DNA methylation is one of the different ways of gene regulation in mammals, and it has been a research hot spot in epigenetics in recent years. DNA methylation is a chemical modification of DNA that can regulate genetic expression without changing the DNA sequence. In a broad sense, DNA methylation refers to the chemical modification process wherein a methyl group is added to a specific base of the DNA sequence via covalent bonding, with S-adenosylmethionine (SAM) as the methyl donor, under the catalysis of DNA methyltransferase (DNMT). Although there are numerous variations in methylation modification, the bases of the modified sites are usually adenine n-6, cytosine n-4, guanine N-7 and cytosine C-5. DNA methylation involved in general research mainly refers to the methylation process of the fifth carbon atom of cytosine in cytosine guanine dinucleotide (CpG) dinucleotide, and the final product of this process is 5-methyl-cytosine.

After fertilization in mammals, entire methylation of DNA molecules inherited from their parents is cleared by specific enzymes. After the embryo is implanted into the uterus, a new round of methylation begins, and the genome of the fertilized egg is methylated again under the action of DNA methylase. The stability of DNA methylation is maintained by DNMT1.

In mammals, DNMT1 and DNMT3 are the most essential DNA methyltransferases. DNMT1 is involved in the maintenance of DNA methylation and is related to the extension of DNA methylation. DNMT3 includes DNMT3a and DNMT3b, which are involved in de novo methylation. Other roles and applications of DNMT family remain under investigation (8,9).

DNA demethylation is regulated by internal fragments of genes and their binding factors. There are two hypotheses that can explain the molecular mechanism underlying DNA demethylation. One hypothesis associated with DNA semi retention replication is passive demethylation. If DNA methylation does not occur after semi-retention replication, the DNA would be in a semi-methylated state. If the semi-retention replication of semi-methylated DNA occurs again and the DNA methylation activity is still inhibited, it indicates that 50% of the cells are in a semi-methylated state. The second hypothesis is independent of the semi-retention replication and is an active process, that is to say, DNA demethylation is catalyzed by DNA demethylase. DNA demethylation is the removal of methylated bases under the action of DNA glycosidase, which is equivalent to the repair of damaged DNA by glycosidases and base free nucleases.

Numerous genes, especially the promoter region of the housekeeping gene, usually have some regions rich in di-nuclear glycoside ‘CG’, called ‘CpG islands’. They are usually found in the promoter and exon regions of genes. Some regions are rich in CpG dinucleotides, with a length of 300–3,000 base pairs. Methylation of specific CpG dinucleotides in the promoter region is involved in transcription regulation. Scientific studies have revealed that DNA methylation modification demonstrates a stronger inertia in vitro (10).

Several studies revealed that the methylation status of tumor cells in the promoter regions of tumor suppressor genes and repair genes is enhanced, resulting in the inhibition of the expression of corresponding tumor suppressor gene (11,12). Moreover, abnormal methylation is common in chronic inflammation (13), which in turn contributes directly to tumors and cancers to a large extent.

In 2015, a study on the CRC subtype alliance summarized four common molecular subtypes (CMS) of CRC: CMS1, CMS2, CMS3 and CMS4. Among them, CMS1 is highly mutated, and all microsatellite-unstable CRCs are CMS1 (14). These kinds of tumors have high methylation status and BRAF mutation rates and a poor prognosis with the other types. However, the other CMSs demonstrate more chromosomal instability (CIN) in CRC typing. Although CMS3 demonstrates moderate hypermethylation levels, the KRAS gene is highly mutated. CMS3 CRC has a high mortality rate once it recurs. In the present study, the classic chromosome instability (CIN) and microsatellite instability (MSI) pathways have been used and other pathways for the pathogenesis of CRC as examples to elucidate the development process of CRC.

Studies have revealed that mutations of the tumor suppressor genes APC regulator of Wnt signaling pathway (APC) and TP53, and activation mutations of KRAS and PIK3CA are related events in CIN pathway-induced tumors. Interestingly, the earliest event in colorectal tumors appears to be mutation inactivation of APC (15), which causes the Wnt signaling pathway to be activated, a feature common to almost all tumors (16).

After the mutation of APC, the mutation of KRAS also occurs (17): When normal, the KRAS image molecular switch can control the pathway that regulates cell proliferation; however, when abnormal, it causes continuous cell proliferation and prevents self-destruction. The KRAS image molecular switch participates in intracellular signal transduction. When the KRAS gene is mutated, it is permanently activated and cannot produce normal RAS protein, leading to intracellular signal transduction error, rampant cell proliferation and canceration (18).

Most sporadic CRC follow the CIN pathway, whereas 85% of sporadic CRC have cyclooxygenase 2 (COX-2) expression (19). This suggests that COX-2 is also an important factor in inducing tumors. During the development of colorectal tumors, COX-2 may convert free arachidonic acid into prostaglandins and regulate the proliferation of CRC cells; and it can be used as an alternative approach for the treatment of CRC (20).

Regionalized hypermethylation is a common feature of CRC with MSI phenotype, including CpG island methylation phenotype (CIMP). In a study by Samowitz et al (21), APC mutations were observed in both MSI and CIMP high cancers, and there was a significant inverse correlation trend between APC mutations and CIMP. However, the mutation frequency of APC and TP53 that was aforementioned is lower in the MSI pathway than in the CIN pathway (22).

Carcinogenesis of the MSI pathway includes the mutation of TGF-βR2. The TGF-βR2 gene encodes a protein that inhibits the proliferation of colonic epithelial cells. The mutated expression product of the gene no longer sends a signal to prevent proliferation, which leads to abnormal proliferation of colonic epithelial cells. Other mutated genes lead to cell cycle arrest (CASP5 and FAS) and abnormal DNA repair (MBD4, BLM and CHK1) (23–27).

The serrated pathway is an alternative multistep mechanism of carcinogenesis. It has been revealed that serrated colorectal lesions rarely harbor truncated APC mutations, most are BRAF mutations, and KRAS mutations remain rare. Regarding the serrated pathway, there was evidence of Fusobacterium nucleatum in 56% of CRCS, which was associated with CIMP-H status and large tumors (28).

In addition, IBD-CRC may represent a distinct tumorigenic pathway. Typical epithelial tumor subtypes associated with Wnt signaling are completely absent in IBD-CRCs, and different mechanisms of Wnt pathway dysregulation bias IBD-CRC toward mesenchymal tumor subtypes (29) (Table I).

Hypermethylation of the tumor suppressor gene CpG island is one of the mechanisms underlying CRC formation. So far, it has been revealed that numerous cancerous genes experience hypermethylation, and only a few protective mechanisms, such as active transcription, active demethylation, replication timing, and prevention of the acquisition of local chromatin structure of deoxyribonucleic acid methyltransferase, can prevent hypermethylation of CpG islands.

Recently, hypermethylation of ADHFE1, CNN1 and NR3C1 has been revealed to play important roles in signal transduction, cell cycle regulation, angiogenesis, as well as CRC (33–35). Suzuki et al (36) reported that the incidence of hypermethylation and downregulation of the SFRP gene (a negative regulator of Wnt signaling pathway) in normal colonic mucosae of patients with CRC is higher than in the mucosae that of patients without CRC.

A meta-analysis revealed a relationship between ITGA4 promoter methylation status and malignancy, that is, IITGA4 hypermethylation was more frequent in tumor samples than in non-tumor samples. ITGA4 methylation analysis enables a reliable method for screening CRC in tissue samples. This has important implications for the detection of early CRC (37).

Hypermethylation of the promoter may contribute to the epigenetic silencing of ADAMTS14 in CRC. ADAMTS14 protein expression was higher in the anterior part of the invasive tumor than in the tumor center or other regions of the tumor. Its high expression is associated with poor prognosis in patients with CRC, suggesting that ADAMTS14 may be a promising indicator for evaluating the prognosis of CRC (38).

Hypomethylation, including CIN and MSI, is considered to promote tumorigenesis by activating proto-oncogenes, such as CMYC and HRAS, or causing genomic instability.

There are two common low functional polymorphic variants of methylenetetrahydrofolate reductase (MTHFR): T variant at nucleotide 677 (MTHFR C677T) and C variant at nucleotide 1298 (MTHFR A1298C). The first variant, C677T, is related to a reduced risk of CRC, and the cancer risk associated with MTHFR polymorphism may be regulated by folate intake. With sufficient folate intake, individuals carrying the variant MTHFR genotype may have a reduced risk of cancer. When folate intake is low, DNA methylation and DNA synthesis/repair in polymorphic individuals may be impaired, resulting in an increased risk of cancer. In animal studies, folate deficiency has been revealed to lead to exon-specific hypomethylation of p53 gene and increased DNA methyltransferase activity (39). However, moderate folate deficiency did not cause DNA methylation (40). It was discovered that MTHFR C677T polymorphism affected DNA methylation status through interaction with folate status. Another study examined the relationship between plasma folate status and colorectal adenoma, by assessing the effect of modification on the polymorphism of its gene (C677T). Compared with subjects with normal CC folate metabolism or diminished CT folate metabolism, subjects with poor TT folate metabolism had a low risk of developing colorectal adenoma at higher plasma folate levels (adjusted odds ratio, 0.58; 95% confidence interval, 0.21 to 1.61) and increased risk when their folate levels were low (adjusted odds ratio, 2.13; 95% confidence interval, 0.82 to 5.54) (41). This indicates that the risk of CRC is reduced when folate intake is low or metabolism is abnormal, that is, when the degree of DNA methylation is low. This also provides a breakthrough for the clinical treatment of CRC by controlling folate intake or using DNA methylation inhibitors.

The present study revealed upregulation and hypomethylation of HER3 gene expression in CRC cases. High expression and hypomethylation of HER3 may play an important role in the occurrence and development of CRC. CpG hypomethylation may be associated with the early stages of tumorigenesis. The discovery of this biomarker provides a powerful approach to improve current diagnostic and therapeutic measures (42).

KRAS is a mouse sarcoma virus oncogene. There are three genes in the RAS gene family associated with human tumors: HRAS, KRAS and NRAS, located on chromosomes 11, 12, and 1, respectively. KRAS encodes a 21-kDa RAS protein, also known as the p21 gene, and has the greatest impact on human cancer. It is like a molecular switch: When normal, it can control the pathways regulating cell proliferation, and when abnormalities occur, it leads to continuous cell proliferation and prevents cell self-destruction. It is involved in intracellular signal transmission. When KRAS is mutated, the gene is permanently activated and cannot produce normal RAS protein, leading to an intracellular signal transduction disorder, that is, the inability to control cell proliferation and thus, cancer (43).

CRC is characterized by a series of mutation events involving APC, KRAS and TP53. KRAS is the most significant oncogenic mutation in CRC, occurring in 30–40% of patients with CRC (44). KRAS cooperates with the Wnt/β-catenin pathway to promote CRC and confers resistance to anti-EGFR antibodies (45).

The SLC25A22 process is required for the survival of CRC cells expressing activated KRAS, which are then rapidly incorporated into the tricarboxylic cycle (glutamine hydrolysis). In the absence of glutamine, cells can proliferate simply via incubation with succinate. The cells with mutated KRAS have a lower ratio of α-ketoglutarate to succinic acid, leading to reduced hypermethylation of 5-hydroxymethyl cytosine (a marker of DNA demethylation) and CpG sites. Numerous hypermethylated genes are located in the pro-cadherin gene cluster in the Wnt signaling pathway and chromosome 5q31. In CRC cells without KRAS mutations or with KRAS mutations and SLC25A22 knockout, the pro-cadherin gene expressed is not methylated at these sites (46).

In summary, in CRC cells expressing activated KRAS, SLC25A22 promotes the accumulation of succinic acid, resulting in increased DNA methylation, activation of Wnt signaling for increased expression of β-catenin, LGR5, proliferation, stem cell properties and resistance to 5-fluorouracil. Strategies that block this pathway may be used to treat CRC in the future (Fig. 2).

DNA methylation is a chemical modification that can alter genetic properties without altering the DNA sequence. The incidence of DNA methylation increases with aging, indicating that methylation may be related to aging and carcinogenesis (47). In the precursors of CRC (colorectal polyps) studies, similar abnormal methylation accumulation has been discovered in DNA sequences of patients, indicating that methylation biomarker detection in CRC could be a major trend in the future. Furthermore, it is considered that it is also possible to trace the origin of tumors through specific methylation of different types of tumors, which is helpful in determining the medication and other treatment strategies for patients. Interestingly, similar methylation also occurs in the colonic mucosae of healthy people, which means that the use of methylation biomarkers may predict the possibility of colon cancer (48). As aforementioned, it is considered by the authors that some of these genes, such as BRAF and KRAS, may provide a key breakthrough in methylation bioassay research.

Furthermore, MSI and CIN are two important types of CRC (49). MSI cancer is caused by the mutation of DNA mismatch repair (MMR) gene and the high methylation of promoter region. MMR is a protein that is produced by correcting the wrong base pair during the process of DNA replication. Therefore, detecting MMR status can predict the incidence of CRC to a certain extent.

In the study by Hinoue et al (50), CRC cases were divided into CIMP-high, CIMP-low, and non-CIMP types. Among them, cancer-related genes in the CIMP-high subgroup demonstrated hypermethylation and the BRAFV600 gene was mutated, while MLHL demonstrated hypermethylation. The CIMP-low subgroup is rich in KRAS mutation, which is characterized by DNA hypermethylation of CIMP-h-related marker subgroup.

Primarily, the study of Jensen et al (51) emphasized the potential utility of using the sensitivity and specificity of DNA methylation markers in blood samples to detect early tumors. Trimethoxy is a minimally invasive method for detecting early CRC, derived from the evaluation of three tumor specific DNA methylation markers in the blood. According to a study (52), the sensitivity of CRC cells to trimethoxy reached 78%, and the specificity was close to 100%. This method undoubtedly lights the way for the early prediction of CRC. Recently, according to Li et al (52), SEPT9, SDC2 and ALX4 methylation status can cover multiple molecular pathways of tumor formation, and combined detection will be hopeful to further improve the sensitivity of CRC detection.

Furthermore, a new combination of plasma DNA methylation-based biomarkers has now been developed and validated using clinical samples from multiple medical centers. It is expected to provide an alternative and cost-effective strategy for early detection of targeted gastrointestinal cancers (53).

Since the high status of MSI has been detected in numerous types of CRC, some scholars have proposed MSI status as a major marker of prognostic analysis (54). However, a meta-analysis has found that MSI-H is not a robust prognostic marker in stage I and stage IV CRC without immunotherapy (55). Interestingly, Popat et al (56) proposed MSI status as a prospective condition for patient management. Consequently, their research revealed that patients with MSI administered with fluorouracil exhibited better survival rates; however, explaining why MSI patients had an improved prognosis and the mechanism of Fu was a great challenge for later researchers. A previous study also posited that gene mutations such as those in the tumor suppressor gene TP53 and KRAS are rare in MSI tumors, and that these gene mutations could be associated with poor prognoses (57).

SMAD4 is a tumor suppressor and a component of the transforming growth factor (TGF)-β signaling pathway, which is associated with cell proliferation, differentiation, migration, and apoptosis. Methylation has been reported to lead to activation of the DPC4/SMAD4 gene, which plays an important role in the development of CRC. The study also pointed out that DPC4/SMAD4 is an important factor in prognostic analysis (58).

SDS2 methylation can be used as a potential biomarker to evaluate preoperative and postoperative fecal DNA in patients with CRC and can be used to determine whether methylated SDC2 in stool DNA returns to normal after surgical resection of CRC (59).

CRC is characterized by hypermethylation and overall hypomethylation of gene promoters. However, DNA methylation is a double-edged sword. It cannot only affect the development of tumor but is also a favorable treatment site. In the present study, several well-studied methylation targeted therapy sites were summarized.

Methylation of p16INK4a can be detected in ~30% of patients with CRC (60), and the mutation of p16INK4a leads to BRAF mutation, which has been proved to be a direct cause of tumorigenesis in the intestinal tract of mice (61). Therefore, p16INK4a can inhibit tumorigenesis, and targeted demethylation could potentially be a novel treatment method.

Hypermethylation of the RASSF1 promoter can be detected in 80% of CRCs. Interestingly, it has been revealed that knockout of the RASSF1 gene in mice increases the risk of CRC, which suggests that RASSF1 may be a tumor suppressor gene (62).

Studies have identified somatic mutations of Cdc7 in CRC, and clinical studies are ongoing. These findings emphasize the potential of Cdc7 in targeted therapy (63).

At the time of the formation of colon cancer, hypermethylation occurs at multiple gene sites on the DNA, resulting in the inactivation of important genes such as tumor suppressors. Therefore, DNA methyltransferase inhibitors can inhibit methylation at specific sites, correct wrong methylation modifications, and directly alter gene expression.

DNA methylation is a physiological process that regulates gene expression. The formation and maintenance of DNA methylation are realized under the action of DNMT. DNMT has three families, namely DNMT1, whose activity is accelerated during DNA semi-retention replication, and DNMT2 and DNMT3, which include DNMT3a, DNMT3b and DNMT3l. The human body possesses DNMT1, DNMT3a and DNMT3b.

DNMT3a and DNMT3b play important roles in catalytic de novo methylation during embryonic development. It has also been suggested that DNMT3a and DNMT3b may correct the errors left by DNMT1, as they are involved in maintaining DNA methylation patterns. The three DNMTs maintain DNA methylation and exhibit low expression in normal tissues and high expression in tumor tissues (64). Gradually, selecting DNMT as the target has become a new direction for drug research and development.

In recent years, Traditional Chinese Medicine (TCM) therapy has become increasingly popular. Since ancient times, there have been TCM compounds used to treat CRC, such as Wumei Wan and Sini Tang (65,66). Among them, numerous natural compounds have been revealed to treat CRC by acting on their methylated genes. Ginseng is one such case. It enhances apoptosis by regulating apoptosis-related genes in CRC cells and down-regulates the expression of DNMTs and reduces the global methylation level in CRC cells (67). Studies that have revealed promising results indicate the role of vitamins as DNA methylation modifiers, but studies with improved study designs are necessary (68).

A clinical study has revealed that 0.5–1 g of resveratrol administered orally daily produces anticancer effects in the human GI tract while being well tolerated in patients with cancer. It actually inhibits cancer spread by regulating epigenetic changes in tumor cells, through DNA methylation (69).

Curcumin from turmeric induces demethylation of specific CpG sites in CRC cells, but it does not induce global DNA methylation changes. Curcumin-induced methylation changes occur in a gene- and cell line-specific manner, and have a direct impact on the transcription of various genes involved in important biological processes (70). A number of experiments have revealed that some natural compounds can treat CRC through DNA methylation, proving that DNA methylation can be used as a target to treat CRC.

There are two classes of DNA methyltransferase inhibitors: Nucleoside and non-nucleoside, wherein the former includes 5-azacytosine and its derivatives, and the latter includes polyphenol epigallocatechin-3-gallate and rgl08 (71).

Although a variety of compounds can inhibit DNA methylation in mammalian cells, such as genistein (GE) and curcumin (72,73), the only widely tested DNA methylase inhibitors are ZCyd (5-azacytidine), DZCyd (5,6-dihydro-5-azacytidine, also known as DHAC) and ZdCyd (decitabine). These three compounds can inhibit DNA methylation only when incorporated into DNA. As aforementioned, ZdCyt is a more effective DNA methylation inhibitor than ZCyd because it binds only with DNA. Additionally, ZCyd is synthesized as a more stable analogue of ZCyd (74), which is at least one order of magnitude weaker than ZdCyt in blocking methylation in vivo (75). This is attributed to the limited incorporation of ZdCyt in DNA because it is inefficiently phosphorylated by cytidine kinase. Data, however, revealed that CRC can be treated with ZCyd.

ZNF671 has been discovered to be an important cancer inhibitor in a variety of tumors. High methylation level of ZNF671 gene promoter region was discovered in CRC, which was negatively correlated with ZN671 expression. It functions as a tumor suppressor in CRC through inactivation of Notch signaling. This suggests that ZNF671 can be used as a candidate target for the treatment of CRC (76).

Changes in DNA methylation may even be an important mediator in the transformation of metastatic cancer: Dramatic and reproducible changes in methylation have been described in several important longitudinal studies. Such studies have been completed for CRC (77), prostate (78) and breast cancers (79). E-cadherin (CDH1) is widely considered to be the target of DNA methylation for breast cancer metastasis (80). It is worth noting that DNA methylation of the metastasis suppressor gene is observed in circulating tumor cells from various tumor types (81). This increases the possibility that DNA methylation may also be a useful prognostic marker for metastasis. It is noteworthy that a pan-cancer meta-analysis identified the methylation site associated with metastasis of breast cancer and CRC (82). Therefore, the potential of methylated DNA as a marker for prognosis, diagnosis and metastasis is an area to be further explored.

Histologically, normal colonic mucosal tissues can be distinguished from colonic tissue with precancerous lesions at the molecular level. Therefore, the change in DNA methylation can be used as a marker of colon cancer. However, the reversibility of DNA methylation also provides a theoretical basis for the treatment of CRC from the perspective of epigenetics, and research revealed that the analyses of the DNA methylation results of specific genes are expected to provide clinicians with useful information such as early diagnosis and disease stages. Although these DNA methylation inhibitors have a potent demethylation effect, they can be accompanied by strong cytotoxicity and lack of specificity in the selection of action sites. Therefore, more studies are needed to solve these problems to facilitate the widespread use of DNA methylation inhibitors in the clinic (Table II).