Epigenetically silenced miR-34b|[sol]|c as a novel faecal-based screening marker for colorectal cancer

MicroRNAs (miRNAs) are a class of small regulatory non-coding RNAs of ~22 nucleotides in length that modulate specific cellular activity post transcriptionally. The miRNAs target the post-translation level of mRNA through a sequence-specific complimentary at the 3′ UTR region of a gene, thus inhibiting gene expression (Lee et al, 1993; Wienholds et al, 2005). The human genome is encoded by at least 3% miRNAs, and up to 30% of human protein encoding genes may be regulated by miRNA modulation (Sassen et al, 2008). The miRNAs are shown to have key roles in normal development and also in differentiation, cell proliferation and apoptosis of cancer cells (Bartel, 2004; He et al, 2005). In the miRBase, thousands of miRNA have been bioinformatically predicted for both prokaryote and eukaryote, and awaiting experimental validation. Each of these miRNA could potentially regulate up to several hundred genes at mRNA post-translation level (Krek et al, 2005). Previous study has shown that more than 50% of the miRNA genes are located in cancer-associated genomic regions or in fragile sites which are hot spots for gene deletion, amplification and mutations (Calin et al, 2004), suggesting major functions for miRNAs in cancer cell biology. Furthermore, miRNAs also have been shown to be involved in a wide variety of human cancers, including colon, pancreatic, breast, prostate, liver and ovarian cancer, suggesting a more negative regulation of cancer cell growth by miRNAs modulation (Michael et al, 2003; Gramantieri et al, 2007; Hurteau et al, 2007; Lee et al, 2007; Shi et al, 2007; Barbarotto et al, 2008; Felicetti et al, 2008; Mitomo et al, 2008). Initially, miR-15 and miR-16 were shown to be involved in the pathogenesis of chronic lymphocytic leukaemia (Calin et al, 2002), and later He et al (2005) and Johnson et al (2005) described a relationship between a miRNA cluster of mir-17–92 with Myc oncogenic pathway. One of the processes that tightly link miRNAs and cancer is the process called epigenetic modification. This epigenetic alteration in cancer has been shown to occur together with the genetic alterations in colorectal cancer (CRC). Thus, the aberrant hypermethylation process drives the initiation and the progression of colorectal polyps towards invasive of advanced stage colorectal tumours. Two recent studies have addressed the involvement of epigenetically silenced miRNA genes in CRC tumourigenesis including miR-34b/c and miR-342. Moreover, miR-148a has also recently been discovered to be involved in metastasis progression in several cancers including in CRC. Therefore, the understanding of the early process of miRNAs methylation signature in CRC is of utmost important to define the actual tumourigenesis process involved, other than that of hypermethylation protein coding tumour suppressor genes (Grady et al, 2008; Toyota et al, 2008). Analysing the promoter methylation of cancer-related genes has been always difficult, as it involved multiple steps for the conversion of unmethylated cytosine residues to uracil, but leaves 5-methylcytosine residues unaffected, particularly when analysing the samples derived from tumour specimens and bodily fluids. In this paper, we have applied the whole bisulfitome amplification method based on isothermal multiple displacement amplification technology. A uniquely processive DNA polymerase with a 3′–5′ exonuclease proofreading activity maintains high fidelity during the replication process. This technique was developed using REPLI-g technology to allow reproducible and representative amplification of bisulfite converted DNA, while maintaining the converted sequence representation (for further details can access from http://www.qiagen.com/). In this study, we investigate miR-34b/c and miR-148a hypermethylation pattern in CRC tissues and correlate it with clinicopathological features. Furthermore, we wanted to confirm whether miR-34b/c could be used as a possible screening marker for the detection of malignant colonocytes in faeces as a novel non-invasive method. We also aimed to determine the suitability of miR-148a as a prognosis factor for CRC. We showed that miR-34b/c could be an ideal candidate target for CRC screening in faecal microenvironment, whereas the involvement of epigenetic silencing factor in miR-148a correlates to poor prognosis. We found that 97.5% (n=79 out of 81) of neoplastic samples showed promoter methylation of miR-34b/c (Figures 2A and B). In contrast, promoter methylation of miR-34b/c was only detected in 14.3% (n=6/42; P<0.001; Figures 2A and B, Table 4) of matched normal colonic tissues. No correlation were found for miR-34b/c methylation pattern with clinicopathological status except for one significant correlation for pTNM stage, P=0.0110 (Table 4). However, this result is conflicting as only two samples were not positive for stage II cancer. We also wanted to determine the prognostic value of miR-148a methylation status in this patient cohort. Surgical approach in patients with colon cancer consisted of standard hemicolectomy (via laparotomic or laparoscopic approach) with negative macroscopic resection margin and regional lymphadenectomy in all cases. The median number (range) of removed lymph nodes was 17 (3–56). In rectal cancer, laparotomic/laparoscopic low anterior resection or abdominoperineal resections with total mesorectal excision were performed; the median number of removed lymph nodes was 13 (range: 4–37). Overall survival was calculated according to the Kaplan–Meier method in 60 patients available for survival analysis, considering cancer related death as the end point. The median follow-up period was 46 months (range: 1–266 months). Comparison between survival curves was performed with the log-rank test. The 10-year survival rate of the entire series was 60% (Figure 3A). Survival was significantly related to the stage of the tumour, and depth of invasion was one of the most important prognostic factors (log-rank test: P<0.05; Figure 3B). Interestingly, we found a trend towards lower overall survival in the patients with methylated miR-148a alleles (10-year survival probability: 48%) compared with patients with unmethylated miR-148a alleles (10-year survival probability: 65%; Figure 3C); however, the difference was not statistically significant (log-rank test: P=0.561) because of the small number of patients available for survival analysis. Previous studies have shown that half of the miRNAs are located in or near CpG islands, which are transcriptionally regulated by DNA methylation process that varies between normal and tumour cells (Lujambio et al, 2007, 2008). For example, the methylation pattern of miR-124a is tumour specific, whereas miR-127 is methylated in both normal and tumour tissues (Saito et al, 2006; Lujambio et al, 2008), and let-7a-3 is methylated in normal tissues, whereas it shows a hypomethylation pattern in lung adenocarcinomas (Brueckener et al, 2007). However, the mechanism underlying the deregulation of miRNAs in cancer has not yet been fully elucidated. With regard to this, we focused our study on the identification of miRNA with low expression in CRC and which may be involved in the development of CRC tumourigenesis. We evaluated expression levels of 46 miRNAs and found that 12 miRNAs were upregulated following AZA treatment. Of these, we decided to further investigate two miRNAs, miR-34b/c and miR-148a, which also previously been reported (Lujambio et al, 2008; Toyota et al, 2008); however, the clinicopathological features have not been fully evaluated in colorectal tumours. Two other main reasons for selecting these miRNAs are: (i) first, to evaluate the methylation pattern of miR-34b/c in tumours tissues in order to extend the work of Toyata et al (2008) and to further evaluate it in faecal samples. This would allow us to develop faecal-DNA-based markers for CRC screening; (ii) second, miR-148a was correlated to metastasis properties, yet no clinical correlation with methylation pattern is available for CRC. We also reasoned that miRNAs may be a better class of tumour marker because of their broad regulatory functions and the ability to measure their expression levels with far better accuracy than is currently achievable for mRNA. Thus, to further elucidate the role of these miRNAs, four CRC cell lines were treated with AZA, which allows us to demonstrate that reduced expression of both miRNAs is directly related to promoter methylation. The most striking evidence from our study is that the CpG island of the promoter region of miR-34b/c are hypermethylated in 97.5% CRC cases, and this methylation is, therefore believed to be tumour specific for CRC. The high percentage of methylation pattern in these CRC cases, and the contribution of miR-34 family on the p53 network, suggest that miR-34b/c may be involved in the response to colorectal tumourigenesis. As induction of cell cycle arrest, senescence and apoptosis are the mechanisms of oncosuppression by miR-34 family (Chang et al, 2007; Hermeking, 2010), permanent inactivation due to epigenetic silencing may result in a selective advantage for cancer cell proliferation. The first report by Toyata et al (2008) on methylation pattern revealed that miR-34b/c was aberrantly hypermethylated in primary CRC tumours. They showed that epigenetically silenced miR-34b/c could possibly be involved in the early process of tumourigenesis. Besides, Corney et al (2010) recently reported downregulation of miR-34b/c in ovarian cancer. They found that miR-34b/c is significantly reduced in stage IV compared with stage III tumours. Accordingly, their data support our observation in which methylation patterns are inversely correlated towards tumour stage in CRC. Furthermore, the results obtained by this study combined with our own observation, could correlate miR-34b/c hypermethylation status into the development of screening markers for CRC detection. We further evaluated miR-34b/c methylation in faecal specimens. Alteration in the genome that would lead to the progression of cancer could be detected using faecal specimens from CRC patients. Faecal DNA analysis was shown to represent a novel non-invasive method for CRC detection (Dong et al, 2001; Kalimutho et al, 2010b). We found that almost 75% of the CRC patients could be detected using faecal specimens. This is considered a good percentage to detect cancer using faeces as a screening marker, as methylated faecal DNA is almost inflexible to detect in the complex microenvironment of faeces. Furthermore, our data are based on a small group of faecal analysis, thus a large randomised blind study should take place to evaluate the actual sensitivity of this marker for CRC screening. In addition, the low percentage of CRC detection in faeces compared with 98% in tumour samples maybe because of the nature of the technique used for MSP observation. Hence, more advanced techniques, such as pyrosequencing or even quantitative-MSP using a specific probe of minor groove binder, would enhance the detection rate of miR-34b/c in faeces. However, as the methylation pattern of miR-34b/c shows almost no trend on tumour stage and possesses a high level of promoter methylation state in CRC, both in tumour and feaces, this would facilitate the feasibility of assaying faecal miR-34b/c as an additional test in clinical settings. Next, we wanted to see whether the hypermethylation of miR-148a could correlate into clinicopathological features of CRC. The miR-148a expression is shown to be downregulated in human breast cancer and undifferentiated in gastric cancer (Lehmann et al, 2008; Katada et al, 2009). DNA methylation-associated silencing of miR-148 expression is identified in human cancer cell lines established from lymph node metastasis of colon, melanoma, and head and neck cancer, suggesting its role in the development of metastasis (Lujambio et al, 2008). Duursma et al (2008) have shown that overexpression of miR-148a leads to a reduction in the expression of de novo DNA-methyl-transferase-3b (DNMT3b) enzyme. The suppression of miR-148a, in contrast, induces increased expression of DNMT3b. The evidence shows that alteration of DNMT3b expression clearly contributes to CRC tumourigenesis. Lin et al (2006) as well Jin et al (2009) showed that the increased expression patterns of DNMT3b protein in ApcMin/+ increases colorectal carcinogenesis because of the hypermethylation of tumour suppressors genes. Also, Ahmed et al (2008) showed that reducing expression of DNMT3b in PC3 tumour cells induces the loss of methylation at the promoters of several tumour suppressor genes such as APC, RB1 and RER-β. Therefore, methylation at the promoter of miR-148a leads to increased expression of DNMT3b, which in turn inactivates its tumour suppressor function in most of the cancers. The correlation of our data with miR-148a promoter methylation pattern, together with Lujambio et al (2008), allows us to link the epigenetically silenced miR-148a with the progression to advanced tumour stage. For example, 65% (n=30 out of 46) of patients with stage T3, 100% (n=5) stage T4 and 78% (n=7 out of 9) with stage N2 of CRC, have a condition of promoter methylation of miR-148a. Very recently, Chen et al (2010) reported the downregulation of miR-148a and miR-152 in gastric and CRC by expression analysis in which they found a strong correlation between these two miRNAs. In support of our data, they also reported that no statistical correlation was found with clinicopathological features including sex, age, tumour location, histological grade, pN stage or lymphatic vessel invasion in gastrointestinal cancers. Despite this, we observed that the positive cases of promoter methylation of miR-148a have a lower survival rate than the negative cases (48 vs 65% at 10 years). In addition, the number of cases in our patient cohort is too small to investigate a potential impact of treatment stratified according to miR-148a methylation status, and this may represent a clinical limitation of this study. Further prospective studies may be designed in order to assess the potential clinical utility of multimodality treatments in tumours with more aggressive biology, identified by miR-148a expression. In conclusion, as many CRC patients present with advanced disease, early detection leads to reduced mortality. Therefore, developing a miR-34b/c methylation assay as a diagnostic tool for early detection of CRC would have substantial clinical benefits. On the other hand, follow-up for the disease progression with targeted molecular markers such as miR-148a would enable classification of the prognosis of the disease as well as being a tool for therapy monitoring in patients having had CRC. Moreover, the reduced expression of miR-34b/c and miR-148a because of the epigenetic silencing could be considered as an important target of antineoplastic therapy development. The authors declare no conflict of interest. This work was supported by grants from the University Hospital Tor Vergata, University of Rome. MK was supported by the pre-doctoral scholarship for foreign student under the International Italian Government University scholarship scheme. We thank to Dr Sandra Van Schaeybroeck, Queen's University, Belfast for the critical reading of the manuscript.