Final Report Summary - GRNES (Analysis of the gene regulatory network controlling ES cell identity) The recognition that undifferentiated ES cells express some genes heterogeneously, and the demonstration that cells with differing levels of the critical pluripotent transcription factor Nanog have variable access to alternate fates represent one of the most intriguing features of ES cells biology. Indeed, fluctuations in Nanog allow ES cells to explore the environment to self-renew or differentiate, with Nanog-positive cells self-renewing efficiently and Nanog-negative cells displaying poor self-renewal efficiency and increased competence to exit the undifferentiated state and undergo differentiation. Before the applicant started his post-doctoral research in Ian Chambers' laboratory, the mechanisms underlying Nanog heterogeneity were largely unknown but believed to be solely dependent on extrinsic signalling such as FGF/Erk. Indeed, while FGF treatment downregulates Nanog, inhibition of Erk activity abrogates Nanog heterogeneity. This belief was further consolidated on the basis that the architecture of the core pluripotency network suggested that intrinsic mechanisms of gene regulation would be irrelevant in terms of generating fluctuations of Nanog transcription: the key pluripotency transcription factors Nanog, Oct4 and Sox2 were proposed to form a coherent positive feedback loop whereby each factor activates transcription of the three genes to reinforce and stabilise the undifferentiated, self-renewing state. We have found that Nanog transcription is under the tight control of Oct4, Sox2 and Nanog itself, as previously proposed. Nanog binds 5kb upstream of its own promoter, whilst Oct4 and Sox2 bind mainly at the Nanog promoter. However, in contrast to the prevalent model, the applicant's work has shown that Nanog autoregulation is based on autorepressive mechanisms which promote the generation and maintenance of Nanog-negative cells.The mechanisms of Nanog autoregulation were dissected using genetic systems of inducible depletion and restoration of Nanog. Nanog depletion was studied in a line in which both Nanog alleles have been deleted and Nanog expression is supported by a randomly integrated transgene that can be deleted upon Tamoxifen treatment (RCNßH). Conversely, a transgene encoding a Nanog-ERT2 fusion was introduced to independent Nanog-null ES cells to restore nuclear Nanog expression upon Tamoxifen treatment (44NERT). To assess the transcriptional activity of endogenous Nanog, qRT-PCR and RNA-FISH with primers and probes designed across a region of Nanog intron 1 that remains intact in Nanog-deleted alleles were designed. In contrast to current models, increased Nanog pre-mRNA levels and transcriptional frequency of the Nanog locus were observed upon loss of Nanog binding in Tamoxifen-treated RCNßH cells. Conversely, reduced levels of Nanog pre-mRNA and transcriptional frequency were observed after restoring Nanog binding in 44NERT cells. As recently shown by others, the RNA-FISH assays showed that Nanog is mainly transcribed monoallelically, and this regardless of Nanog activity.To study the dynamics of Nanog-negative cells generation relative to Nanog gene activity, the 44NERT line was used. In these cells, one Nanog allele is replaced by the GFP, the second allele by a selection cassette, and Nanog expression is supported from a transgene producing a Nanog-ERT2 fusion protein. After FACS-sorting Nanog:GFP-positive 44NERT cells, the presence of nuclear exogenous Nanog (in the presence of Tamoxifen) was associated to an efficient transition to the negative state. Conversely, after FACS-sorting Nanog:GFP-negative 44NERT cells, only in the absence of Tamoxifen they transited efficiently to the positive state and this even after 20 days of continous culture. Therefore, Nanog autorepression appears to be a major regulatory step of Nanog heterogeneity by promoting and maintaining the negative state.To address the molecular mechanisms governing Nanog autorepression, the distribution of the transcriptional apparatus across the Nanog locus was established by ChIP. A virtually complete transcription initiation complex was found at the -5kb region at levels similar or even higher than at the promoter itself. ChIP analyses of the phosphorylation forms of the RNAPII, combined with RT-(Q)PCR and RNA-FISH showed that the -5kb and Nanog promoter regions display bidirectional transcription exclusively in undifferentiated cells. Loss of Nanog binding in RCNßH cells led to an increase of phospho-S5 RNAPII from the -5kb region to the Nanog promoter, which was associated with increased bidirectional transcription and H3K4me3 enrichment at the promoter. Conversely, the restoration of Nanog binding to 44NERT cells led to reduced enrichment of phospho-S5 RNAPII across the Nanog upstream region in association with lower levels of bidirectional transcription and H3K4me3. Notably, Nanog was found to enhance binding of NELF, which blocks RNAPII release from paused promoters enriched in phospho-S5 RNAPII, at the -5kb region.In accord with several biological systems, this work provides the first evidences indicating that intrinsic activities of the core pluripotency network, in this case Nanog autorepression, are required to maximise Nanog heterogeneity: Nanog autorepression establishes the conditions facilitating switching between Nanog expression states. In the absence of Nanog, high levels of bidirectional transcription between the -5kb region and the Nanog promoter confer a high probability for Nanog expression by maintaining a chromatin state competent for transcription and an efficient loading of RNAPII.S5Pi at both regulatory regions. Binding of Nanog to the -5kb region increases NELF recruitment and tethers RNAPII to inhibit the release of RNAPII.S5Pi. This entails the erosion of the crosstalk of the two regulatory regions such that the probability of Nanog transcription drops, thereby opening a window of opportunity for switching to the Nanog-negative state. In agreement with this, ES lines harbouring a randomly integrated Nanog:GFP reporter vector that lacks the -5kb region exhibit relatively homogeneous and reduced transcription activity as compared to a line in which the GFP is driven by the endogenous Nanog locus.
