DNMT3A splice variants and DNA methylation in normal and malignant hematopoiesis

DNA methylation (DNAm) is an epigenetic process that guides development and cellular differentiation. Particularly the de novo DNA methyltransferase 3A (DNMT3A) plays a pivotal role for differentiation of hematopoietic stem and progenitor cells (HSPCs), clonal hematopoiesis, and development of blood disorders. It is known that DNMT3A can be spliced in a disease- and tissue-specific manner, but it is unclear if different splice variants have a specific function in normal or malignant hematopoiesis. Acute myeloid leukemia (AML) patients often harbor a mutation in DNMT3A and their epigenetic makeup is largely dysregulated. This was used to develop biomarkers based on DNAm for identification and classification of AML patients. In this thesis, it was shown that modulation of DNMT3A variants caused transcript-specific DNAm and gene expression changes. While transcript 2 seemed to be important for the loss of a primitive immunophenotype, transcript 4 was relevant for the erythroid differentiation. Furthermore, changes caused by modulation of transcript 2 were reflected in AML and its expression correlated with survival of AML patients. It was shown that DNMT3A expressed also a novel fifth transcript and expression of different splice variants was altered in myeloid malignancies. Dysregulated DNAm in AML patients was utilized to develop an epigenetic biomarker based on a single CpG site located in C1R to support outcome prediction of AML patients. DNAm at C1R seemed to be of independent prognostic relevance and might be relevant for other cancers. Moreover, four AML-associated regions were selected to reliably discern between AML patients and healthy individuals based on single CpG sites or DNAm patterns of neighboring CpGs of those four regions. Finally, a technique for determining absolute blood cell counts was developed for fresh or frozen blood samples and is based on a non-methylated reference DNA that is mixed with the sample. Altogether, we demonstrated that DNMT3A splice variants play an important role for differentiation of HSPCs, target different genomic regions and particularly transcript 2 is relevant for AML. Furthermore, it was shown that DNAm can be used to develop biomarkers that support identification and stratification of AML