Investigating the role of DNA methylation in pluripotency and differentiation using an embryoid body model
Epigenetics is the study of heritable alterations in phenotype caused by changes in cellular properties, but where the genotype is unchanged. At the molecular level these changes include chemical modifications of DNA and histones in chromatin. Specific chromatin states are associated with gene activity or silencing. The proper functioning of these mechanisms is critical for mammalian survival, particularly during embryonic development. One of the best studied epigenetic modifications is DNA methylation, wherein methyl-groups are placed on cytosines in CpG dinucleotide contexts by DNA methyltransferases to form 5mC. The malfunction of this mechanism is associated with failure of embryogenesis and many adult human disease pathologies, including cancer. However, questions remain about how the 5mC patterns are established de novo, how the patterns can change between different cell types, and why some cell types can tolerate the absence of 5mC but not others. Genetic removal of the de novo (Dnmt3a and Dnmt3b) or maintenance methyltransferases (Dnmt1) in somatic cells can lead to cell death. However, mouse embryonic stem cells (ESCs) can proliferate normally in the absence of all three of these proteins. This suggests that DNA methylation becomes essential at some point after the cells exit pluripotency and begin differentiation. Being able to identify this window of time during development is important for better understanding the dynamics of recruitment of these proteins and deposition of the mark and could therefore provide insight into how their misregulation contributes to disease. In this thesis I investigate the role of DNA methylation during ESC differentiation using a combination of conditional and reporter cell lines, in vitro differentiation models, RNA sequencing, genetic engineering, and high-resolution imaging. The ESC lines include those in which i) the expression of Dnmt1 and key pluripotency genes can be tuned separately and in combination, ii) contain germ layer differentiation reporters, and iii) contain reporters of 5mC distribution. I allow these cells to differentiate to form embryoid bodies (EBs), a validated embryogenesis-like model that enables simulation of the pluripotent-to-differentiated transition in vitro. By combining these approaches, I was able to investigate the impact of loss of Dnmt1 on gene expression pathways including apoptosis, primordial germ cell (PGC) and 2C-like cell formation, germ layer differentiation, and changes in transposable element expression. I was also able to delineate differentiation trajectories by comparing my bulk RNA sequencing data with published single cell RNA sequencing data. Overall, I observed that inhibition of Dnmt1 activity consistently led to a significant reduction in EB size though germ layer differentiation was still able to occur. Likewise, EBs were still able to form in the absence of master pluripotency factor Oct4, although to a reduced capacity. Loss of both proteins led to smaller EBs than the wild type. Interestingly Oct4-/-, Dnmt1+/+ EBs were the most affected. There were no significant changes in frequency of apoptotic cells, and only LTR-family transposable elements were de-repressed. By comparing data, I was able to identify that loss of Dnmt1 enriched EBs for PGC-like cell marker genes at late stages of differentiation. I conclude that under normal differentiation initiation conditions the two systems, the pluripotency network and the DNA methylation network, work synergistically to control the gradual switch of cells from the pluripotent to the differentiated state enabling the formation of a limited population of PGC-like cells prior to advanced differentiation of germ layer cell lineages. The purpose of this may be to protect emerging PGC-like cells from Transposon Element activity. It may also be to allow the number of pluripotent cells to reach a threshold level prior to initiating differentiation to prevent the EB size being limited in a mechanism similar to that suggested for primordial dwarfism.