSI-high (19) < (26) < (62) <.0001 MSI-low (41) 31 (48) 26 (41) MSS (45) Referent 307 (49) Referent 223 (35) Referent CIMP status CIMP-high (13) < (19) < (74) <.0001 CIMP-low (39) 129 (43) 58 (48) CIMP (49) Referent 218 (57) Referent 42 (33) Referent KRAS mutation (+) (44) 134 (49) 98 (74) <.0001 ( ) (39) 225 (45) 215 (43) Referent BRAF mutation (+) (16) < (23) < (68) <.0001 ( ) (45) Referent 336 (49) Referent 246 (36) Referent p53 expression (+) (45) (53) (33).0006 ( ) (37) Referent 187 (40) Referent 207 (45) Referent COX-2 overexpression Strong (+) (58) (56) 19 (35) (+) (42) 155 (44) 142 (41) ( ) (31) Referent 36 (40) 37 (41) D5S346 (APC) LOH (+) (47) 75 (55) 40 (29) ( ) (45) 123 (47) 98 (37) Only significant P values are described.

6 574 B-Catenin, COX-2 and CIMP in Colorectal Cancer Kawasaki et al. Table 3. b-catenin Activation Score and Various Molecular Features in Colorectal Cancer. Molecular Feature Total n b-catenin Score* z 2 b-catenin Score* z 4 n (%) P n (%) P All cases (67) 352 (42) MSI status MSI-high (44) < (19) <.0001 MSI-low (70) 29 (45) MSS (71) Referent 292 (46) Referent CIMP status CIMP-high (38) < (12) <.0001 CIMP-low (64) 118 (39) CIMP (78) Referent 207 (54) Referent KRAS mutation (+) (70) 131 (45) ( ) (66) 203 (41) BRAF mutation (+) (44) < (13) <.0001 ( ) (70) Referent 321 (46) Referent p53 expression (+) (73) (49).0003 ( ) (63) Referent 171 (37) Referent COX-2 overexpression Strong (+) (78) 32 (58).003 (+) (68) 144 (41) ( ) (68) 30 (33) Referent D5S346 (APC) LOH (+) (75) 71 (52) ( ) (66) 117 (45) Only significant P values are described. *The b-catenin activation score is based on the criteria of Jass et al. [26]. on b-catenin expression in CIMP-low/CIMP-0 tumors could not be completely excluded. COX-2 Overexpression Is Significantly Correlated with Cytoplasmic -Catenin Expression, Even after Stratification By CIMP Status Interestingly, compared to COX-2 negative tumors, COX- 2 strong positive tumors showed a significantly higher frequency of cytoplasmic b-catenin expression (58% vs 31%, P =.001), but no significant relationship was present between Figure 3. Frequencies of -catenin expression and -catenin scores z 4 in colorectal cancer stratified by CIMP, KRAS (K), and BRAF (B) statuses. COX-2 and nuclear (or membrane) b-catenin expression (Table 2). COX-2 strong positive tumors showed a significantly higher frequency of b-catenin scores z 4 than did COX- 2 negative tumors (58% vs 33%, P =.003) (Table 3). COX-2 overexpression was inversely associated with CIMP in our previous analysis [41]. To examine the effect of COX-2 (or CIMP) on b-catenin expression independent of CIMP status (or COX-2 status), we classified tumors according to combined CIMP and COX-2 statuses (Figure 4). It appeared that COX-2 overexpression correlated with cytoplasmic b-catenin expression after stratification by CIMP status and that CIMP-high inversely correlated with cytoplasmic b-catenin expression and high b-catenin score independent of COX-2 status. Nuclear or membrane b-catenin expression did not correlate with COX-2 expression after stratification by CIMP status (data not shown). Because we have previously shown the synergistic effect of COX-2 and p53 expressions (on CIMP and MSI) [41], we examined whether the effect of COX-2 on cytoplasmic b-catenin was modified by p53 status. As shown in Figure 5, there appeared to be a synergistic effect of COX-2 and p53 on cytoplasmic b-catenin expression, and the effect of COX-2 on Figure 2. Frequencies of -catenin expression and -catenin scores z 4 in colorectal cancer stratified by MSI and CIMP statuses. Note that the relationship between CIMP and -catenin is independent of MSI status.

7 B-Catenin, COX-2 and CIMP in Colorectal Cancer Kawasaki et al. 575 Figure 4. Frequency of -catenin expression and -catenin scores z 4 in colorectal cancer stratified by CIMP and COX-2 statuses. cytoplasmic b-catenin appeared to be substantial, whereas there was some (but smaller) effect of p53. Discussion We conducted this study to examine molecular correlates with b-catenin activation in colorectal cancer, using a large number of samples and robust DNA methylation detection methods. Discovering molecular correlates is important in cancer research because it may: 1) provide clues to pathogenesis; 2) propose or support the existence of a new molecular subtype; 3) alert investigators to potential confounding in association studies; and 4) suggest surrogate markers in clinical or research settings. We used quantitative DNA methylation assays (MethyLight), which are essential to reproducibly differentiate low-level methylation from highlevel methylation [38]. Compared to high-level methylation, low-level promoter methylation does not generally silence gene expression and, thus, can be regarded as biologic noise [38]. Our resource of a large number of samples of colorectal cancer (relatively unbiased samples compared to retrospective or single hospital based samples), derived from two large prospective cohorts, has enabled us to precisely estimate the frequency of colorectal cancers with specific molecular features (e.g., nuclear b-catenin expression, CIMP-high, MSI-high, and so on). In particular, we sought to decipher the relationship between b-catenin and MSI/CIMP status. Molecular classification based on MSI and CIMP statuses is increasingly important [25] because MSI and CIMP statuses reflect global genomic and epigenomic aberrations in colorectal cancer cells. Promoter CpG island methylation in tumor-suppressor genes has been shown to be an important mechanism in the development of various human malignancies, including colorectal cancer [1,44 47]. We have demonstrated that nuclear and cytoplasmic b-catenin expressions (i.e., b-catenin activation) are inversely correlated with CIMP-high, independent of MSI status. Previous studies have shown an inverse relationship between b-catenin activation and MSI-high [24 26]. We have further demonstrated that the inverse relation between MSI and b-catenin is indirect and is mediated by CIMP because MSI status is not correlated with b-catenin expression after stratification by CIMP status. Interestingly, APC promoter methylation has been shown to be inversely correlated with features of CIMP in colorectal cancer [15]. Together with our results, it is likely that not only APC promoter methylation but also overall WNT/b-catenin activation is inversely correlated with CIMP. We assessed b-catenin activation status by immunohistochemistry because b-catenin localization reflects the status of b-catenin activation. Regardless of the mechanism of b-catenin activation (i.e., a mutation in APC or other mechanisms), b-catenin accumulates in the cytoplasm and nucleus, then activates the WNT signaling pathway. Although previous studies have shown positive correlations between CTNNB1 (the b-catenin gene) mutations and MSI-high in colorectal cancer [22,23], these studies did not examine the frequencies of cytoplasmic or nuclear b-catenin localization (i.e., the overall frequencies of b-catenin activation). In fact, CTNNB1 mutation constitutes a mechanism of b-catenin/wnt activation in only a minority of colorectal cancers [19]. Thus, b- catenin localization is a reasonable surrogate marker when one evaluates the overall frequency of b-catenin activation. Interestingly, we have shown that COX-2 overexpression is correlated positively with cytoplasmic b-catenin expression, but not significantly with nuclear or membrane expression. A previous study also did not show a significant relationship between COX-2 expression and nuclear b-catenin [48]. A possible link between COX-2 and the WNT signaling pathway has been suggested [10]. COX-2 has been shown to activate b-catenin through prostaglandin E 2 and G protein coupled receptor EP2 [11]. A recent study has also shown that b-catenin stabilizes COX-2 mrna by interacting with AU-rich elements in a 3V untranslated region [12]. Considering these data, it may be possible that COX-2 and b-catenin may form a positive feedback loop. Our data support the role of cytoplasmic b-catenin in stabilizing COX-2 (PTGS2) mrna. That might be the reason why the relationship between COX-2 and b-catenin expression appears to be limited to cytoplasmic b-catenin. We have analyzed b-catenin expression in each cellular compartment (cytoplasm, nucleus, and membrane), and we have also used the b-catenin scoring system based on the Figure 5. Frequency of cytoplasmic -catenin expression in colorectal cancer stratified by p53 and COX-2 statuses.

8 576 B-Catenin, COX-2 and CIMP in Colorectal Cancer Kawasaki et al. methods of Jass et al. [26]. We have shown that separately analyzing cytoplasmic b-catenin expression and nuclear expression is also valuable because associations might be limited to expression in only one of the compartments (as shown by the association between COX-2 and cytoplasmic b-catenin in this study). Thus, molecular correlates with b-catenin expression in each compartment (in particular, cytoplasm and nucleus) can be valuable. In conclusion, b-catenin activation is inversely associated with CIMP-high in colorectal cancer independent of MSI or BRAF status. Cytoplasmic b-catenin localization is correlated positively with COX-2 overexpression. Our results indicate that one should examine CIMP status when analyzing the WNT/ b-catenin pathway in relation to clinical and/or other molecular variables because CIMP status reflects global epigenomic status in tumor cells and may be a confounding factor. The exact mechanisms of these molecular correlates in colorectal cancer need to be elucidated by additional studies. Acknowledgements We deeply thank the Nurses Health Study and Health Professionals Follow-up Study cohort participants who have generously agreed to provide us with biologic specimens and information through responses to questionnaires. We thank Graham Colditz, Walter Willett, and many other staff members who implemented and maintained the cohort studies. We deeply thank Peter Laird, Daniel Weisenberger, and Mihaela Campan for assisting in the development of MethyLight assay. References [1] Laird PW (2005). Cancer epigenetics. Hum Mol Genet 14 (1), R65 R76. [2] Toyota M, Ahuja N, Ohe-Toyota M, Herman JG, Baylin SB, and Issa JP (1999). CpG island methylator phenotype in colorectal cancer. Proc Natl Acad Sci USA 96, [3] Whitehall VL, Wynter CV, Walsh MD, Simms LA, Purdie D, Pandeya N, Young J, Meltzer SJ, Leggett BA, and Jass JR (2002). Morphological and molecular heterogeneity within nonmicrosatellite instability high colorectal cancer. Cancer Res 62, [4] Samowitz W, Albertsen H, Herrick J, Levin TR, Sweeney C, Murtaugh MA, Wolff RK, and Slattery ML (2005). Evaluation of a large, populationbased sample supports a CpG island methylator phenotype in colon cancer. Gastroenterology 129, [5] Ogino S, Cantor M, Kawasaki T, Brahmandam M, Kirkner G, Weisenberger DJ, Campan M, Laird PW, Loda M, and Fuchs CS (2006). CpG island methylator phenotype (CIMP) of colorectal cancer is best characterised by quantitative DNA methylation analysis and prospective cohort studies. Gut 55, [6] Weisenberger DJ, Siegmund KD, Campan M, Young J, Long TI, Faasse MA, Kang GH, Widschwendter M, Weener D, Buchanan D, et al. (2006). CpG island methylator phenotype underlies sporadic microsatellite instability and is tightly associated with BRAF mutation in colorectal cancer. Nat Genet 38, [7] Jass JR (2005). Serrated adenoma of the colorectum and the DNAmethylator phenotype. Nat Clin Pract Oncol 2, [8] Clevers H (2004). Wnt breakers in colon cancer. Cancer Cell 5, 5 6. [9] Attisano L and Labbe E (2004). TGFbeta and Wnt pathway cross-talk. Cancer Metastasis Rev 23, [10] Buchanan FG and DuBois RN (2006). Connecting COX-2 and Wnt in cancer. Cancer Cell 9, 6 8. [11] Castellone MD, Teramoto H, Williams BO, Druey KM, and Gutkind JS (2005). Prostaglandin E 2 promotes colon cancer cell growth through a Gs axin beta-catenin signaling axis. Science 310, [12] Lee HK and Jeong S (2006). Beta-catenin stabilizes cyclooxygenase-2 mrna by interacting with AU-rich elements of 3V-UTR. Nucleic Acids Res 34, [13] Araki Y, Okamura S, Hussain SP, Nagashima M, He P, Shiseki M, Miura K, and Harris CC (2003). Regulation of cyclooxygenase-2 expression by the Wnt and ras pathways. Cancer Res 63, [14] Chen J, Rocken C, Lofton-Day C, Schulz HU, Muller O, Kutzner N, Malfertheiner P, and Ebert MP (2005). Molecular analysis of APC promoter methylation and protein expression in colorectal cancer metastasis. Carcinogenesis 26, [15] Iacopetta B, Grieu F, Li W, Ruszkiewicz A, Caruso M, Moore J, Watanabe G, and Kawakami K (2006). APC gene methylation is inversely correlated with features of the CpG island methylator phenotype in colorectal cancer. Int J Cancer 119, [16] Samowitz WS, Slattery ML, Sweeney C, Herrick J, Wolff RK, and Albertsen H (2007). APC mutations and other genetic and epigenetic changes in colon cancer. Mol Cancer Res 5, [17] Zeng G, Germinaro M, Micsenyi A, Monga NK, Bell A, Sood A, Malhotra V, Sood N, Midda V, Monga DK, et al. (2006). Aberrant Wnt/beta-catenin signaling in pancreatic adenocarcinoma. Neoplasia 8, [18] Tan M, Gallegos JR, Gu Q, Huang Y, Li J, Jin Y, Lu H, and Sun Y (2006). SAG/ROC-SCF beta-trcp E3 ubiquitin ligase promotes pro-caspase-3 degradation as a mechanism of apoptosis protection. Neoplasia 8, [19] Thorstensen L, Lind GE, Lovig T, Diep CB, Meling GI, Rognum TO, and Lothe RA (2005). Genetic and epigenetic changes of components affecting the WNT pathway in colorectal carcinomas stratified by microsatellite instability. Neoplasia 7, [20] Corvinus FM, Orth C, Moriggl R, Tsareva SA, Wagner S, Pfitzner EB, Baus D, Kaufmann R, Huber LA, Zatloukal K, et al. (2005). Persistent STAT3 activation in colon cancer is associated with enhanced cell proliferation and tumor growth. Neoplasia 7, [21] Lovig T, Meling GI, Diep CB, Thorstensen L, Norheim Andersen S, Lothe RA, and Rognum TO (2002). APC and CTNNB1 mutations in a large series of sporadic colorectal carcinomas stratified by the microsatellite instability status. Scand J Gastroenterol 37, [22] Kim IJ, Kang HC, Park JH, Shin Y, Ku JL, Lim SB, Park SY, Jung SY, Kim HK, and Park JG (2003). Development and applications of a betacatenin oligonucleotide microarray: beta-catenin mutations are dominantly found in the proximal colon cancers with microsatellite instability. Clin Cancer Res 9, [23] Shitoh K, Furukawa T, Kojima M, Konishi F, Miyaki M, Tsukamoto T, and Nagai H (2001). Frequent activation of the beta-catenin Tcf signaling pathway in nonfamilial colorectal carcinomas with microsatellite instability. Genes Chromosomes Cancer 30, [24] Wong NA, Morris RG, McCondochie A, Bader S, Jodrell DI, and Harrison DJ (2002). Cyclin D1 overexpression in colorectal carcinoma in vivo is dependent on beta-catenin protein dysregulation, but not k- ras mutation. J Pathol 197, [25] Jass JR (2007). Classification of colorectal cancer based on correlation of clinical, morphological and molecular features. Histopathology 50, [26] Jass JR, Biden KG, Cummings MC, Simms LA, Walsh M, Schoch E, Meltzer SJ, Wright C, Searle J, Young J, et al. (1999). Characterisation of a subtype of colorectal cancer combining features of the suppressor and mild mutator pathways. J Clin Pathol 52, [27] Ricciardiello L, Ceccarelli C, Angiolini G, Pariali M, Chieco P, Paterini P, Biasco G, Martinelli GN, Roda E, and Bazzoli F (2005). High thymidylate synthase expression in colorectal cancer with microsatellite instability: implications for chemotherapeutic strategies. Clin Cancer Res 11, [28] Colditz GA and Hankinson SE (2005). The Nurses Health Study: lifestyle and health among women. Nat Rev Cancer 5, [29] Giovannucci E, Liu Y, Rimm EB, Hollis BW, Fuchs CS, Stampfer MJ, and Willett WC (2006). Prospective study of predictors of vitamin D status and cancer incidence and mortality in men. J Natl Cancer Inst 98, [30] Chan AT, Ogino S, and Fuchs CS (2007). Aspirin use and risk of colorectal cancer according to cyclooxygenase-2 expression. New Engl J Med 356, [31] Ogino S, Kawasaki T, Kirkner GJ, Kraft P, Loda M, and Fuchs CS (2007). Evaluation of markers for CpG island methylator phenotype (CIMP) in colorectal cancer using a large population-based sample. J Mol Diagn 9, [32] Ogino S, Kawasaki T, Brahmandam M, Yan L, Cantor M, Namgyal C, Mino-Kenudson M, Lauwers GY, Loda M, and Fuchs CS (2005).

9 B-Catenin, COX-2 and CIMP in Colorectal Cancer Kawasaki et al. 577 Sensitive sequencing method for KRAS mutation detection by Pyrosequencing. J Mol Diagn 7, [33] Dietmaier W, Hartmann A, Wallinger S, Heinmoller E, Kerner T, Endl E, Jauch KW, Hofstadter F, and Ruschoff J (1999). Multiple mutation analyses in single tumor cells with improved whole genome amplification. Am J Pathol 154, [34] Ogino S, Brahmandam M, Cantor M, Namgyal C, Kawasaki T, Kirkner G, Meyerhardt JA, Loda M, and Fuchs CS (2006). Distinct molecular features of colorectal carcinoma with signet ring cell component and colorectal carcinoma with mucinous component. Mod Pathol 19, [35] Ogino S, Kawasaki T, Ogawa A, Kirkner GJ, Loda M, and Fuchs CS (2007). TGFBR2 mutation is correlated with CpG island methylator phenotype (CIMP) in microsatellite instability high colorectal cancer. Hum Pathol 38, [36] Boland CR, Thibodeau SN, Hamilton SR, Sidransky D, Eshleman JR, Burt RW, Meltzer SJ, Rodriguez-Bigas MA, Fodde R, Ranzani GN, et al. (1998). A National Cancer Institute Workshop on Microsatellite Instability for cancer detection and familial predisposition: development of international criteria for the determination of microsatellite instability in colorectal cancer. Cancer Res 58, [37] Ogino S, Kawasaki T, Kirkner GJ, Loda M, and Fuchs CS (2006). CpG island methylator phenotype low (CIMP-low) in colorectal cancer: possible associations with male sex and KRAS mutations. J Mol Diagn 8, [38] Ogino S, Kawasaki T, Brahmandam M, Cantor M, Kirkner GJ, Spiegelman D, Makrigiorgos GM, Weisenberger DJ, Laird PW, Loda M, et al. (2006). Precision and performance characteristics of bisulfite conversion and real-time PCR (MethyLight) for quantitative DNA methylation analysis. J Mol Diagn 8, [39] Eads CA, Danenberg KD, Kawakami K, Saltz LB, Blake C, Shibata D, Danenberg PV, and Laird PW (2000). MethyLight: a high-throughput assay to measure DNA methylation. Nucleic Acids Res 28, E32. [40] Widschwendter M, Siegmund KD, Muller HM, Fiegl H, Marth C, Muller- Holzner E, Jones PA, and Laird PW (2004). Association of breast cancer DNA methylation profiles with hormone receptor status and response to tamoxifen. Cancer Res 64, [41] Ogino S, Brahmandam M, Kawasaki T, Kirkner GJ, Loda M, and Fuchs CS (2006). Combined analysis of COX-2 and p53 expressions reveals synergistic inverse correlations with microsatellite instability and CpG island methylator phenotype in colorectal cancer. Neoplasia 8, [42] Camp RL, Charette LA, and Rimm DL (2000). Validation of tissue microarray technology in breast carcinoma. Lab Invest 80, [43] Ogino S, Kawasaki T, Kirkner GJ, Yamaji T, Loda M, and Fuchs CS (2007). Loss of nuclear p27 (CDKN1B/KIP1) in colorectal cancer is correlated with microsatellite instability and CIMP. Mod Pathol 20, [44] Kwong J, Lee JY, Wong KK, Zhou X, Wong DT, Lo KW, Welch WR, Berkowitz RS, and Mok SC (2006). Candidate tumor-suppressor gene DLEC1 is frequently downregulated by promoter hypermethylation and histone hypoacetylation in human epithelial ovarian cancer. Neoplasia 8, [45] Liu Y, Lan Q, Siegfried JM, Luketich JD, and Keohavong P (2006). Aberrant promoter methylation of p16 and MGMT genes in lung tumors from smoking and never-smoking lung cancer patients. Neoplasia 8, [46] Kuester D, Dar AA, Moskaluk CC, Krueger S, Meyer F, Hartig R, Stolte M, Malfertheiner P, Lippert H, Roessner A, et al. (2007). Early involvement of death-associated protein kinase promoter hypermethylation in the carcinogenesis of Barrett s esophageal adenocarcinoma and its association with clinical progression. Neoplasia 9, [47] Ebert MP, Mooney SH, Tonnes-Priddy L, Lograsso J, Hoffmann J, Chen J, Rocken C, Schulz HU, Malfertheiner P, and Lofton-Day C (2005). Hypermethylation of the TPEF/HPP1 gene in primary and metastatic colorectal cancers. Neoplasia 7, [48] Tatsuguchi A, Kishida T, Fujimori S, Tanaka S, Gudis K, Shinji S, Furukawa K, Tajiri T, Sugisaki Y, Fukuda Y, et al. (2006). Differential expression of cyclo-oxygenase-2 and nuclear beta-catenin in colorectal cancer tissue. Aliment Pharmacol Ther 24 (Suppl 4),