Successful mammalian development to term requires that embryonic and extra-embryonic tissues communicate and grow in coordination, to form the body. After implanting into the uterus, the mouse embryo is comprised of three cell lineages: first, the embryonic epiblast (EPI) that forms the embryo proper, second, the extra-embryonic ectoderm (ExE) which contributes to the foetal portion of the placenta, and third, the visceral endoderm (VE) that contributes to the yolk sac. These three tissues form a characteristic ‘egg-cylinder’ structure, which allows signals to be exchanged between them and sets the stage for body axis establishment and subsequent tissue patterning. The mechanisms underlying this process are difficult to study in vivo because a different genetically manipulated mouse line must be generated to investigate each factor involved. This difficulty has prompted efforts to model mammalian embryogenesis in vitro, using cell lines, which are more amenable to genetic manipulation. The pluripotent state of the EPI can be captured in vitro as mammalian embryonic stem cells (ESCs). Although mouse ESCs have been shown to contribute to all adult tissues in chimeric embryos, they cannot undertake embryogenesis when allowed to differentiate in culture. Previous studies have shown that ESCs formed into three-dimensional (3D) aggregates, called embryoid bodies, can become patterned and express genes associated with early tissue differentiation. However, embryoid bodies cannot recapitulate embryonic architecture and therefore may not accurately reflect what happens in the embryo. In this study, a new technique was developed to model early mouse development which is more faithful to the embryo. ESCs were co-cultured with stem cells derived from the ExE, termed trophoblast stem cells (TSCs), embedded within extracellular matrix (ECM). These culture conditions lead to the self-assembly of embryo-like structures with similar architecture to the mouse egg cylinder. They were comprised of an embryonic compartment derived from ESCs abutting an extra-embryonic compartment derived from TSCs, and hence were named ‘ETS-embryos’. These structures developed a continuous cavity at their centre, which formed via a similar sequence of events to those that lead to pro-amniotic cavity formation in the mouse embryo, and required active Nodal/Activin signalling. After cavitation, ‘ETS-embryos’ developed regionalised mesodermal tissue and primordial germ cell-like cells originating at the boundary between embryonic and extra-embryonic compartments. Inhibitor studies revealed that this occurred in response to endogenous Wnt and BMP signalling, pathways which also govern these tissue specification events in the early mouse embryo. To demonstrate that ‘ETS-embryos’ were comparable to mouse embryos at the global transcriptional level, RNA-sequencing was then performed on different tissue regions of ‘ETS-embryos’ and the resulting transcriptomes were compared to datasets from mouse embryos. These data showed that ‘ETS-embryos’ were highly similar to mouse embryos at post-implantation stages in their overall gene expression patterns. Taken together, these results indicate that ‘ETS-embryos’ are an accurate in vitro model of mammalian embryogenesis, which can be used to complement studies undertaken in vivo to investigate early development.

DNA methylation entails the addition of a methyl group to the 5-carbon of the cytosine base of the DNA. This modification is important during many biological processes such as imprinting, X-chromosome inactivation, cell differentiation as well as silencing of transposable elements (TEs). DNA methylation is dynamic during early mammalian development, despite being a more static mark in somatic cells. Global hypomethylation is a hallmark of epigenetic reprogramming in mammalian primordial germ cells (PGCs), the early embryo and in naïve embryonic stem cells (ESCs). Genome integrity is crucial during early development, as the germline DNA needs to be protected for future generations. Therefore, epigenetic reprogramming presents a critical phase for TE defence since presumably alternative silencing pathways need to be employed to limit their activity. In this thesis, I investigate the role of small RNAs to control TEs during global waves of DNA demethylation in cellular reprogramming, naïve pluripotency as well as early mammalian development. Following an introduction to the research questions, in chapter 3 I investigate the mechanism of TE regulation in an in vitro model of Dnmt1 deletion in mouse ES cells to recapitulate in vivo epigenetic reprogramming. I find that certain classes of TEs become transcriptionally upregulated and subsequently resilenced by a mechanism independent of DNA methylation. I identify ARGONAUTE 2 (AGO2) bound siRNAs as the prominent mechanism to control certain classes of TEs, while others appear to be regulated by redistribution of repressive histone modifications. In chapter 4, I construct Dicer constitutive and conditional KO ESCs in the background of the Dnmt1f l/f l ESCs using CRISPR-Cas9. I dissect the role of DNA methylation and of DICER dependent small RNAs on transcriptional changes of ESCs. Additionally, I find that DICER dependent small interfering RNAs (siRNAs) re-silence transcriptionally active TE classes. Finally, in chapter 5, I examine the role of small RNAs in TE silencing in different models of global hypomethylation in vivo and in vitro PGCs, during iPSC reprogramming and in a transition from serum to 2i culturing of mouse ESCs.

<p>Mammalian development is orchestrated by global transcriptional changes, which drive cellular differentiation, giving rise to diverse cell types. The mechanisms that mediate the temporal control of early differentiation can be studied using embryonic stem cell (ESCs) and embryonal carcinoma cells (ECCs) as model systems. In these stem cells, differentiation signals induce transcriptional repression of genes that maintain pluripotency (PpG) and activation of genes required for lineage specification. Expression of PpGs is controlled by these genes’ proximal and distal regulatory elements, promoters and enhancers, respectively. Previously published work from our laboratory showed that during differentiation of ESCs, the repression of PpGs is accompanied by enhancer silencing mediated by the Lsd1/Mi2-NuRD-Dnmt3a complex. The enzymes in this complex catalyze histone H3K27Ac deacetylation and H3K4me1/2 demethylation followed by a gain of DNA methylation mediated by the DNA methyltransferase, Dnmt3a. The absence of these chromatin changes at PpG enhancers during ESC differentiation leads to their incomplete repression. In cancer, abnormal expression of PpG is commonly observed. Our studies show that in differentiating F9 embryonal carcinoma cells (F9 ECCs), PpG maintain substantial expression concomitant with an absence of Lsd1-mediated H3K4me1 demethylation at their respective enhancers. The continued presence of H3K4me1 blocks the downstream activity of Dnmt3a, leading to the absence of DNA methylation at these sites. The absence of Lsd1 activity at PpG enhancers establishes a “primed” chromatin state distinguished by the absence of DNA methylation and the presence of H3K4me1. We further established that the activity of Lsd1 in these cells was inhibited by Oct3/4, which was partially repressed post-differentiation. Our data reveal that sustained expression of the pioneer pluripotency factor Oct3/4 disrupts the enhancer silencing mechanism. This generates an aberrant “primed” enhancer state, which is susceptible to activation and supports tumorigenicity. </p> <p>As differentiation proceeds and multiple layers of cells are produced in the early embryo, the inner cells are depleted of O₂, which triggers endothelial cell differentiation. These cells form vascular structures that allow transport of O₂ and nutrients to cells. Using ESC differentiation to endothelial cells as a model system, studies covered in this thesis work elucidated a mechanism by which the transcription factor Vascular endothelial zinc finger 1 (Vezf1) regulates endothelial differentiation and formation of vascular structures. Our data show that Vezf1-deficient ESCs fail to upregulate the expression of pro-angiogenic genes in response to endothelial differentiation induction. This defect was shown to be the result of the elevated expression of the stemness factor Cbp/p300-interacting transactivator 2 (Cited2) at the onset of differentiation. The improper expression of Cited2 sequesters histone acetyltransferase p300 from depositing active histone modifications at the regulatory elements of angiogenesis-specific genes that, in turn, impedes their activation. </p> <p>Besides the discovery of epigenetic mechanisms that regulate gene expression during differentiation, our studies also include development of a sensitive method to identify activities of a specific DNA methyltransferase at genomic regions. In mammals, DNA methylation occurs at the C5 position of cytosine bases. The addition of this chemical modification is catalyzed by a family of enzymes called DNA methyltransferases (Dnmts). Current methodologies, which determine the distribution of Dnmts or DNA methylation levels in genomes, show the combined activity of multiple Dnmts at their target sites. To determine the activity of a particular Dnmt in response to an external stimulus, we developed a method, Transition State Covalent Crosslinking DNA Immunoprecipitation (TSCC-DIP), which traps catalytically active Dnmts at their transition state with the DNA substrate. Our goal is to produce a strategy that would enable the determination of the direct genomic targets of specific Dnmts, creating a valuable tool for studying the dynamic changes in DNA methylation in any biological process.</p>