Mitosis imposes major constrains on gene regulation: the chromatin is modified, condensed and re-arranged, leading to rod-shaped metaphase chromosomes, the eviction of transcriptional regulators from the chromatin, the shut-down of transcription and the dismantlement of gene regulatory processes (Coux et al. Transcription. 2020, PMID: 33054514; Gonzalez et al. Curr Opin Cell Biol. 2021, PMID: 33454629). Yet, we showed that a transcriptional memory after mitosis does indeed exist in mouse embryonic stem (ES) cells: all active genes get promptly reactivated after mitosis, through a hyper-transcriptional state that allows ES cells to get ready for all physiological tasks before transcription is again repressed by DNA replication (Chervova et al. Embo Reports 2023, PMID: 36330771). We then provided a mechanistic scenario for such reactivation, based on the concept of mitotic bookmarking TFs, special cases where TFs remain active during mitosis. Given the major role played by Oct4, Sox2, Nanog and Esrrb in ES self-renewal, we initially explored whether these factors act as mitotic bookmarking TFs. We showed that only Esrrb behaves as a mitotic bookmarking TF (Festuccia et al. Genome Research 2019, PMID: 30655337). We further showed that another important TF, CTCF, also behaves as a mitotic bookmarking factor (Owens et al. Elife 2019, PMID: 31599722). Strikingly, Esrrb and CTCF preserve exquisitely organized nucleosome arrays at their targets: if their binding is lost during mitosis, then nucleosomes invade the regions and occlude TF motifs.
Despite the promising function of CTCF, which mitotic binding takes place in the vicinity of the most rapidly reactivated genes, we showed that, alone, it cannot explain the widespread strong post-mitotic reactivation of the genome (Chervova et al. Embo Reports 2023, PMID: 36330771).This observation, together with the known role of CTCF in 3D topology in interphase, prompted us to address whether mitotic CTCF could instead impose 3D constrains to Condensins, the loop extruders that fold mitotic chromosomes. We found that mitotic CTCF does not block Condensins as it blocks Cohesins in interphase (Oomen et al. BioRxiv 2024). While of importance, these observations did not provide a satisfactory explanation to post-mitotic gene reactivation dynamics. However, bioinformatic analyses showed that the vast majority of the most active genes after mitosis harbor MAX-MYC binding sites at their promoters (Chervova et al. Embo Reports 2023, PMID: 36330771). Accordingly, we found that MAX remains bound to its targets during mitosis, preferentially at promoters, and facilitates early recruitment of MYC following mitosis to trigger a global hyper-transcriptional state (Gonzalez et al. BiorXiv 2023).
While Max/Myc act on all genes, they lack specificity for cell identity genes, in contrast to Esrrb. Despite our efforts, linking it to a clear functional consequence was unsatisfactory. However, we found that other nuclear receptors (like Esrrb itself) and known cofactors, bind with Esrrb in interphase and in mitosis. Focusing on another nuclear receptor, Nr5a2, we showed that nuclear receptors redundantly bookmark a sizable fraction of the mouse ES cell genome for their efficient post-mitotic reactivation, especially genes important for the pluripotency network (Chervova et al. Nat Struct Mol Biol. 2024, PMID: 38196033). The finding of bookmarking redundancy among nuclear receptors, prompted us to generate inducible full double knock-outs of both Esrrb and Nr5a2. We found hat in their absence the pluripotency network collapses and ES cells differentiate (Festuccia et al. Development 2021, PMID: 34397088). Next, we asked if this redundancy was also operating in vivo, during the establishment of pluripotency in the blastocyst. We found that the loss of Nr5a2 alone led to pronounced and fully penetrant phenotypes before the formation of the blastocyst: the embryos degenerate at the morula stage, they cannot pass the 8-cell stage due to a large number of defects including mitotic issues (Festuccia et al. Science 2024, PMID: 39361745). This work identified the mitotic bookmarking factor Nr5a2 as a determinant regulator of the morula. Interestingly, Esrrb alone has been shown by others to be required to form the primitive endoderm, another lineage of the blastocyst; yet, neither Esrrb nor Nr5a2 are expressed in this tissue, indicating that Esrrb has an early role only. Using in vitro models of the primitive endoderm we could show that another mitotic bookmarking nuclear receptor, Esrra, is also strictly required for differentiation along this lineage (Coux et al. BiorXiv 2024).
