This Division has been focusing on epigenetic mechanisms of carcinogenesis, represented by aberrant DNA methylation. Using our original genome-scanning technique for differences in DNA methylation, methylation- sensitive-representational difference analysis (MS-RDA), many aberrantly methylated DNA fragments were identified in various cancers, i.e. gastric cancers, breast cancers, pancreatic cancers, lung cancers, ovarian cancers, neuroblastomas, and melanomas. These led to identification of a novel tumor-suppressor gene in gastric cancers, development of a novel and powerful prognostic marker for neuroblastomas, and revelation of an "epigenetic field for cancerization". This Division is now trying to develop these findings into clinical markers, in addition to continuing to identify aberrant DNA methylation in cancers. It is also interested in the elucidation of the molecular mechanisms and etiology of induction of aberrant methylation.

In human gastric cancers, it had been known that most tumor-suppressor genes are more frequently inactivated by DNA methylation than by mutations. By quantifying the amounts of aberrantly methylated DNA molecules in healthy volunteers with and without current Helicobacter pylori (HP) infection, it was found that HP infection induces aberrant methylation in the gastric mucosa. Among individuals without current HP infection, an increasing trend in methylation levels was found from the gastric mucosa of healthy volunteers to that of cases with a single gastric cancer and then that of cases with multiple gastric cancers. These findings showed that epigenetic alterations have already accumulated earlier in the gastric mucosae of gastric cancer cases, which can be termed as an epigenetic field for cancerization (epigenetic field defect) ⁽¹⁰⁴⁾. This Division is now trying to use the presence of an epigenetic field for cancerization as a cancer risk marker.

An association between the methylation levels in non-cancerous gastric mucosae and cancer tissues was further investigated using 66 paired samples. Surprisingly, the number of methylated CpG islands (CGI) in a cancer was not associated with the methylation levels in the background mucosa. This was considered to be because the methylation analyzed in the gastric mucosa was not necessarily that in the cancer precursor cells ⁽¹⁰⁵⁾.

Aberrant methylation of a specific gene or a methylation profile in cancers can be associated with clinically relevant information, such as patient prognosis or response to a specific therapy.

This Division previously demonstrated that the CpG island methylator phenotype (CIMP) in neuroblastomas (NBLs) was closely associated with poor overall survival in Japanese cases. Considering that ethnic differences could be present and that, generally, genome-wide screenings tend to produce "too good" results, German NBL cases were examined. The strong association between the CIMP and poor survival was faithfully reproduced, including in cases without N-myc amplification, and CIMP is, therefore, considered to be a strong and universal prognostic marker for NBL cases ⁽¹⁰⁶⁾.

Human melanomas were analyzed by a chemical genomic screening using the demethylating agent, 5-aza-2'-deoxycytidine, and an expression microarray. Eighteen genes were identified as silenced in melanoma cell lines, including the Tissue factor pathway inhibitor-2 gene (TFPI-2), which is involved in repression of the invasive potential of malignant melanomas. TFPI-2 was methylated in 5 of 17 metastatic site specimens, while it was not in any of 20 primary site specimens. TFPI-2 silencing was suggested to be involved in melanoma metastasis ⁽¹⁰⁷⁾.

Human ovarian cancers were analyzed by MS-RDA, and 33 CGIs in putative promoter regions were identified as methylated in ovarian cancer cell lines. Among these, PRTFDC1 was silenced, and GPR150, ITGA8, and HOXD11 were methylated in primary ovarian cancers ⁽¹⁰⁸⁾.

Regardless of the fact that aberrant DNA methylation is widely involved in human cancers and possibly in other human disorders, very limited information is available on the mechanisms of how aberrant DNA methylation is induced and what factors are involved. There seem to be diverse mechanisms, such as HP infection and the one involved in the CIMP of NBLs.

This Division revealed a unique mechanism for CIMP in gastric cancer cell lines. Some cancer cell lines showed increased rates of methylation of scattered CpG sites within a CGI, denoted as "methylation seeds". It was previously shown that increased production of methylation seeds can lead to dense methylation of an entire CGI. This year, it was shown that increased production of methylation seeds can override gene transcription, which is known to confer resistance to methylation, and is associated with a larger number of methylation-silenced genes ⁽¹⁰⁹⁾.

As factors that induce aberrant DNA methylation, only a limited number of chemicals are known ⁽¹¹⁰⁾. There are many chemicals that can induce cancers without inducing mutations, and some of them can be carcinogens with epigenetic actions. The lack of efficient screening systems for such chemicals may be the reason for the limited information. Screening systems for chemicals with epigenetic actions are being constructed.

To identify factors that promote or repress induction of aberrant DNA methylation in vivo, animal models with defined methylation-silenced genes are indispensable. Appropriate animal models are being explored.

Cyclin D1 expression and localization were immunohistochemically analyzed in a gastric cancer model in collaboration ⁽¹¹¹⁾.