Colorectal cancer DNA methylation patterns from patients in Manaus, Brazil

Background DNA methylation is commonly linked with the silencing of the gene expression for many tumor suppressor genes. As such, determining DNA methylation patterns should aid, in times to come, in the diagnosis and personal treatment for various types of cancers. Here, we analyzed the methylation pattern from five colorectal cancer patients from the Amazon state in Brazil for four tumor suppressor genes, viz.: DAPK, CDH1, CDKN2A, and TIMP2 by employing a polymerase chain reaction (PCR) specific to methylation. Efforts in the study of colorectal cancer are fundamental as it is the third most of highest incidence in the world. Results Tumor biopsies were methylated in 1/5 (20 %), 2/5 (40 %), 4/5 (80 %), and 4/5 (80 %) for CDH1, CDKN2A, DAPK, and TIMP2 genes, respectively. The margin biopsies were methylated in 3/7 (43 %), 2/7 (28 %), 7/7 (100 %), and 6/7 (86 %) for CDH1, CDKN2A, DAPK, and TIMP2, respectively. Conclusions Our findings showed DAPK and TIMP2 to be methylated in most samples from both tumor tissues and adjacent non-neoplastic margins; thus presenting distinct methylation patterns. This emphasizes the importance of better understanding of the relation of these patterns with cancer in the context of different populations.


Background
Colorectal cancer (CRC) comprehends tumors that affect the colon and rectum; it is the third type of cancer with the highest incidence in the world [1]. According to the 2014 estimates for Brazil, about 576,000 new cases of cancer were expected [2] of which 15,070 of those being for CRC in men and 17,530 for women. In particular, for the Amazon state (Brazil), a rate of 4.65 cases per 100,000 inhabitants [2] was estimated. The standard treatment for localized colorectal cancer involves surgical resection by open surgery of the primary tumor and regional lymph nodes. It has been shown that excessive consumption of red meat, alcoholism, body and abdominal fat, smoking, family history of CRC, genetic susceptibility to the development of chronic intestine disease, and age are all positively correlated to CRC [3,4]. Nevertheless, about 90 % of CRC occurs sporadically and without family history or genetic predisposition; i.e., less than 10 % of the cases are believed to be linked to one's genetics [5].
The study of epigenetics alterations has gained increasing attention; it comprises investigating how external or environmental factors affect the control of gene expression by turning genes "on" or "off " [6]. An example of a key epigenetic event is the methylation that occurs in small regions of DNA called CpG island which are located in the promoter region of genes [7]. Methylation defines the process of adding a methyl (CH 3 -) group in the promoter cytosine bases at the 5′ position to form 5-methylcytosine (5-MC). DNA methylation is usually linked with the silencing of gene expression for many tumor suppressor genes, such as, CDH1 [8], CDKN2A [9], TIMP2 [10], and DAPK [11]. The CDH1 gene expresses E-cadherin which is a transmembrane protein contained between epithelial cells and represents one of the main proteins involved in cell adhesion [12]. The loss of adhesion mediated by E-cadherin appears to have a fundamental importance in neoplastic processes. A down regulation of this gene is correlated with a decrease in the efficiency of cellular adhesion and therefore facilitating cellular motility; this in turn facilitates cancer cells for invading surrounding tissues [8]. Another gene strongly correlated to cancer is the CDKN2A; one of its products is the p16, a regulatory protein that inhibits the progression of cells through the G1 phase of the cell cycle [11]. Inactivation of p16 can cause abnormal cells and uncontrolled cell growth [13]. The TIMP2, for tissue inhibitor of metalloproteinases 2, is known to antagonize the activity of matrix metalloproteinases and suppress tumor growth, angiogenesis, invasion, and metastasis [14]; the mechanism of inhibition of this gene's expression remains unknown. Finally, DAPK (for death-associated protein kinase) participates in several functions in the cell, among them, playing a key role in the reorganization of the cytoskeleton in cytokine stimulation and induction of apoptosis [11]. Understanding DAPK's role in transcriptional regulation may lead to the discovery of novel therapeutics to combat cancer and inflammation associated diseases.
According to literature, abnormal methylation patterns of these genes is an indicative of cancer [15][16][17][18]. Studies have reported such alterations, for example, in the mucosa of colorectal tumor and suggested these genes as cancer markers [19,20]. In another work, the hypermethylation of DAPK and CDKN2A in the normal colonic mucosa of patients with CRC was evaluated by the methylight assay and the results showed their levels of methylation to be relatively low [mean percentage of a methylated reference (PMR) <1] [20]. In the same work, the authors correlated patients with high levels of methylation of INK4A and DAPK with advanced age (p < 0.1), supporting the hypothesis that age is one of the risk factors for this pathology. Finally, a panel of methylation markers for CRC diagnosis comprising CDKN2A, MGMT, MLH1, and SFRP was proposed [21].
Even though epigenetic markers are being increasingly applied in the screening for colorectal neoplasia's samples, further investigations, especially in different populations, is fundamental for a proper understanding of the applicability of these markers to the clinical practice. In the present study, we evaluated the methylation of the CDH1, DAPK, CDKN2A, and TIMP2 genes in five colorectal cancer patients from Manaus, the capital of the Amazon state in Brazil. Although the Amazon region presents a predominance of Indians, with approximately 342,800 indigenous people [22], this study included only patients recognized as having brown or white skin. This is the first study of this type performed in the population of the Amazon region. Table 1 shows the clinical characteristics of the five patients included in this study; we note that 20 % (1/5) of the tumors were located in colon and 80 % (4/5) in the rectum. The median age of the patients was 62 years. All the patients had resection margins free of neoplasia according to a histopathological exam. These results were compared to the methylation status obtained here. All of the tumor-node-metastasis (TMN) stages analyzed here had one or more genes methylated. Table 2 shows the gene methylation status for the CDH1, CDKN2A, DAPK, and TIMP2 genes.
Patient 4 had the tumor biopsy (4T) and margins (4PM and 4DM) methylated for all, CDH1, CDKN2A, DAPK, and TIMP2, except for 4DM not being methylated for CDH1. The tumor tissue of patient 8 (8T) and the respective adjacent margin (8DM) were not methylated for

Discussions
The DNA methylation leads to an altered gene expression resulting in changes in the control of cell proliferation and therefore being recognized as a major epigenetic modification in human genes [19]. There is increasing evidence that an aberrant methylation pattern is actively involved in early carcinogenesis [7]. In colorectal cancer, both hyper-and hypomethylation have been observed across different stages of progression. Here, we evaluated the prevalence of hypermethylation in the CDH1, CDKN2A, DAPK, and TIMP2 suppressor tumors genes of colorectal tumors tissue and their non-neoplastic adjacent margin using methylation specific PCR (MSP), which is a highly sensitive and specific technique [23]. Subjects having either brown or white skin were eligible for participating in this study. We note that the brown classification decurrently from the so called caboclos, descendants of Indians and Europeans-mostly English, French, and Spanish, that are commonly found in the northern part of Brazil [24]. It is the first time a study addressing the methylation pattern in colorectal cancer patients from the Amazon is performed. Our results show that each individual presented a distinct panel of methylated genes. In general, the methylated tumor tissues had their respective margin methylated. Likewise, genes, not methylated in tumor tissues were not methylated in their adjacent margin. Yet, patient 7 showed an opposing result by not presenting methylation on the tumor for genes DAPK and CDH1; these genes were methylated only in the respective adjacent margins (7PM and 7DM). We hypothesize that this may have happened considering that methylation is a reversible process [25] and, in this way, the gene could have been previously methylated in the tumor but metabolic factors reversed it. Another probable explanation is that tumor cells have a great heterogeneity [26] and, therefore, the examined area was not methylated.
The literature provides several studies that compare the methylation status of histologically normal mucosa with the neoplastic tissues. While some show no correlation between the methylation with the pathology, many studies, from around the world, were able correlate methylation of the tumor region and the corresponding adjacent margin [27][28][29][30]. Li et al. reported that 2.4 and 57.4 % of the samples were methylated in their tumor and adjacent tissue and that no methylation was detected in all the control tissue. A meta-analysis study revealed the frequency of the CDH1 promoter methylation in CRC tissues to be higher than those in control tissues (OR = 2.61, 95 % CI = 1:24-5:50, p = 0.012) [31]. When the ethnicity factor was considered, the CDH1 promoter methylation showed to be closely linked to the pathogenesis of CRC among Asians and Africans, but not among Caucasians. Our results show that the CDH1 promoter methylation in tumor tissue was lower than in the adjacent margins. CDH1 gene promoter methylation may lead to aberrant expression of E-cadherin [32]. The loss of adhesion mediated by E-cadherin appears to have a fundamental importance during the neoplastic processes, which allows cells having a controlled stop normal growth signaling, resulting in the loss of differentiation and increase in cell proliferation associated with the invasive behavior [8].
Xing et al. showed that hypermethylation of the CDKN2A gene correlated with a poor prognosis for CRC in European and Asian patients [28], but not for patients from other locations. In this regard, Xing et al., highlights the importance of more studies in different geographical locations to determine and clarify whether such Table 2 Methylation status of CDH1, CDKN2A, DAPK and TIMP2 genes suppressor genes serve as a prognostic factor for patients with CRC. In our study, the percentage of methylation for CDKN2A was 2/5 (40 %) for tumors tissue and 2/7 (28 %) for adjacent margins. Kuan et al. showed that advanced stage CRC patients presenting CDKN2A/p16 methylation were associated with higher risk of CRC recurrence as compared to those with tumor tissues that were not methylated [30].
In the literature, the methylation status of TIMP2 and DAPK is far from a global picture. Here, 80 % (4/5) of tumor samples and 86 % (6/7) of the adjacent margins, respectively, were methylated for TIMP2. However, in a histologically exam all margins were diagnosed as being free of tumor. For DAPK, 80 % of tumor samples and 100 % of the margins samples were methylated. Since DAPK is methylated, its anti-metastatic function is presumed to be lost [33]. Therefore, this work is aligned with the thesis that DAPK methylation poses as an important marker for the invasion, metastasis, and apoptotic processes in colorectal neoplasias and thus, ultimately, a potential marker for diagnosis and prognosis.
Even though epigenetic markers are being increasingly applied in the screening for colorectal neoplasia samples, there is still need for further investigations, especially in different populations, such as in Manaus. Although the number of patients in this report is small to draw solid conclusions on the importance of these markers in the context of Manaus population, the TIMP2 and DAPK were methylated on almost all tumors. Furthermore, our results showed a correlation with previous works, thus, supporting our claims.

Conclusion
Taken together, our results disclose methylation patterns for the CDH1, CDKN2A, DAPK, and TIMP2 genes on tumors and their respective margins. Although this report was limited to five patients, our results can serve as building blocks for future studies comprising metaanalysis of these genes and thus ultimately contributing to generating models with more statistical power. Such large-scale studies should consider establishing panels tailored towards specific population. Our observations pinpointed the heterogeneity of specific molecular abnormalities in each patient; most noticeably, the TIMP2 and DAPK genes were methylated in almost all samples of tumor and their adjacent margins. Yet, adjacent margins were diagnosed as disease-free according to the histopathological assessment. These facts make evident that a better understanding of these methylation patterns are fundamental, in time to come, to aid in developing more effective treatments and premature diagnosis of colorectal cancer.

Patients
This study was approved by the Ethics Committee of the Federal University of Amazonas (UFAM: MEMO, No. 27598614.1.0000.5020, CAAE). The samples were collected at the Oncology Control Center Foundation of Amazonas State (FCECON). After signing informed consent, five patients diagnosed with adenocarcinoma were evaluated. We obtained biopsies of the tumor (T), distal margin (DM), and proximal margin (PM) from two patients (Fig. 1) and only T and DM from the remaining three. Taken together, there were 12 biopsies. These samples were stored at −80 °C. The study includes patients of both genders, all aging between 18 and 80 years and presenting CRC diagnosed by colonoscopy and biopsy.

DNA extraction
The tissues were pulverized with liquid nitrogen and the DNAs were extracted from each sample using the kit InnuPrep Forensic (Biometra, Germany) according manufacturer's recommendations. The DNAs were quantified using NanoDrop 2000 (Thermo, USA).

Sodium bisulfite modification
Two micrograms of DNA were used to a final volume of 20 µL in sodium bisulfite modification according the recommendations of the EZ DNA Methylation-Lightning kit manufacturer (Synapse, Germany). Briefly, 130 µL of lightning conversion reagent were added to 20 µL containing DNA, vortexed and incubated at thermocycler (Mastercycler Gradient, Eppendorff ) under the conditions: 98 °C for 8 min, 54 °C for 60 min and 4 °C for 10 min to finish. After transferring the sample to the column, we added 600 µL of M-binding buffer, mixed and centrifuged for 30 s at 10,000×g. The supernatant was

Methylation-specific PCR (MSP)
DNA from the biopsies was subjected to bisulfite treatment and later was amplified by MSP, using primers specific to distinguish methylated from unmethylated DNA in bisulfite-modified DNA, taking advantage of the sequence differences resulting from bisulfite modification. The treatment with bisulfite results in conversion of unmethylated cytosines to uracil and leaves methylated cytosine intact. To detect these changes, specific primers were used (Table 3), as described previously [34][35][36][37].
For PCR analysis, 1 µL of bisulfite modified DNA in a final volume of 12.5 µL of reaction mixture was used containing 1× PCR buffer, MgCl 2 1.25 mM, 0.2 mM dNTPs, 0.4 M of each primer, 0.3U Taq polymerase (5U/µL). The reactions methylated and unmethylated were performed in a mastercycler thermocycler (Eppendorff ) and the parameters were: 96 °C for 1 min followed by 35 cycles of 94 °C for 1 min, 55 °C for 1 min, 72 °C for 1 min and final extension step of 72 °C for 7 min. The PCR product (10 µL) was directly loaded onto 10 % polyacrylamide gel stained with silver. Negative controls used in PCR were reaction mixture without DNA.

Table 3 Oligonucleotides used in the MSP
Oligonucleotides were used for regions containing frequent cytosines (to distinguish between modified and unmodified DNA) and contained CpG dinucleotides at the 3′ end (to provide maximal discrimination between methylated and unmethylated DNA). For each row, the gene names are listed in the primer pair column viz.: 5′-GTAGGTGAATTTTTAGTTAATTAGTGGTA3′ 5′-ACCCATAACTAACCAAAAACACCA-3′ [36]