Introduction

The mutated in colorectal cancer (MCC) gene was discovered in 1991 through its linkage to the region containing the susceptibility locus for familial adenomatous polyposis on chromosome 5q, which was subsequently shown to be the adenomatous polyposis coli (APC) gene (Groden et al., 1991; Kinzler et al., 1991; Nishisho et al., 1991). The APC gene is altered somatically in ∼60–80% of sporadic colorectal cancers and this is considered an important early event in carcinogenesis. In contrast, little is known about the role of MCC in the pathogenesis of colorectal cancer. Early reports indicated that loss of heterozygosity (LOH) at the MCC locus is common, that is, ∼42% in colorectal cancers (Ashton-Rickardt et al., 1991), whereas somatic mutation of MCC is relatively rare, 3–7% (Nishisho et al., 1991). More recent studies have shown that the wild-type MCC protein may have a functional role in important cellular processes relevant in carcinogenesis, such as cell cycle regulation and the nuclear factor-κB (NF-κB) pathway (Matsumine et al., 1996; Bouwmeester et al., 2004). The activation and translocation of NF-κB to the nucleus is part of a signal-transduction pathway, which forms a basis for several physiologic and pathologic processes including cancer. These findings warrant further investigation of MCC as a potential tumor suppressor gene.

The molecular alterations involved in the progression from adenoma to colorectal carcinoma are now well established (Vogelstein et al., 1988). LOH in chromosome 5q and mutations in the APC locus are considered among the earliest changes in adenoma formation. Hypermethylation of gene promoters is an important alternative mechanism by which tumor suppressor genes are silenced during colon carcinogenesis. A subset of sporadic colorectal cancers displays concordant methylation of multiple promoters and this is known as the CpG island methylator phenotype (CIMP) (Toyota et al., 1999), although the definition of CIMP is controversial. CIMP comprises 20–30% of sporadic colon cancers and includes, but is not confined to, tumors with high microsatellite instability (MSI-H), which are largely caused by hypermethylation of the MLH1 mismatch repair gene (Samowitz et al., 2005a; Weisenberger et al., 2006). Hypermethylation of gene promoters is seen early in colon carcinogenesis and is found in a subset of both adenomas and hyperplastic polyps (HP) (Rashid et al., 2001; O'Brien et al., 2004). However, the MSI-H phenotype is more common in HP than in sporadic adenomas. HP were previously considered benign lesions with no malignant potential. The finding of MSI-H in a subset of these lesions (Jass et al., 2000; Hawkins and Ward, 2001) led to the hypothesis that they represent an alternative pathway of colon carcinogenesis, called ‘the serrated pathway’. This pathway also features CIMP and implicates other types of serrated polyps including ‘serrated adenomas (SA)’, ‘mixed polyps’ and ‘sessile serrated adenomas (SSA)’ (Jass, 2005). A subset of serrated polyps are also characterized by the V600E mutation of the BRAF oncogene, which is found in 5–15% of colorectal cancers and is associated with MSI-H and CIMP but inversely correlated with KRAS mutations (Kambara et al., 2004).

MCC is a highly conserved cytoplasmic protein with orthologs (>95% identity) across many species. It is expressed in the surface epithelium of the colon and other well-differentiated cells (Senda et al., 1999). However, its loss of expression in colorectal cancer has not been studied previously. Given the importance of the chromosome 5 region in colorectal cancer and the emerging functional evidence relating MCC to pathways that are relevant in carcinogenesis, we set out to establish whether the MCC gene could be inactivated through promoter hypermethylation in colorectal cancer. We also wanted to determine any relationships of this feature with other molecular and clinicopathologic variables.

Results

Characterization of CpG island methylation in the MCC promoter

A 514-bp section of the MCC promoter was obtained from the Ensembl Genome Browser and sequenced from 11 bisulfite-treated colon cancer cell lines. The sequencing revealed extensive methylation (48 CpG sites) in the HCT-15, HT-29, KM12SM, LISP-1 and LS411N colon cancer cell lines and partial methylation in the SW480 line. Most of the CpG sites in the other cell lines (Caco-2, HCT-116, LIM1215, LoVo, SW620) appeared unmethylated. Methylation-specific PCR (MSP) and quantitative PCR were designed for the area of highest CpG site density upstream of the transcription start site (TSS) (Figure 1). A PCR product was amplified only in the methylated cell lines (Figure 2a) and the two sets of primers produced the same results. The primers specific for the unmethylated promoter were amplified only in Caco-2, HCT 116, LIM1215, LoVo, SW620 and SW480 (Figure 2a).

Figure 1
figure 1

Schematic representation of the MCC promoter region and CpG island with the TSS 220 bp upstream from codon 1. The vertical bars indicate the position of each CpG site. Six CpG sites located in the middle of the CpG island were resistant to bisulfite modification (indicated by a dotted line). MSP and qMSP tests were designed upstream from the TSS in the region of highest CpG density.

Full size image
Figure 2
figure 2

Correspondence of MCC gene promoter hypermethylation with loss of RNA or protein expression in cell lines (a and b) and primary colorectal cancers (d and e). (a and d) MSP using primers specific for the methylated (M, 99 bp) and unmethylated (U, 94 bp) MCC promoter and control PCR using MYOD1 primers in bisulfite-treated DNA samples. (b and c) cDNA amplification from colorectal cancer cell lines with primers for MCC and housekeeping genes ALAS1 and GAPDH. Addition of Aza to the control (Con) medium reversed MCC expression in HCT-15, KM12SM and LS411N cell lines, but not in Caco-2 cells, which are unmethylated. (e) Western blotting of primary colorectal cancer (C) and matching normal (N) tissue specimens. Loss or reduced MCC protein expression is evident in patients (P) 3, 4, 6 and in the cell line (CL) HCT-15.

Full size image

Reverse transcription (RT)–PCR was carried out for 10 of the 11 cell lines, to correlate MCC promoter methylation with loss of gene expression. The predicted 265-bp fragment was consistently amplified in the unmethylated cell lines HCT-116, LIM1215, SW620, and in the partially methylated SW480, but not in Caco-2, which may have a different MCC defect (Figure 2b). The fully methylated cell lines HCT-15, HT-29, KM12SM, LISP-1 and LS411N had no detectable MCC expression. Expression of MCC mRNA was restored with the demethylation agent 5-aza-2′-deoxycytidine (Aza) treatment in HCT-15, KM12SM and LS411N but not in the Caco-2 line (Figure 2c). Western blotting of 10 tumor biopsies revealed loss or clearly reduced MCC protein expression in at least three strongly methylated tumors compared with matching normal tissue (Figure 2d and e). Another two methylated specimens were less conclusive possibly because of significant contamination by normal cells, including stromal and inflammatory cells.

Correlation of MCC methylation with clinicopathologic variables

MCC promoter methylation was first assessed using the MSP in 103 clinicopathologic stage C colon cancers. MCC methylation was detected in 52% of the cancers but was not found in the matched normal tissue from the same patients. Frequency of CDKN2A methylation was 33% and there was a highly significant correlation between the two methylated genes (P=0.0001) (Table 1). MCC methylation was also associated with high tumor grade (P=0.0135) and metastasis to four or more lymph nodes (P=0.0260). MCC methylation was not associated with differential overall survival (hazard ratio (HR) 1.2, confidence interval (CI) 0.7–1.8) nor with survival to death due to colon cancer (HR 1.7, CI 0.9–3.2). There was no association with gender, tumor size, apical node or vein involvement. MCC methylation was more common in proximal than in distal colon cancer (P=0.0011) (Table 1). There was a trend for more frequent MCC methylation in tumors that showed loss of MLH1 protein expression (P=0.0642) or MSI-H (P=0.0930) but this was not statistically significant.

Table 1 Comparison of MCC promoter methylation with CDKN2A promoter methylation, MLH1 protein expression, MSI and other clinicopathologic variables in 103 clinicopathologic stage C colon cancer patients
Full size table

Correlation of MCC methylation with the BRAF mutation and CIMP

In a separate cohort of colorectal cancer patients of known CIMP and BRAF/KRAS mutation status, MCC methylation was detected in 38 out of 84 sporadic cancer specimens (45%) and one out of seven hereditary non-polyposis colon cancer (HNPCC) tumors using quantitative MSP. MCC methylation was detected in all CIMP+ tumors (P<0.0001) and in all 17 tumors with the BRAF V600E mutation (P<0.0001; Table 2). There was no correlation with the presence of KRAS mutation. MSI-H cancers showed MCC methylation more often (P=0.0444). There was no correlation with the stage of cancer but there was a trend for more frequent methylation in proximal compared with distal cancers. APC methylation was detected in 25% of the specimens and did not correlate with MCC methylation (P=0.3191), CIMP (P=0.0858) or BRAF V600E (P=0.3388). Only two out of 20 CIMP+ tumors and two out of 16 tumors with the BRAF mutation showed APC methylation.

Table 2 Comparison of MCC promoter methylation with other molecular variables
Full size table

Hyperplastic polyps and adenomas

In a well-characterized series of 42 precursor lesions, MCC methylation was significantly more common in serrated polyps (20/25; 80%) than in traditional adenomas (AD) (6/17; 35%, P=0.0084), including eight out of nine SSA (Figure 3). APC methylation was less common in serrated polyps (4/18; 22%) than in AD (9/16; 56%) but this was not statistically significant (P=0.0764). There was no association between APC and MCC methylation (P=0.2376).

Figure 3
figure 3

MCC and APC methylation in a series of hyperplastic polyps (HP), serrated adenomas (SA), sessile SA (SSA), a mixed polyp (MP) and traditional adenomas (AD). Black cells indicate presence of methylation, white cells lack of methylation and shaded cells indicate those specimens where no data are available. HP2-3, HP5-6, HP9, SSA1-SSA9 and SA3 are from hyperplastic polyposis patients, the other serrated polyps are sporadic.

Full size image

Discussion

This study has shown for the first time that the MCC gene is frequently methylated in colorectal cancer, which leads to loss of RNA and protein expression of this gene. This raises questions about the biological significance of the MCC defect and the molecular pathways involved. Here, we also provide new evidence that MCC methylation is an early event in colon carcinogenesis and correlates with CIMP but not with APC methylation. This suggests that the two linked chromosome 5 genes are methylated independently and exert independent effects on colorectal cancer phenotype.

APC mutations, methylation and LOH are common in adenomas and are considered important early events in colorectal carcinogenesis. In this study, we detected a clear difference in the spectrum of MCC methylation in early lesions. MCC methylation was more common in serrated polyps than in AD. The series analysed here included nine SSA, three SA and one mixed polyp, of which all but one showed MCC methylation. Advanced serrated polyps have recently emerged as a precursor lesion for the MSI-H subset of colorectal cancers that also display high-frequency methylation of tumor suppressor genes (Kambara et al., 2004; Wynter et al., 2004). This has been extensively studied in the setting of hyperplastic polyposis, where patients present with multiple, large, mostly proximal SSA. However, not all HP show extensive methylation, in particular the common small polyps that are typically present in the distal colon and rectum. Our data support the view that CIMP could be a marker for those HP that have malignant potential (Wynter et al., 2004).

MCC methylation was further analysed in two independent cancer patient cohorts. The stage C cohort included extensive clinicopathologic follow-up data, whereas the second cohort was more suitable for correlation with other gene alterations. The latter cohort was a subset of the patients recently analysed for 195 methylated gene markers, which did not include the MCC gene (Weisenberger et al., 2006). Both cohorts had been analysed for MSI and CDKN2A methylation. Striking positive associations were found with the BRAF V600E mutation, CDKN2A methylation and CIMP. There was no association with APC methylation or with KRAS mutations. MCC methylation was found in all but one specimen with the BRAF mutation, out of a total of 29 cancers and serrated polyps. BRAF and KRAS oncogenes both upregulate the RAS/RAF/ERK pathway with wide-ranging effects on gene expression, cell proliferation and differentiation, cytoskeletal rearrangements and chromatin remodeling. The BRAF V600E mutation also increases NF-κB signaling (Ikenoue et al., 2003). This mutation is found in serrated polyps but not in adenomas (Kambara et al., 2004) and displays a striking correlation with MSI-H and CIMP (Nagasaka et al., 2004). KRAS mutations are found in a wider spectrum of early lesions, but never together with BRAF V600E, and are possibly caused by a defect in the MGMT repair gene (Esteller et al., 2000). Therefore, our findings add to the speculation that there is a causal relationship between the BRAF mutation and one or more genes methylated in CIMP.

There are strong indications that MCC is involved in important cellular functions. The MCC protein becomes highly phosphorylated during the G1- to S-phase transition in vitro and is a potential negative regulator of cell cycle progression (Matsumine et al., 1996). Overexpression of wild-type MCC protein in vitro resulted in a decrease in the number of cells entering the cell cycle S phase. This cell cycle inhibitory effect was completely abolished using mutant forms of the MCC protein (Matsumine et al., 1996). One mutation tested was the same as that initially identified in a tumor from a colorectal cancer patient (Kinzler et al., 1991). Particularly interesting is the finding that MCC appears to stabilize the IκBβ protein that is an essential inhibitor of NF-κB transcription factor activation (Bouwmeester et al., 2004). There is evidence that NF-κB activation is involved in colorectal carcinogenesis (Karin et al., 2002) and that it can be caused by the BRAF V600E mutation in vitro (Ikenoue et al., 2003). The more precise role of an MCC defect in this process can now be tested experimentally.

One novel clinicopathologic association detected here was that MCC methylation was more commonly found in clinicopathologic stage C tumors with four or more involved lymph nodes. This may suggest an effect on tumor metastasis. However, a survival analysis did not reveal any difference between patient groups with or without MCC methylation. One study has demonstrated that CIMP is associated with adverse prognosis, but only in those patients who do not have MSI-H tumors (Ward et al., 2003). Similarly, the BRAF V600E mutation is associated with poor survival but only in MSS colon cancers (Samowitz et al., 2005b). Therefore, it is possible that an adverse effect of gene methylation on survival is difficult to demonstrate because of the survival benefit attributed to the MSI-H phenotype, which comprises a subgroup of CIMP. Detailed molecular analysis of much larger cohorts of patients will be required to further define the outcome of these patient subgroups.

In conclusion, we have characterized a new methylated gene marker for early colorectal carcinogenesis. The finding of a higher frequency of MCC methylation in serrated polyps than in adenomas provides an interesting contrast with the APC gene that is closely linked with MCC but is methylated in a different spectrum of early lesions. Further study is required to define the role of MCC in colorectal carcinogenesis.

Materials and methods

Patients and tumor specimens

Paraffin-embedded tumor and matched normal tissue specimens were analysed from 103 clinicopathologic stage C colon cancer patients from the Concord Hospital. These were previously examined for microsatellite instability, CDKN2A methylation, MGMT and mismatch repair protein expression (Kohonen-Corish et al., 2005). Frozen biopsies of tumor and matching normal tissue were available for protein analysis from an additional 10 cancers from eight patients. We also examined a further cohort of patients from the Royal Brisbane Hospital who were previously analysed for MSI, BRAF and KRAS mutations and CIMP (Kambara et al., 2004; Wynter et al., 2004; Weisenberger et al., 2006), This cohort consisted of 84 sporadic and seven hereditary non-polyposis colon cancers and 42 adenomas and HP. The early lesions were divided into two groups for statistical analysis: group 1 included nine SSA, three SA, one mixed polyp and 12 HP; group 2 consisted of 10 adenomas, two tubular adenomas and five tubulovillous adenomas. The Royal Brisbane Hospital cohort of 91 cancer patients was a subset of the patients analysed in a recent study that redefined the classification of CIMP by using five new methylated gene markers (Weisenberger et al., 2006). If at least three out of five of these genes showed methylation, the cancer was classified as CIMP+ and methylation in 0–2 genes was classified as CIMP−. The study was approved by the South-Western Sydney Area Health Service Human Research Ethics Committee and the Human Research Ethics Committee of the Queensland Institute of Medical Research.

Bisulfite sequencing of the MCC promoter

A 360-kb genomic sequence containing the MCC gene was obtained from the Ensembl Genome Browser (ENSG00000171444; http://www.ensembl.org). This includes a 90.7-kb region upstream of the translation initiation codon. A 522-bp CpG island was identified using the EMBOSS-CpGPlot program (EMBL-European Bioinformatics Institute; http://www.ebi.ac.uk). A TSS is located in this CpG island 220 bp upstream of codon 1 (Figure 1) (Ensembl Transcript ID ENST00000302475).

DNA from cancer cell lines was bisulfite-treated using standard methods (Herman et al., 1996). The MCC promoter region was amplified using primers MCC-promF 5′-GGTTAGTAGTTAGATAGTTGT-3′, and MCC-promR 5′-TACTTAATCCCTTCTACCAC-3′ (Figure 1). PCR conditions were 95°C (30 s), 53°C (45 s) and 72°C (90 s) for 40 cycles. The 514-bp fragment was sequenced using BigDye on an ABI PRISM 3700 DNA Analyzer (Applied Biosystems, Foster City, CA, USA).

Methylation-specific PCR

The MCC promoter was amplified from bisulfite-treated tumor and cell line DNA specimens using methylation-specific primers MCC-metF1 5′-TATTGTTTCGGAACGGGGCGT-3′ and MCC-metR1 5′-CAAAAAACTCGATAACGCGACG-3′ (Figure 1). Primers specific for the unmethylated promoter were MCC-unmetF1 5′-GGTATTGTTTTGGAATGGGGTG-3′ and MCC-unmetR1 5′-CTCAATAACACAACACACTCAC-3′. PCR conditions were 95°C (30 s), 61°C (45 s) and 72°C (30 s) for 35 cycles. CDKN2A promoter methylation was analysed as previously described (Herman et al., 1996; Kohonen-Corish et al., 2005). The MYOD1 primers, which do not contain any CpG nucleotide sequences, were used as an internal reference PCR to ensure integrity of each DNA specimen (Eads et al., 1999, 2000). PCR conditions were 95°C (45 s), 57°C (45 s) and 72°C (60 s) for 38 cycles. All PCR reactions in this study were carried out using AmpliTaq Gold (Applied Biosystems, Foster City, CA, USA) in a DNA engine DYAD (MJ Research, Waltham, MA, USA). The cycling conditions included an initial denaturation step at 95°C for 12 min and a final elongation at 72°C for 7 min.

Methylation-specific quantitative PCR

Methylation-specific primers for the real-time PCR assay were MCC-metF2 5′-TTTCGTCGTTGTCGTAGTTG-3′ and MCC-metR2 5′-TTCACGAACGAACCATTACTA-3′ (Figure 1). The methylated amplicon-specific fluorogenic hybridization probe was 6FAM5′-CGCGTCGCGTTATCGAGTT-3′BHQ-1. The PCR was carried out in duplicate for each tumor specimen using the RealMasterMix Probe ROX (Eppendorf, Hamburg, Germany) with 4 mM Mg2+ in a Rotorgene 3000 (Corbett Research, Sydney, Australia). PCR conditions were 95°C (15 s) and 60°C (60 s) for 50 cycles, with an initial denaturation at 95°C for 2 min. APC methylation and the internal reference gene MYOD1 were analysed following the previously described MethyLight protocol (Eads et al., 1999, 2000). Following this method, the threshold cycle C(T) value for each test and reference PCR is compared to a standard curve to obtain the quantitation value for each reaction. The standard curve is obtained using serial dilutions of a fully methylated control sample with known concentration. The relative quantitation of methylation in a tumor specimen is then determined by calculating the ratio of the test gene value (MCC or APC) to the reference gene value (MYOD1) for the same specimen (Eads et al., 1999, 2000). Limit of detection of each assay was 1.2 ng/μl.

cDNA and protein analysis

For the Aza treatment colon cancer cells were freshly seeded and allowed to grow overnight. The culture medium was then replaced with fresh medium containing 1 μ M of Aza (Sigma-Aldrich Corporation, St Louis, MO, USA). Cells were allowed to grow for 72 h, with changing of Aza-containing medium every 24 h, and then harvested. A cell viability of >70% was retained after 72 h of treatment. RNA was extracted from cell lines using an RNeasy Mini kit (Qiagen, Hilden, Germany). cDNA was prepared with the Expand Reverse Transcriptase kit (Roche, Mannheim, Germany). A 265-bp fragment was amplified from exons 1 and 2 using primers MCC-cDNAF1 5′-CTGGGTGAAAATGGCTGTCT-3′ and MCC-cDNAR1 5′-GTTCCCTCTCTGTTTGCTGG-3′. PCR conditions were 95°C (30 s), 60°C (30 s) and 72°C (30 s) for 35 cycles.

Whole biopsies of 10 primary tumors were frozen after surgery for protein extraction. Protein (20 μg) from total tissue lysate was separated using a 10% SDS–polyacrylamide gel at 100 V for 1.5 h, and transferred to PVDF membrane for 1 h at 120 V. Membranes were blocked in 5% skim milk powder/Tris-buffered-saline-Triton (0.05%), and incubated with primary antibody, 1:2000 α-MCC (Pharmingen, San Diego, CA, USA) or 1:48 000 α-GAPDH (Ambion, Austin, TX, USA), overnight at 4°C. Horseradish peroxidase-conjugated anti-mouse IgG (Amersham, Piscataway, NJ, USA) was used as the secondary antibody. Bands were visualized using ECL.

Survival analysis and statistics

The χ2 test was used to evaluate the significance of differences in contingency tables. Patients were followed annually until death or December 2004. Overall survival was measured from the date of resection to the date of death owing to any cause, censored patients being those alive at the close of the study or lost to follow-up. For death owing to colon cancer, censored patients were those alive at the close of the study, lost to follow-up or deceased owing to other causes. Proportional hazards regression was used to determine HR and their 95% CI. The level for statistical significance was set at ⩽0.05.