Milea Timbergen

104 methylation has been the subject of various sarcoma-related publications. A study by Röhrich et al. based the classification of peripheral nerve sheath tumours (benign versus malignant) on methylation patterns 39 . Tombolan et al. was able to distinguish metastatic and non-metastatic rhabdomyosarcomas based on their methylation patterns 18 . In this study, we hypothesized a role for aberrant methylation patterns based on the differences in clinical behaviour between the different CTNNB1 mutations found in DTF. To our knowledge, this is the first study that explored DNA methylation patterns in DTF. Whole genome DNA methylation was examined using MeD-seq. This is a novel and powerful technique to perform genome-wide DNA methylation analysis 28 . Since this technique focusses the sequencing resources on methylated regions only, and because the restriction enzyme LpnPI is restricted by a short template size, the generated fragments are consistent in size which results in accurate identification of DMRs genome-wide 28 . The MeD-Seq method compared very favourable to other methods such as whole-genome bisulfite sequencing, MeDIP and the 450K Infinium bead-chip technology 28 . Furthermore, MeD-seq uses single base pair resolution which allows us to identify methylation on one specific CG site. In case of marker development, this would create the opportunity to use information from this single nucleotide for primer selection. Here we focused on differences between CTNNB1 T41A and S45F mutated tumours as they occur frequently, are mutually exclusive and are reported to exhibit a divergent clinical behaviour. Overall, the detected differences in DNA methylation were few, subtle and unable to discriminate between S45F and T41A tumours in a cluster analysis. Only some DMRs were found to be differentially methylated with a fold change ≥ 1.5, and only a single DMR, related to the CCDC6 gene had a fold change of ≥ 2. Most of the DMRs (fold change ≥ 1.5) appeared to be situated within gene bodies. The relatively small fold changes in DMRs observed, the intertumoural heterogeneity and the low amount of statistically significant DMRs identified in this study, suggests that there is no distinct difference in DNA methylation patterns between S45F and T41A DTF tumours. In the current study the effects of differential methylation on gene expression were assessed using an independent Affymetrix mRNA expression dataset which only revealed significant expression differences between S45F and T41A tumours for CCDC6 (p = 0.034) and FOXK2 (p = 0.004) but not for all probes capable of detecting these genes. To explain these observations, one may speculate that gene body methylation affects differential splicing yielding splice products that hybridize only with some capture probe. To obtain biological insight why different β-catenin mutants would affect DNA methylation differently, the interaction between 4