Pluripotency is defined as the capacity to differentiate into cells from each of the three primary germ layers, the ectoderm, mesoderm and endoderm. This is a property of cells located in the inner cell mass (ICM) of preimplantation blastocysts and in the epiblast layer of postimplantation, presomite embryos. Preimplantation and postimplantation pluripotency can be captured indefinitely in cultured embryonic stem (ES) cells and epiblast stem cells (EpiSCs) respectively. Preimplantation pluripotency in ES cells is regulated by a network of genes centred on three transcription factors (TFs) Oct4, Sox2 and Nanog. Oct4 and Sox2 form a mutually-reinforcing circuit and cooperatively stimulate transcription of downstream genes, including Nanog. All three TFs are expressed in EpiSCs and in the postimplantation epiblast. Functional studies established a role for Oct4 and Nanog in the specification of ICM cell identity, and a role for Oct4 in the maintenance of postimplantation pluripotency. Although the role of Sox2 in preimplantation ICM cells is unclear, it is critical for the establishment of egg cylinder following implantation and indispensable for ES cell pluripotency. However, despite the presence of Sox2 in postimplantation pluripotent cells the role of Sox2 in postimplantation pluripotency is unknown. In this thesis the role of Sox2 in the regulation of postimplantation pluripotency was examined. In contrast to the situation in the preimplantation ICM, Sox2 and Nanog are expressed in opposing gradients in the gastrulation-stage postimplantation epiblast, with Sox2 highest anteriorly and Nanog highest posteriorly. Interestingly the posterior epiblast of neural-plate (NP)-staged embryos was shown not to be pluripotent. Furthermore, forced expression of Sox2 but not Oct4 in this region rescued pluripotency. The ability of Oct4 to reinstate pluripotency in the somitogenesis-stage embryo is limited to Sox2-positive tissues. This strongly suggests that coexpression of Sox2 and Oct4 is important for establishing postimplantation pluripotent identity. Sox2HIGH cultured EpiSCs were not positively correlated with NanogHIGH cells. This reciprocal relationship emerged during the transition from ES cells to EpiSCs in culture. Using mutant cells with reduced levels of Sox2 or Nanog, Sox2 positively influences Nanog but Nanog negatively influences Sox2 expression post-transcriptionally. The negative influence of Nanog on Sox2 protein level was confirmed using doxycycline-inducible Nanog overexpressing EpiSCs. This negative relationship indicates that the regulation of Sox2 expression is different in postimplantation pluripotency and that Nanog may negatively regulate Sox2 on the protein level in the posterior epiblast. Sox2 is expressed at a lower level in EpiSCs than ES cells and the significance of this was further investigated by microarray transcription profiling using cells in which a fluorescent reporter (tdTomato) was knocked in to the Sox2 gene. Sox2- tdTomatoHIGH cells cultured in LIF/FCS/GMEMβ correlate with an undifferentiated cell identity and Sox2-tdTomatoLOW cells are associated with non-neural differentiation. Interestingly the global profile of ES cells and EpiSCs that share similar Sox2-tdTomato signal are non-identical. This suggests that Sox2 has different roles in different pluripotent states. ES cells with enforced Sox2 expression were unable to enter the EpiSC state, while ES cells with lowered Sox2 levels were inefficient in neural differentiation. Therefore, levels of Sox2 are critical for cell fate decisions. Strikingly, given the apparent requirement for Sox2 during Oct4-induced reinstatement of post-implantation pluripotency, deletion of Sox2 had no effect on the maintenance of EpiSC pluripotency. This is likely due to the presence of redundant Sox factors and indeed Sox3 is able to rescue the Sox2-null phenotype in ES cells. Taken together, these results suggest the hypothesis that postimplantation pluripotency is maintained by multiple Sox factors, while Nanog negatively regulates Sox2 post-transcriptionally to repress neural specification in the posterior epbilast. The positive influence of Sox2 on Nanog protein level suggests a possible negative feedback loop to balance the proneural and pluripotent properties of Sox2 in postimplantation pluripotency.

Elongation of the mouse anteroposterior axis depends on stem cell-like axial progenitors including a neuromesodermal (NM) bi-fated population existing in the primitive streak and later in the tail bud. Fate mapping experiments have demonstrated these NM progenitors reside in precise locations of the embryo. At E8.5, these cells are found in the node-streak border (NSB) and anterior epiblast on either side of the primitive streak. At tail bud stages (E10.5-E13.5), these progenitors reside in the chordoneural hinge (CNH). The coexpression of the transcription factors T (brachyury) and Sox2 has been proposed as a good marker to identify NM progenitors in vertebrates. However, this cell signature has never been thoroughly assessed during mouse axis elongation. In this thesis, I performed T and Sox2 double immunofluorescent stainings on different stages of mouse embryos and reconstructed their expression domains in the 3D images to investigate the spatiotemporal dynamics of NM progenitors during axis elongation. The results show the transient existence of T+Sox2+ cells in the posterior progenitor zone, from the headfold stage (E8.0) to the end of axis elongation (E13.5, 65somites). Moreover, the number of T+Sox2+ cells increases between E8.5 and E9.5 but gradually declines afterwards. I then investigated the time points for initiation and loss of NM progenitors by performing a series of heterotopic grafting experiments. It has been previously shown that distal epiblast (Sox2+T- cells) at LS-EB stages (E7.5) are fated to become NSB cells in E8.5 embryos. However, when cells from the distal region of LS-EB stage embryos (E7.5) were grafted to E8.5 NSB, these cells contribute extensity to the notochord but not either neural tissues or paraxial mesoderm. This indicates that NM progenitors may be not yet specified before the onset of T and Sox2 coexpression, while the notochord progenitors are already specified at E7.5. The grafting experiments also show the loss of NM progenitors at E14.5 after the end of axis elongation, which coincides with the disappearance of T+Sox2+ cells in the tail. Collectively, these results indicate that T+Sox2+ cells may represent a distinct cell state that defines NM progenitors. Wnt/β-catenin signalling has been shown to play an important role in maintaining the posterior progenitor zone. However, due to the wide expression of β-catenin and the early lethality of β-catenin null embryos, the exact effect of losing β-catenin in NM progenitors is still unknown. In this study, I took advantage of the Cre-ERT2 system and grafting technique to conditionally delete β-catenin specifically in NM progenitors during ex vivo culture. The results show that Wnt/β-catenin signalling is required cell autonomously for initiating mesoderm fate choice in NM progenitors. In its absence, mesoderm fated NM progenitors convert their fate and differentiate to neural derivatives. Moreover, the interchangeability between neural and mesodermal fate only exists in NM progenitors, as the loss of β-catenin in mesoderm committed progenitors does not affect their fate choice. Using image analysis and quantification software, I also show that Wnt/β-catenin signalling is crucial for the expansion of T+Sox2+ NM progenitors during axis elongation. Due to difficult access and a limited number of NM progenitors in vivo, in vitro generated NM progenitors from pluripotent cells, such as epiblast stem cells (EpiSCs), can offer an insight into the maintenance and differentiation of NM progenitors. Since the in vivo potential of EpiSCs had never been successfully demonstrated before, I first grafted EpiSCs into postimplantation embryos and cultured them ex vivo for 24-48 hours to assess their cell integration. The results show that EpiSCs can integrate successfully in streak stage embryos (E6.5-E7.5), but not at early somite stages (E8.5), when the epiblast has lost its pluripotency. I then further investigated the in vivo potential of EpiSC derivatives. The results show that increasing Wnt signalling in EpiSCs inhibits their ability to generate anterior neural tissues in vivo, which is consistent with the previous in vitro data. Recently, it has been demonstrated that NM progenitors can be derived from EpiSCs. These in vitro derived NM progenitors can incorporate into E8.5 embryos and give rise to both neural and mesodermal derivatives. In this thesis, I show that these in vitro derived NM progenitors do not incorporate successfully in E7.5 embryos. Collectively, by combining grafting experiments with a chimeric embryo formation assay, I can identify the in vivo stage of the in vitro counterparts of the embryonic cell types.

The governance of cell fate decisions during development is a fundamental biological problem. An important aspect of this is how cells exit a multipotent state and choose their fates in a correct manner and proportion. To tackle an aspect of this problem, I have focused on 2 multipotent models: one infinite self-renewal pluripotency in an artificial environment, and the other, bipotent progenitors in the context of the mouse embryo. The first model aimed to explore the effects of chromatin-associated factors on the ability of pluripotent mouse Embryonic Stem Cells (ESCs) to self-renew, via monitoring gene expression heterogeneity of key genes. The second model focused on Neural Mesodermal Progenitors (NMPs), a bipotent cell population found in the Caudal Lateral Epiblast (CLE) of mammalian embryos, which contributes to the spinal cord and paraxial mesoderm. The aim here was to derive NMPs in vitro which exhibit similar gene expression patterns and function like their mouse embryo counterpart and study their renewal and differentiation in detail. The first multipotent model explores the effects of chromatin remodelling on cell fate decisions, specifically investigating the consequences of inhibiting the histone acetyltransferase Kat2a on the ESCs fate. I found first, that the effect of Kat2a inhibition depends on the pluripotent state of the cells; cells in a ground state exhibit a resistance to Kat2a inhibition and maintain their pluripotency, whereas cells in a naïve state experience destabilization of their pluripotency gene regulatory network and shift towards differentiation. Second, that Kat2a inhibition in the naïve state results in a decline in the gene expression noise strength contributed by the promoter activation operation, which suggests that when ESCs become lineage-primed their transcriptional noise is constrained. In the bipotent model, the NMPs are identified as cells coexpressing Sox2 and T/Brachyury, a criterion used to derive NMP-like cells from ESCs in vitro. Comparison between the different NMPs protocols stresses that Epiblast Stem Cells (EpiSCs) are an effective source for deriving a multipotent population resembling the embryo Caudal Epiblast (CE), that generates NMPs. Furthermore, self-organization of this CE-like population, resulted in axially organized aggregates. Exploiting the mouse embryo CLE as a reference shows that EpiSCs derived NMPs, monolayers and aggregates, consist of a high proportion of cells with the embryo's NMP signature. Importantly, studying this system in vitro sheds light on the sequence of events which lead to NMP emergence in vivo. On this basis, I conclude that understanding the initial state of cells at a crossroads is important to reveal the limitations it imposes on the cells fate exploration, hence makes it possible to mimic more precisely the fate decision process in vitro.

Les cellules souches pluripotentes sont capables de donner naissance à tous les types cellulaires constituant un organisme, ce qui leur confère un fort intérêt thérapeutique. A partir de l’embryon de souris on peut en dériver deux types : les cellules souches embryonnaires (ES) au stade blastocyste et les cellules souches d’épiblaste (EpiSC) au stade œuf cylindre. Ces deux types de cellules partagent leurs propriétés pluripotentes mais se distinguent par de nombreux aspects comme leurs conditions de culture et les gènes qu’elles expriment. Nous avons montré que la reprogrammation par clonage par transfert de noyau permet d’obtenir des EpiSC présentant un méthylome et un transcriptome similaires à ceux des EpiSC issues d’embryons après fécondation. Nous avons également caractérisé le profil de méthylation de l’ADN des EpiSC, et montré une tendance à l’hyperméthylation des promoteurs des EpiSC par-rapport aux cellules ES et à l’épiblaste. De plus, l’absence de méthylation empêche la conversion des cellules ES en EpiSC. Les EpiSC semblent donc dépendre fortement de la méthylation de l’ADN pour réguler l’expression de leurs gènes, ce qui les distingue des cellules ES. / Pluripotent stem cells are of great therapeutic interest because of their capability to give rise to all the cells composing an organism. We can derive two types of these stem cells from the mouse embryo: embryonic stem cells (ESCs) from the blastocyst and epiblast stem cells (EpiSCs) from the egg cylinder stage. These two cell types share their pluripotent properties but are distinct on several features, like their culture conditions and gene expression. We showed that reprogramming using cloning by nuclear transfer allows the obtention of EpiSCs with a methylome and a transcriptome similar to those of EpiSCs derived from embryo after fertilisation. We also characterised the DNA methylation pattern of EpiSCs and showed their tendency to present a hypermethylation at their promoters compared to ESCs and epiblast. We also observed that the absence of DNA methylation blocks the conversion of ESCs into EpiSCs. As a conclusion, it seems that EpiSCs are strongly dependant of DNA methylation to regulate gene expression, which distinguishes them from ESCs.