Human malignant tumors are characterized by changes in the patterns of DNA methylation, which include a globally hypomethylated tumor cell genome and the focal hypermethylation of numerous CpG islands; many of them are associated with gene promoters [
38]. Promoter hypermethylation events can lead to silencing of genes functioning in tumor-relevant pathways. Together with our collaborators at Guangxi Medical University, we examined aberrant DNA methylation of nasopharyngeal carcinoma tissues of patients in the endemic area of Southern China. We used mRNA microarrays to analyze gene expression changes in human nasopharyngeal carcinoma cell lines treated with a demethylation reagent, 5-aza-2′-deoxycytidine. Among the upregulated candidate genes, we confirmed promoter DNA hypermethylation of RRAD (RAS associated with diabetes) in biopsy specimens and cancer cell lines. Transfection of RRAD in cancer cells suppressed cell proliferation, colony formation, and migration, suggesting that RRAD has tumor-suppressive functions in nasopharyngeal carcinoma [
39]. We also detected downregulation of RERG (RAS-like estrogen-regulated growth inhibitor) by DNA methylation in the cancer tissues. RERG-overexpressing cells showed significantly slower growth and less angiogenesis in tumor xenografts in nude mice [
40]. DNA methylation occurs in
Helicobacter pylori infection-related gastric cancer [
41,
42] and ulcerative colitis-associated colorectal cancer [
43], indicating that DNA methylation is a key event of inflammation-related carcinogenesis.
Profiling of DNA methylation across the genome is extremely important to understanding the influence of epigenetics. There has been a revolution in DNA methylation analysis technology over the past decade, and analyses that previously were restricted to specific loci can now be performed on a genome scale [
44]. The main principles of DNA methylation analysis are restriction enzyme-based, affinity-based, and bisulfite deamination-based pretreatments, followed by the analytical methods such as endpoint PCR, real-time PCR, microarray, and next-generation sequencing [
44]. In our early studies, we used bisulfite deamination-based methods, such as methylation-specific PCR (MSP) and bisulfite genomic sequencing (BGS) [
39,
45]. Genomic DNA was treated by sodium bisulfite to deaminate unmethylated cytosine to uracil (finally detected as thymine), but methylated cytosine (5-methylcytosine) was not affected (detected as cytosine). Based on this sequence change, DNA methylation was analyzed by endpoint PCR (MSP) or Sanger sequencing of amplified bisulfited DNA (BGS). Nowadays, whole-genome bisulfite sequencing can bioinformatically lift out regions of interest, and sequencing costs are reduced. However, it remains expensive to achieve sufficient sequencing depth for high accuracy, and it is difficult to apply this technique to small samples. We recently explored methylated tumor suppressor genes using affinity-based enrichment with a methyl-CpG-binding domain protein, followed by a next-generation sequencer (methyl-capture sequencing) [
46]. This method can be applied to relatively small samples at less cost. Combined with the gene expression microarray, we identified candidate genes that are silenced by promoter DNA methylation in cancer tissues. We quantified the methylation rates of the candidate genes by bisulfite amplicon sequencing (BAS, also known as amplicon bisulfite sequencing: AmpliconBS). BAS involves targeted sequencing of PCR amplicons generated from bisulfite-deaminated DNA. It is a flexible, cost-effective way to study methylation of a sample at single CpG resolution and to perform subsequent multi-target, multi-sample comparisons [
47]. The BLUEPRINT consortium evaluated BAS as one of the best all-round methods for use in DNA methylation assays in large-scale validation studies, biomarker development, and clinical diagnostics [
48]. We found several differentially methylated candidate genes in nasopharyngeal carcinoma tissues compared to normal nasopharynx tissues that may be useful biomarkers for cancer screening [
46]. Among these candidates, we measured the methylation rates of RERG in nasopharynx biopsy specimens by using restriction enzyme-based real-time PCR, which is more convenient than BAS. The methylation rate of RERG in cancer tissues was significantly higher than that in normal tissues, with 78% sensitivity and 100% specificity to screen nasopharyngeal carcinoma [
40]. It is advantageous to apply this biomarker for less-invasive specimens.