In this study, we demonstrated a unique DNA methylation pattern along the promoter CpG island of the MT3 gene. The MT3 CpG island demonstrated variable degrees of methylation. Interestingly, several CpG nucleotides were highly methylated in tumor and normal samples (from −372 to −306) whereas CpG nucleotides closest to the TSS (from −4 and +3) remained unmethylated in all normal and most tumor samples. On the other hand, CpG nucleotides in two regions (from −139 to −49 and +296 to +344) were more significantly hypermethylated in EACs as compared to normal samples (FDR<0.001, −log10(FDR)>3.0). The methylation pattern in BE was similar to that in EACs; however, BE had an extended region with unmethylated CpG nucleotides around the TSS (from −69 to +3). These results suggest the existence of some protective mechanism(s) that prevent the core region of the TSS from methylation in normal conditions 
. Although the mechanisms explaining progressive methylation patterns remain unclear, recent studies suggested that dimethylation of lysine 4 of histone 3 (H3K4me2) may protect against DNA methylation 
. Our ChIP results also support this hypothesis by showing a high level of H3K4me2 in an active MT3 promoter (in particular, in ChIP 2) in the FLO-1 cell line and a significantly decreased level of H3K4me2 in a suppressive promoter in OE33.
The unexpected variable methylation levels in the MT3 promoter CpG island highlight the critical importance of selecting the proper promoter sites for DNA methylation studies in cancer. Comprehensive analysis of the promoter regions using quantitative techniques such as Pyrosequencing may be important for identifying crucial promoter region(s) that regulate gene expression. This is best exemplified by our results in the region from −372 to −306, where DNA methylation was high in all samples including normal tissues without any correlation with the gene expression levels. On the other hand, the strongest inverse correlation between DNA methylation and gene expression was present in R2 (from −127 to −8), suggesting that this region may be critical in regulating MT3 gene expression. Significance analysis demonstrated that R2 has the smallest false discovery rate (FDR<0.001, −log10(FDR)>3) in differentiating EACs from both NS and NG. The difference between BE and normal tumors (NS+NG) and between EACs and BE were also statistically significant (p<0.01, respectively). Taken together, our findings suggest that R2 is the most suitable region to examine DNA methylation that possibly plays a critical role in transcription regulation of MT3. We, and others, have shown that DNA methylation could be an early event in carcinogenesis 
. Our analysis of a few BE samples that were available for DNA methylation analysis suggested that R1 and R3 of the MT3 promoter may be early events occurring in BE. However, a study using a large panel of precancerous lesions, including Barrett's dysplasia samples, is needed to fully understand the progressive mode of MT3 methylation in the early stages of Barrett's carcinogenesis.
Using an in vitro
cell model of EAC, we confirmed that DNA methylation is the mechanism underlying MT3 gene silencing. In a cell model of OE33, the treatment with 5-Aza demethylated the promoter region and restored MT3 gene expression, concordant with many other reports 
. Histone modifications including histone acetylation and methylation are important regulators of gene expression. Generally, active chromatin is associated with histone acetylation of H3K9 and methylation of H3K4, H3K36 and H3K79 while repressive chromatin is associated with histone deacetylating of H3K9 and methylation of H3K9, H3K27 and H4K20 
. The final outcome of gene expression may be determined by the overall changes of these modifications, in addition to DNA methylation. We carried out ChIP assay followed by qPCR using two esophageal adenocarcinoma cell lines; FLO-1, which expresses MT3 and has a low level of DNA methylation; and OE33, which has hypermethylated promoter and silenced MT3. Analysis of three different regions of the MT3 promoter demonstrated significantly high levels of H3K4me2 and H3K9ace, and low levels of H3K9me2 in all three regions in FLO-1 cells. OE33 showed the converse results with an increased ratio of repressive to active histone H3K9 (H3K9me2/H3K9ace). The 5-Aza administration in the methylated OE33 cells reversed the ratio of repressive to active histone H3K9, showing higher levels of H3K9ace than H3K9me2. These results indicate that DNA methylation of the MT3 gene is linked with unique histone modifications in regulating gene expression. Surprisingly, the H3K4me2 level did not disappear completely, even in the highly methylated OE33 promoter. After 5-Aza treatment of OE33 cells, H3K4me2 showed a significant increase only in the ChIP3 region not in the ChIP2 region where the strongest correlation between DNA methylation and MT3 expression was observed. These results suggest that H3K4me2 may not be directly associated with MT3 gene expression. While some reports indicated that H3K4me2 is associated with active chromatin 
others demonstrated that H3K4me2 could be present on poised and inactive genes 
. The histone modifications and their relationship with DNA methylation and gene expression are quite complex 
. Further studies are needed to understand the complexity of epigenetic mechanisms in transcription regulation. Interestingly, we found that DNA methylation of the MT3 promoter region (R2, from −127 to −8 sites) was significantly associated with advanced tumor stages and lymph node metastasis (), indicating that dysfunction of MT3 through DNA methylation of the core promoter region may be associated with tumor progression. These results further confirm the importance of selecting the appropriate promoter regions for the study of DNA methylation in cancer.
In conclusion, we demonstrated frequent silencing of MT3 expression in esophageal adenocarcinomas due to unique DNA methylation changes and histone modifications. The biological functions of MT3 in regulating ROS and oxidative stress in other diseases call for investigations along this line in esophageal carcinogenesis. The choice of a promoter region to examine DNA methylation is a critical step that should be carefully considered in molecular studies of carcinogenesis and in the development of early detection biomarkers.