The immediate goal of this project was to investigate differences in DNA methylation between fusion-positive and fusion-negative rhabdomyosarcoma (RMS) tumors. In previous studies, we used Illumina methylation arrays to examine DNA methylation in 20 fusion-positive and 17 fusion-negative RMS samples. Unsupervised hierarchical clustering and principal component analyses showed that the RMS tumors clustered into two groups consisting exclusively of fusion-positive or fusion-negative cases. We also found a similar clustering in a panel of 5 fusion-positive and 5 fusion-negative RMS cell lines. A supervised analysis identified probes that were significantly hypermethylated and other probes that were significantly hypomethylated in fusion-positive compared to fusion-negative RMS tumors. To investigate whether any methylation difference represents an aberrant event in fusion-positive or fusion-negative RMS, we compared methylation in RMS with two normal tissues, skeletal muscle and bone marrow. Using the CpG sites that are differentially methylated between the two RMS subtypes, hierarchical clustering revealed that the normal tissue samples were tightly clustered, and were grouped together on one main branch with the fusion-negative RMS samples while all fusion-positive samples were on the other main branch. Though the methylation status of numerous differentially methylated sites is similar in fusion-negative tumors and the normal tissues, there are also differentially methylated sites for which the methylation status in fusion-positive tumors closely resembles either or both normal tissues. These methylation studies show similarities of both fusion subtypes to normal tissues for different sets of genes, and we propose that there are aberrant hypo- and hypermethylation events occurring in fusion-positive tumors, and other aberrant hyper- and hypomethylation events occurring in fusion-negative tumors. To further examine the relationship between methylation and expression, we treated 6 fusion-positive and 5 fusion-negative RMS lines with the DNA methyltransferase inhibitor 5-aza-2'-deoxycytidine (aza-dC). We focused on the EMILIN1 gene and used pyrosequencing to find very high (80%) DNA methylation levels in 5 of 6 control-treated fusion-positive lines and only 1 of 5 control-treated fusion-negative lines. After aza-dC treatment, there was a significant decrease in EMILIN1 promoter methylation levels in the 1 fusion negative and 5 fusion-positive lines. Quantitative RT-PCR analysis showed that there are very low EMILIN1 mRNA levels with control treatment and significantly increased mRNA levels with aza-dC treatment of the same 1 fusion negative and 5 fusion-positive lines. These findings indicate that EMILIN1 promoter methylation and expression are inversely correlated, and that loss of methylation results in increased EMILIN1 mRNA expression.
