Page 179 - Molecular features of low-grade developmental brain tumours

mutations beyond TSC1/TSC2 most likely do not contribute to development and growth of SEGAs. Alternatively, changes at the methylation level and/or changes in gene expression might play a more prominent role.
DNA methylation profile and transcriptional landscape of SEGAs
Changes in DNA methylation have been well studied in CNS tumours and have also been seen in neuro-psychiatric diseases such as autism spectrum disorder, epilepsy and TSC 27,45- 48. Previous studies showed that SEGAs are a unique entity among CNS tumours based on their methylation profile, but did not study the molecular mechanisms targeted by these methylation changes 27,49,50. Additionally, gene expression studies on SEGA are limited and only focus on the expression of protein-coding genes using either a microarray or RNA- sequencing (RNA-Seq) 27,51. In chapter 3 and 4 we investigated the methylations patterns and the coding and non-coding landscape of SEGAs in TSC, respectively and show that the methylation profile as well as the transcriptome of SEGAs are enriched for GO terms including the adaptive immune system, T cell activation, leukocyte mediated immunity, extracellular structure organization and the ERK1 & ERK2 cascade. Several of the enriched pathways found in chapter 4 are also related to the biological processes found in previous transcriptome-based SEGA studies 27,51. Analysis at the protein level confirms the presence of inflammation markers in chapters 2 & 3, activation of ERK/MAPK in chapter 4 and the dysregulation of extracellular matrix (ECM) proteins in chapter 5. Our methylation data suggests that these biological processes might already be affected on DNA level and could therefore be important drivers in SEGA pathogenesis. Since these pathways are also affected in cortical tubers it could be of interest to investigate these methylation patterns in other TSC lesions 27,52,53. In unpublished data from our laboratory, we see that in tubers, several genes, including IL18, IL21R & IL6ST are hypo methylated indicating that also in tubers, genes related to the adaptive immune system are affected on methylation level. However, the overall methylation changes in tubers are minor (Figure 1). By investigating the mTOR pathway directly we found only a few genes differentially expressed or methylated in SEGAs compared to control tissue. Furthermore, GO term analysis on the differentially methylated genes between SEGAs and controls did not reveal the mTOR pathway, indicating that DNA methylation changes most likely do not contribute directly to the mTOR activation in SEGA.
Interestingly, we identified two subgroups in the SEGA methylation data with one group further subdividing into two smaller groups. Reanalyzing the methylation data with an extended cohort showed that the two groups appear to be robust (Figure 2). However, no correlation was found between the methylation and RNA expression of genes that were differentially methylated between the two groups. Furthermore, in chapter 4 we did not find any subgrouping based on the coding and small non-coding transcriptome of SEGAs. However, due to the complexity of regulating RNA expression, methylation changes might not be directly reflected in RNA expression data. The differentially methylated genes between the two groups are enriched for the GO terms related to the adaptive immune response and the MAPK cascade. Additionally, differences in expression of the T cell marker CD3 were
GENERAL DISCUSSION
177
7