Toshikazu Ushijima
National Cancer Center Research Institute
Allele-specific DNA methylation is an important mechanism of genomic imprinting. To make a genome scanning for regions differentially methylated in two genomes, methylation-sensitive representational difference analysis (MS-RDA) was developed (Ushijima et al., PNAS 94: 2284, 1997). Each of two sources of genomic DNA is digested with a methylation-sensitive restriction enzyme, HpaII, and PCR is performed to amplify the entire restriction product using a universal adaptor and primer. When a specific genomic region is unmethylated in one genome, DNA fragments whose sizes are suitable for PCR will be produced by the restriction digestion and amplified by PCR. On the other hand, when the region is methylated in the other genome, no fragments suitable for PCR will be produced and no fragments from the region will be present in the PCR product. These steps convert a difference in the methylation of a specific genomic region into a difference in the presence/absence of DNA fragments derived from the region. The difference in the presence/absence can be easily identified by the following genomic subtraction technique, RDA (Lisitsyn et al., Science 259: 946, 1993). MS-RDA was performed to isolate DNA fragments that are unmethylated in a normal human bronchial cell line (NHBE) and methylated in human lung cancer cell lines. A total of 16 DNA fragments that were flanked by CpG islands and hypermethylated also in primary lung cancers were isolated. Among the 16 fragments, five were located in the vicinities of known genes. The expression levels of the five genes were analyzed, and those of EDN-1, MEIS1 and 5HTR1B were shown to be reduced in human lung cancers. Bisulfite sequencing showed that a CpG island in intron 6 of MEIS1 was methylated in almost half of the DNA molecules in NHBE and in almost all of the DNA molecules in EBC-1 and LK-2 carcinoma cell lines. A polymorphism in the region is being searched to determine the allelic status of the methylation. MS-RDA was shown to detect differences between 50% and 100% in the number of methylated DNA molecules, and is expected to be useful in finding out differentially methylated genomic regions.