Colorectal cancer (CRC) is the third most common cancer in men and women, accounting for 11% of all cancer-related deaths. The majority of cases are diagnosed in advanced stages when a curative treatment is less likely to occur and chemotherapy is the only option [1]. The identification of the molecular, genetic, and epigenetic changes underlying the adenoma-carcinoma sequence [2] leading to CRC has been the focus of many researches [3]. It is now widely accepted that sporadic CRC frequently arises from preneoplastic lesions through the activation of proto-oncogenes, such as K-ras, and the inactivation of tumor suppressor genes (TSG), such as APC, p53, DCC, and the mismatch repair genes [4, 5]. Apart from mutations, gene expression may also be modified by altering of DNA methylation [6]. Two general phenomena have until now been observed. The first one is global DNA hypomethylation with decreased 5-methylcytosine content which results in both enhanced expression of proto-oncogenes [7] and genomic instability [8]. The second event is represented by local DNA hypermethylation of CpG islands, short sequences rich in CpG dinucleotides in the 5'untranscribed region (5'-UTR). This event occurs in approximately half of all human genes [9, 10] silences specific TSG, and accelerates cancer formation [11, 12]. Treatment with DNA demethylating drugs, such as 5-aza-2'-deoxycytidine (5-Aza-CdR or Decitabine), was shown to reverse the hypermethylation and restore expression of TSG [13]. Therefore, cancer-specific promoter methylation may by itself serve as a valuable clue to uncover novel TSG.

In the present study, we aimed to uncover novel targets of promoter methylation in CRC, by combining gene expression profile data, already highlighted by our group [14], with results of demethylating assay and in silico screening for CpG islands.



Peripheral blood, primary tumour and matching normal tissue samples from a cohort of 30 consecutive CRC patients undergoing curative surgery at our Institution were collected. Clinical data, tumour location, and AJCC staging of these patients are shown in Table 1. Primary tumour and matching normal tissue samples were obtained from a second cohort of 8 CRC patients, and used in a validation assay. Genomic DNA was isolated from peripheral blood samples using standard techniques. Tissue samples were immediately frozen in liquid nitrogen and stored at -80°C until nucleic acids extraction. The study was approved by the Ethics Committee at our Institution, and all patients gave their informed written consent.

Table 1 Clinical data of colorectal cancer (CRC) patients

RNA extraction from fresh frozen tissue

About 150–200 mg fresh frozen tissues were used to isolate total RNA by phenol extraction (TRIzol Reagent, Invitrogen Corporation, Carlsbad, CA, USA) which was subsequently purified by column chromatography (RNeasy Mini Kit, Qiagen, Valencia, CA, USA). RNA integrity was monitored using MOPS gel electrophoresis.

Cell culture, 5-Aza-CdR Treatments

The HCT116, CaCo2 and SW480 cell lines were purchased from the American Type Culture Collection (ATCC, Rockville, MD, USA), and maintained in DMEM medium (Invitrogen Corporation, Carlsbad, CA) supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin in a humidified 5% CO2 atmosphere at 37°C. For demethylation studies, cells were seeded at a density of 1 × 106 cells per 100-mm dish, and incubated for 24 hrs in a growth media. Subsequently, 5-Aza-CdR (Merck Chemicals Ltd., Nottingham, UK) was added to the incubation mixture following two different protocols. In the acute treatment 1 μM 5-Aza-CdR was added to incubation mixture for 24 hrs; afterwards, the medium was changed once daily for 3 consecutive days; DNA and RNA content were checked at 2nd, 4th and 6th days [15, 16]. In the chronic treatment, 2 μM 5-Aza-CdR was added for 24 hrs at day 1st, 3rd and 5th [17]; at each experimental day, the cells were placed in fresh medium and harvested at day 6th to isolate DNA and RNA.

qPCR Assay

For qPCR, 1.0 μg of total RNA from CRC cell lines and normal and tumour tissues of the second cohort of CRC samples was used with hexamer random primers to run the first strand cDNA synthesis by the RT-System kit (Promega Corporation, Madison, WI). Oligonucleotide sequences were designed by means of the PrimerExpress program (Applied Biosystems, Applera, Foster City, CA) with default parameters in every case; whenever possible, the oligos were designed to span an intron region (Table 2). To ensure specificity, amplicon sequences were checked by both BLAST and BLAT programs against the human genome. The efficiency of each oligo pairs was checked by diluting a series of control cDNAs. All qPCRs were performed in a 10-μl final volume, in three replicates per sample, set up in a 384-well plate format with the Biomek 2000 robot (Beckman Coulter, Inc., Miami, FL). The assays were run in an ABI 7900 Sequence Detection System (Applied Biosystems, Applera, Foster City, CA, USA) with the following amplification conditions: 50°C for 2 min, 95°C for 10 min, and 50 cycles at 95°C for 15 s and at 60°C for 1 min. Expression of mRNA from candidate genes was analysed quantitatively by means of SYBR Green Real Time PCR (Invitrogen Corporation, Carlsbad, CA) and raw Ct values calculated with SDS2.0.

Table 2 Primer Sequences and Conditions for qRT-PCR, Bisulfite sequencing and MSP Analysis

In SilicoSearch and Bisulfite Sequencing Analysis (BSA)

The presence of CpG islands, overlapping the 5'-UTR, was examined by means of the MethPrimer, according to CpG islands definition.

Bisulfite modification of DNA from colon cancer cell lines, peripheral blood and frozen tissues of patients was assayed, as reported by Herman et al. [18]; normal lymphocytes (NL) and in vitro methylated DNA (IVD) were used as negative and positive controls, respectively. In the assay, 1 μg of DNA was denaturated by treatment with NaOH at 37°C for 10 min, followed by incubation with hydroquinone and sodium bisulfite at 50°C for 16–17 h in the dark. After treatment, DNA was purified using DNA cleanup kit (Promega Corporation, Madison, WI), incubated with NaOH at 37°C for 15 min, precipitated with ammonium acetate and 100% ethanol, washed with 70% ethanol and, finally, re-suspended in 25 μl of distilled water. DNA methylation patterns in the CpG islands were determined by BSA using the primers listed in Table 2. The PCR conditions were 3 min at 94°C, 30 cycles of 94°C for 30 sec, specific annealing temperature for 30 sec, and 72°C for 60 sec. The sequence of the PCR products was analysed by using Sequencing Analysis 3.4.1 (Applied Biosystems, Applera, Foster City, CA, USA).

MSP Assay

Qualitative analysis of CpG islands in the promoter region of the NDRG2, p16, APC, and MLH1 genes in 30 patients of CRC, in cell lines, in NL and IVD was carried out by MSP assay [18]. The primers for unmethylated and methylated DNA are listed in Table 2. For the NDRG2 promoter region we used two different primers (NDRG2_UnM/NDRG2_M and NDRG2_UnM2/NDRG2_M2) to cover the same region sequenced by the bisulfite assays. PCR reaction was carried out in a 25 μl mixture containing 0,2 mM each dNTP, 1.5 mM MgCl2, primers (10 μM each), bisulfite-modified DNA (50 ng), and 0.75 U of Amplitaq Taq Gold polymerase (Applied Biosystems, Applera, Foster City, CA) for 35 cycles (95°C for 12 min, 94°C for 1 min, TA for 1 min, then 72°C for 1 min, followed by a final extension at 72° for 5 min) and analysed on a 3% agarose gel stained with ethidium bromide. All reactions were run in duplicate to ensure consistent and reproducible results.

Statistical Analysis

We carried out three separate statistical analyses. In the initial analysis, qPCR data were used to validate our previous array results [14]. Calculations were made using the Comparative CT method [19, 20]. We used three genes, that is hEEF1A1, hGAPDH and hHRPT1, to normalize input cDNA for each sample, with cDNA of normal tissue used as calibrator. The chi-square method with one degree of freedom (χ2 1), calculated by BMDP Statistical Software (BMDP Statistical Software, Cork Technology Park, Model Farm Road, Cork, Ireland) [21], was used to asses statistical significance of expression difference for each gene of the 8 paired samples. The second analysis concerned the variation of gene expression before and after 5'-Aza-CdR treatment by means of the T test, calculated by BMDP Statistical Software. Expression level of the post-treatment specimen compared to the pre-treatment specimen was calculated as a log-transformed ratio. A gene was classified as up-regulated following the 5-Aza-CdR treatment when relative mRNA expression was greater or equivalent to 1.65-fold in at least one treatment condition in one cell line. Genes with no change or very low expression levels in post treatment specimens were no further considered in the analysis.

The final part of our analysis evaluated the BSA data. The median number of full CpG islands, present in normal and tumour tissues, was calculated and compared in tumour (T) matched normal (N) tissue of the same patients. The χ2 1 method was used to asses significant difference; a number of CpG islands higher than 5 was taken as statistically significant.


qPCR validation of deregulated genes

Among the genes higlighted as significantly deregulated in our previous microarray study [14], 24 genes were selected for further validation by using the quantitative Real-Time PCR (qPCR), based on their cellular function, such as transport, signal transduction, intracellular and cell surface signalling, cell cycle, replication-repair of DNA, and protein folding. (Table 3).

Table 3 List of genes selected from microarrays analysis comparing normal mucosa matched tumour colon tissue. The different expression in tumoural tissue was showed.

qPCR assay was applied to analyse mRNA expression of the 24 selected genes, as well as of three control housekeeping genes (GAPDH, EEF1A1 and HRPT1) in the tumour and normal tissues taken from new cohort of 8 patients with CRC.

Compared to normal tissue with an expression profile normalized to 1, in tumour samples 7 genes (ABCA8, AQP8, CLCA4, HPGD1, PRDX6, SLC26A3, and STX12) were uniformly under expressed in all 8 CRC patients, 3 genes (MXI1, NDRG2 and SCNN1B) in 7 patients, and 3 other genes (SGK2, NR3C2, and SGK1) in 6 patients. Eight genes, that is CSE1L, GTF2IRD1, HSPH1, NEBL1, RFC3, SLC12A2, FOXM1 and SOX9, were specifically over-expressed in tumour tissues from 7 patients (Fig. 1). The remaining 3 genes (TPX2, UBE2C, CCNB1) were excluded from the analysis because of indeterminate qPCR values. In general, qPCR results were in agreement with the microarray data.

Figure 1
figure 1

Logarithmic expression profile value of twenty-one genes determined by quantitative reverse transcription-PCR by using the Comparative CT method. Three housekeeping genes were used to normalize input cDNA for each sample with colorectal cancer, with cDNA of normal tissue used as calibrator. Crosses represents mean of triplicate determinations. *P < 0.05; ° P < 0.01; ^ P < 0.001.

Gene expression before and after 5-Aza-CdR in colon cell lines

To investigate the role of methylated CpG islands in the modulation of gene expression, HCT-116, CaCo2 and SW480 human colon cancer cell lines were cultured with different doses of 5-Aza-CdR to induce a demethylation event, and gene expression levels were measured by means of qPCR. Two 5-Aza-CdR challenge regimens were used to obtain expression data under different cellular conditions: the acute treatment focused on moderate DNA demethylation to minimize cell viability, and the chronic treatment to maximize DNA demethylation (see Additional file 1).

Ten out of the 21 validated genes, i.e. ABCA8, AQP8, HPGD, PRDX6, SLC26A3, STX12, NDRG2, MXI1, SGK2, and SCNNB1, were analysed for the impact of their DNA hypermethylation on epigenetic events (Fig. 2). Other genes had no epigenetic influence and were, therefore, excluded from the analysis. Indeed, the underexpression of NR3C2 and SGK1 has been related to aldosterone regulation pathway [22], whereas the 14-3-3ε gene modulates the CLCA4 gene by interacting with the calmodulin-dependent pathway [23]. Demethylation of the 5'-UTRs of some genes with a concomitant increase in mRNA expression was documented in cell lines (Fig. 2). For HPGD and NDRG2 genes a 1.4 and 1.3-fold increase in CaCo2, a 7.0 and 1.6-fold increase in HCT116, and a 4.1 and 1.3-fold increase in SW480, respectively, was observed. PRDX6 gene expression increased 1.5 and 3.0-fold in Caco2 and HCT116 cells, respectively. MXI1 showed a 1.8 and 5.0-fold increase in Caco2 and SW480 cells, respectively (Fig. 2).

Figure 2
figure 2

Histograms depict expression levels of AQP8, HPGD, PRDX6, MXI1, NDRG2, SCNNB1 and SGK2 genes in CaCo2 cell line (A), HCT116 cell line (B) and SW480 cell line (C) before and after exposure to 5-Aza-CdR determined by quantitative real-time PCR using the Comparative CT method. 2^DCt indicates the ratio between the values of CT normalized to three housekeeping genes and compared to cDNA of untreated cells used as calibrator. Error bars indicate standard deviation from triplicate experiments. p16 gene was used as positive control in the experiments. Statistical significance threshold p-value (p < 0.001) for both acute and chronic treatment are shown. Asterisks indicate P < 0.05 values and double asterisks indicate P < 0.001 values.

In silicosearch verification and Bisulfite Sequencing Analysis (BSA)

Two of the demethylated genes had CpG islands overlapping their putative promoter regions at in silico confirmation. By means of the MethPrimer software, we found 16 CpGs islands located between nucleotides 20,563,460 and 20,564,147 in NDRG2 genes (Fig. 3A and Additional file 2), and 26 CpGs islands located immediately at the 5' of the transcription start site and exon 1 (nucleotides 171,170,862 and 171,713,160) in the PRDX6 gene (Fig. 3A).

Figure 3
figure 3

CpG islands present in putative promoter regions of the two genes of interest. A. Promoter structure of NDRG2 gene (the sequence are shown in Additional file 2). The black arrows correspond to NDRG2 primers for BSA assay, the grey arrows correspond to NDRG2_M_2 and NDRG2_M primers for MSP assays. B. CpG islands present in NDRG2 (numbered from 1 to 16) and PRDX6 (numbered from 1 to 26) genes obtained by MethPrimer software. Methylation status of CpG sites in vitro methylated DNA (IVD), normal lymphocytes (NL), three colon cancer cell lines, and in normal (N) and tumor (T) tissue of one patient with colorectal cancer. Methylated and unmethylated cytosine residues are indicated with filled and small circles while open circles denote partially methylated sites.

The methylation status of promoter regions of these putative tumour-suppressor genes was assessed by BSA in untreated cell lines, in tumour matched to normal tissues of one patient, in vitro methylated DNA (IVD) and in normal lymphocytes (NL). PRDX6 showed dense methylation only in IVD; NDRG2 showed a significant methylation in cell lines and in tumour tissue compared to normal tissue, suggesting a potential epigenetic regulation of the gene (Fig. 3B and Additional file 3). Full methylation in all 16 CpG sites of the NDRG2 gene was found in HCT116 and CaCo2 cell lines, and partial methylation at the 3th CpG site in the SW480 cell line (Fig. 3B).

Quantitation of NDRG2 methylation in paired tumour and normal tissue samples of CRC patients

To determine whether hypermethylation of the NDRG2 gene could be ascertained in primary CRC (Table 1), BSA of 30 primary colon tumour tissues matched to normal tissues was performed. When compared to their paired normal tissues, a relative increase of methylation in tumours was observed in 19 of 30 (63%) CRC patients (Fig. 4), but the increase was significant in only 3 patients (χ2 > 5, df = 1, p < 0.05). In four tissue pairs, the relative methylation was apparently decreased, likely due a low sensitivity of the detection method. In the remaining patients figures were unchanged.

Figure 4
figure 4

Methylation analysis of the NDRG2 promoter in 30 tumour and normal tissue pairs, evaluated by the bisulfite sequencing analysis. Data are expressed as the difference in methylation status between tumour and normal tissue.

Methylation-specific PCR (MSP) assay in colon cancer cell lines and primary CRC samples

MSP was performed to examine the methylation status of CpG islands identified in the NDRG2 gene (see Additional file 3). The methylation status of the gene was compared with that observed in three usually hypermethilated genes (p16, APC, and MLH1) in 3 colon cancer cell lines (HCT116, CaCo2 and SW480) and in 30 paired tumour-normal tissues. CpG methylation in NDRG2 was detected in all cell lines and in 8 of the 30 (27%) colorectal cancer patients (Table 4). No methylation was detected in 30 samples from normal tissue. Hypermethylation of APC, p16, and MLH1 genes in tumour tissue was found in 3 (10%), 4 (13%) and 6 (20%) patients, respectively, but not in normal colonic tissue (Table 5).

Table 4 Comparison of the clinicopathological features of 30 CRC patients according to the presence of NDRG2 methylation
Table 5 Promoter gene methylation rates in tumour and normal tissue from patients with colorectal cancer (CRC) sorted by tumour location

When relating the NDRG2 methylation status to clinical pathologic features, no association with age, gender, tumour site, and MSI status was observed. Conversely, a significant correlation was found between the NDRG2 methylation and the AJCC stage of the cancer (Z test, p < 0,05) (Table 4).


Carcinogenesis is a complex event characterized by the progressive development of genetic and epigenetic aberrations which ultimately result in loss of physiological control of cell growth and differentiation. The two most important epigenetic mechanism are represented by the DNA methylation, the conversion of cytosine into methyl-cytosine catalyzed by the DNA methyltransferase And histone modifications [24]. Changes in the DNA methylation pattern may occur everywhere in the DNA molecule. Global DNA hypomethylation generally occurs in centromeric repeats and repetitive sequences and contributes to carcinogenesis by causing chromosomal instability, reactivation of transposable elements, and loss of imprinting [25]. Hypermethylation is especially frequent in CpG islands, i.e. short DNA sequences rich in CpG dinucleotides, mostly located in the 5'-untranslated region (5'-UTR) of genes [24]. When CpG islands are heavily methylated, transcriptional gene silencing generally occurs. Although the fine mechanisms of regulation of the "epigenetic" machinery are still poorly understood, the DNA methylation may switch on or off several genes and, in particular, those regulating important biological phenomena, such as cell growth and differentiation [25]. In normal cells, this epigenetic mechanism is involved in several physiological events, such as the inactivation of X chromosome in female cells, silencing either paternal or maternal alleles of "imprinted" genes, and transcriptional blocking of exogenous integrated genes potentially dangerous for the cell life. However, aberrant DNA methylation is also relatively common in cancer cells and is likely to play an important role in cancer initiation and progression [26].

Since the pioneer studies of Baylin et al. [27], it has been widely recognized that cancer cells are characterised by two opposite events: a global hypomethylation which results in either up-regulation of proto-oncogenes and induction of genomic instability, favouring both uncontrolled cell growth [9] and mutations, and CpG islands hypermethylation of other genes, the so-called tumour-suppressor-genes (TSG), which contributes to loss of the negative control of the cell cycle [12]. Searching for up-regulated oncogenes and down-regulated TSG is important in basic science, especially when an epigenetic mechanism (hypomethylation or hypermethylation) is suspected. In fact, oncogenes and TSG not only may elucidate the highly complex molecular derangement in cancer cells, but also may be used as potential targets for new therapeutic approaches. DNA methylation is a reversible phenomenon which can be modulated by specific agents. An example is represented by demethylating drugs which can globally reduce the DNA methylation level of TSG promoters, restoring their normal activity. Interestingly, some in vitro experiments have shown that cancer cell lines reverted to normal phenotype after treatment with demethylating agent.

The current study was carried out with a three-step design. First, we specifically looked at up- and down-regulated genes not yet firmly associated with colon carcinogenesis, and selected 24 genes for validation with qPCR. A straight correlation between results obtained from qPCR and those from DNA microarray was found, implying that DNA microarray technology is a reliable tool to search for new genes significantly deregulated in cancer [28]. Second, we selected 10 of 21 genes (ABCA8, AQP8, HPGD, PRDX6, SLC26A3, STX12, NDRG2, MXI1, SGK2, and SCNNB1) as possible targets of epigenetic modifications in colon cancer, and after treatment with a demethylating agent, seven of them showed a significant increase of mRNA expression (AQP8, HPGD, PRDX6, MXI1, SCNNB1, SGK2 and NDRG2). From an in silico screening, only 2 genes (PRDX6 and NDRG2) were considered as possible candidates for the presence of CpG islands in their 5'-UTR. For the excluded genes, additional mechanisms of transcriptional regulation were hypothesized to be responsible for their differential expression. Third, to evaluate the methylation status of PRDX6 an NDRG2 genes in normal and cancer tissues, as well as in colon cancer cell lines, bisulphite sequencing analysis was used. In the PDRX6 gene the methylation status was not different from that observed in normal tissue. In the NDRG2 gene a significant methylation status either in colon cancer cell lines and in tumour tissue compared to normal tissue was observed. The underexpression of the PRDX6 protein responsible for the red-ox regulation of the cell, was found to be correlated with loss of function of NKX3.1 gene, known as TSG [29].

Using these approaches, the NDRG2 gene was selected for further analysis because: (i) it was suppressed in all colon cancer cell lines, (ii) its expression may be up-regulated in all cell lines by 5Aza-CdR treatments, and (iii) it is involved in important biological process such as cell growth [30], differentiation [31] and apoptosis [32]. The NDRG2 gene is a new member of the N-myc downstream-regulated gene (NDRG) family, that is located on chromosome 14q11.2 and encodes for a 41 kDa protein. It has been proposed that the NDRG2 gene is a candidate TSG, and its expression is low or undetectable in several primary tumour and tumour cell lines [30, 33, 34]. Liu et al. [35] revealed that the down-regulation reported in cancer be driven by promoter methylation, mutation, and genomic deletion of the NDRG2 gene. Recently, it has been shown that expression of the NDRG2 protein is modulated by the insulin-stimulated Akt-dependent phosphorylation [36]. Several studies have suggested that the NDRG2 mRNA is down-regulated or undetectable in a number of human primary cancers, such as squamous cell carcinoma, pancreatic cancer [37], glioblastoma [30], and cancer cell-lines. Recently, Zhang et al. [38] have demonstrated that c-Myc represses NDRG2 gene expression via Miz-1-dependent interaction with NDRG2 core promoter region, and this inverse regulatory relationship induces cell differentiation and proliferation.

The MSP assay was used to check for NDRG2 methylation status in 30 primary colon tumour tissues compared to normal colonic mucosal samples. After sorting colon cancer patients by age, gender, tumour site, and MSI status, no statistically significant association was observed between these features and the NDRG2 methylation. Nevertheless, there was a trend towards NDRG2 methylation status with an advanced tumour stage of the CRC samples, with significant value detected in patients with AJCC stage IV (p < 0.05). These results are in agreement with those reported in other cancer types [34, 39] where NDRG2 expression is reduced in high-grade compared to low-grade tumours. In particular, Lorentzen et al. [39] suggested that in CRC samples the down-regulation of NDRG2 expression occurs during the progression from adenoma to carcinoma.


In conclusion, we showed that NDRG2 expression is frequently suppressed in colon cancer cell lines in conjunction with aberrant DNA methylation, and that the loss of expression of this gene could be related to advanced colon tumour stage.