RYBP regulates Pax6 during in vitro neural differentiation of mouse embryonic stem cells

We have previously reported that RING1 and YY1 binding protein (RYBP) is important for central nervous system development in mice and that Rybp null mutant (Rybp−/−) mouse embryonic stem (ES) cells form more progenitors and less terminally differentiated neural cells than the wild type cells in vitro. Accelerated progenitor formation coincided with a high level of Pax6 expression in the Rybp−/− neural cultures. Since Pax6 is a retinoic acid (RA) inducible gene, we have analyzed whether altered RA signaling contributes to the accelerated progenitor formation and impaired differentiation ability of the Rybp−/− cells. Results suggested that elevated Pax6 expression was driven by the increased activity of the RA signaling pathway in the Rybp−/− neural cultures. RYBP was able to repress Pax6 through its P1 promoter. The repression was further attenuated when RING1, a core member of ncPRC1s was also present. According to this, RYBP and PAX6 were rarely localized in the same wild type cells during in vitro neural differentiation. These results suggest polycomb dependent regulation of Pax6 by RYBP during in vitro neural differentiation. Our results thus provide novel insights on the dynamic regulation of Pax6 and RA signaling by RYBP during mouse neural development.

We have previously reported that Rybp is essential for mouse embryonic development and decreased dosage of Rybp caused exencephaly in mice 22 . Furthermore, ES cells lacking functional Rybp formed less matured neural cell types from existing progenitors in vitro than the wild type cells and exhibited altered expression of a series of key neural marker genes including Pax6 23 . RYBP, is Zinc finger protein, which can recognize nucleosomes containing H2AK119Ub marks, It can bind to ubiquitylated proteins and as a part of ncPRC1s, RYBP facilitates H2A ubiquitylation of targets 15 . Besides the Zinc finger motif of RYBP, it also contains intrinsically disordered stretches conferring multiple tertiary structure of the protein. Partially due to these characteristics, RYBP interacts with a versatile set of proteins with diverse functions (moonlighting function). RYBP was first described as interactor of the mammalian polycomb group proteins and as an epigenetic regulator involved in chromatin compaction 24 . Besides its role in neural development, the function of RYBP has been investigated during haematopoetic, cardiac and germ cell development and also in tumour formation [25][26][27][28] , RYBP was also described as a pro-apoptotic protein and interactor of procaspase-8 and DEDD via a non-epistatic mechanism 29 . However multiple biological functions of RYBP has been described, the precise molecular mechanism of specific functions are not well established and target genes of RYBP actions are not well defined.
In the current study, we have investigated the molecular mechanisms that might be responsible for the increased progenitor formation of the Rybp −/− ES cells. We examined whether the modulation of the RA signaling pathway members could transmit the effects of RYBP and whether this can be the reason of the elevated Pax6 level in the Rybp −/− neural cultures. We used a stem cell based in vitro differentiation system and compared the effect of RA on the expression of Pax6 and other members of the RA signaling pathway in the presence and absence of Rybp during in vitro neural differentiation. Since RYBP exerts its functions partially as a core member of the ncPRC1s, we investigated whether the function of RYBP is mediated via ncPRC1s to regulate Pax6 gene expression. Here we report that in the lack of Rybp, the expression of the RA signaling pathway members and their target genes, such as Pax6, were upregulated during neural differentiation. By using luciferase reporter assays, we revealed that RYBP is able to directly repress the P1 promoter of the Pax6 gene and that this repression was enhanced in the presence of RING1. These findings suggest that RYBP regulates Pax6 expression partially by governing the expression of several RA pathway genes and by repressing the activity of Pax6 P1 promoter via ncPRC1s.

Results
The expression of RA pathway members is altered in the lack of Rybp in ES cells. In order to investigate whether the lack of Rybp altered RA signaling related gene expression, we analyzed our previously published transcriptome dataset from wild type and Rybp −/− ES cells 30 (GEO acc. GSE151349) (Materials and Methods) focusing on the expression of genes related to RA signaling.
Next, we checked the H2AK119ub1-a major repression mark in the wild type and Rybp KO ES cells using previously published ChIP-seq datasets 33 (SRA acc. SRP247114) at the genomic loci of those 7 genes which were upregulated in Rybp −/− ES cells (Fig. 1a). We found loss of H2AK119ub1 mark in Rybp −/− cells at the genomic loci of 4 genes namely Rbp7, Crabp1, Rbp4, Aldh1b1 (Fig. 1b). The loss of H2AK119ub1 marks in the genomic loci of specific RA pathway members in the Rybp −/− cells indicated that RYBP might regulate these genes in a polycomb dependent way. These results, and the elevated Pax6 gene expression in the Rybp −/− neural cultures 23 encouraged us to further analyze the detailed expression changes of RA pathway members during in vitro neural differentiation.
Expression of several RA signaling pathway members are amplified during in vitro neural differentiation of the Rybp −/− stem cells. Since Pax6 is robustly activated by RA signaling during neural development, we examined whether the expression of RA signaling pathway members is attenuated after alltrans retinoic acid (atRA, also known as tretinoin) treatment and whether the Rybp −/− neural cultures express more Pax6 than the wild type cells. To answer this question, wild type and Rybp −/− ES cells were differentiated to neural lineages in vitro for 14 days (d) by using the modified protocol of Bibel et al. 34  www.nature.com/scientificreports/ treatment), day 10 (d10; representing an early stage of neural differentiation) and day14 (d14; representing the late stage of neural differentiation) ( Fig S2a). For monitoring the mRNA expression changes, qRT-PCR (Materials and methods, Quantitative real-time PCR (qRT-PCR)) was performed using primers specific to Stra6, Retinol binding protein 1 (Rbp1, also known as Crbp1), Retinol binding protein 2 (Rbp2, also known as Crbp2), Retinol dehydrogenase 10 (Rdh10) and Retinol dehydrogenase 14 (Rdh14), Aldehyde dehydrogenase 1 family member A1 (Aldh1a1, also known as Raldh1) and A2 (Aldh1a2, also known as Raldh2), Crabp1, Cellular retinoic acid binding protein 2 (Crabp2), Retinoic acid receptor alpha (Rarα), Rarβ, Retinoid X receptor alpha (Rxrα), Retinoid X receptor beta (Rxrβ) and Cytochrome P450 Family 26 Subfamily B Member 1 (Cyp26b1) were monitored by qRT-PCR (primer details in Table S1).
As expected, Stra6 was robustly induced by atRA in both cell lines at every examined time point (d7, d10, d14). Furthermore, Stra6 was induced at a higher extent in the Rybp −/− cells compared to the wild type cells (Fig. 2a). The expression of both Rbp1 (Fig. 2b) and Rbp2 (Fig. 2c) increased after atRA treatment in the wild type cells, and the mRNA levels were significantly higher in the Rybp −/− cells. We also assessed the expression changes of the retinol metabolizing Rdh10 and Rdh14, the retinal metabolizing Aldh1a1 and Aldh1a2 dehydrogenase enzymes. The expression of these genes was strongly enhanced after atRA treatment (d7) and Rdh10 expression reached the highest level in the Rybp −/− cells by the end of differentiation (d10 and d14) (Fig. 2d). There were no significant differences in Rdh14 expression between the two cell lines, its level gradually increased after atRA treatment until the endpoint of in vitro differentiation (Fig. 2e). Aldh1a1 expression kinetics was different in the two cell lines: it peaked right after atRA induction (d7) in the wild type cells, and it was most extensively expressed by the end of differentiation (d14) in the Rybp −/− cells and the induction was 2.5 times higher compared to the wild type at this time point (Fig. 2f). In contrast, Aldh1a2 gene expression was higher at d7 and d10 in the absence of Rybp in comparison to the wild type (Fig. 2g). The expression changes of the cytoplasmic RA binding complex members Crabp1 and Crabp2 peaked after atRA treatment (day 7). At d7 and d14 Crabp1 expression was also more pronounced (Fig. 2h) than Crabp2 (Fig. 2i) in the Rybp −/− mutant.
Furthermore, gene expression analysis revealed upregulation of the nuclear RA binding Rar/Rxr complex members. We found that Rarα, exhibited persistently higher expression peaks in the Rybp −/− cells compared to the wild type at d10 and d14 (Fig. 2j). The two cell lines did not show any difference in the Rarβ expression www.nature.com/scientificreports/ (Fig. 2k). The expression of both Rxr members showed a gradually increasing level in the Rybp −/− cells. Rxrα level was higher at all examined time points in the Rybp −/− cell line and its expression kinetics did not decline by the end of differentiation as it did in the case of the wild type (Fig. 2l). Rxrβ level was also higher in the Rybp −/− cells, but compared to the Rxrα, it showed higher mRNA level at d0, d3 and the highest at d7 (after atRA treatment) (Fig. 2m). The mRNA level of RA degrading enzyme Cyp26a1 was strongly induced by atRA treatment (d7), and the level of induction was more pronounced in the Rybp −/− compared to the wild type (Fig. 2n). There were no differences in Cyp26b1 expression in the two cell lines (Fig. 2o). These results showed that the expression of all examined genes, which are responsible for retinoid binding and metabolism in the cytoplasm (Rbp1, Rbp2, Rdh10, Aldh1a2 and Crabp1) was higher in the Rybp −/− cells from neuroectoderm formation (after day 4) until the last day (d14) of in vitro neural differentiation. Notably, the www.nature.com/scientificreports/ expression level of RA inducible genes, such as Stra6 and Cyp26a1, were significantly higher after atRA treatment (d7) in the absence of Rybp. Taken together, these data suggested that the lack of Rybp led to an amplified RA response in the cells during in vitro neural differentiation.
PAX6 is upregulated in the Rybp −/− neural cultures in response to RA signaling. Next, we examined whether the elevated level of Pax6 indeed depends on the atRA treatment. First, we investigated whether PAX6 is profoundly present in the Rybp −/− cells even without atRA treatment (Fig. 3a). In this experiment, both the wild type and the Rybp −/− ES cells were grown in suspension culture for promoting EB formation for 14 days with and without atRA (Materials and methods, Embryoid body formation and atRA treatment). We added atRA on d4 to the cultures in case of the atRA treated samples, to facilitate neuroectoderm formation (Fig S2b). Samples were collected at different time points (d3, d7, d10 and d14) and were proceeded for ICC (Materials and www.nature.com/scientificreports/ methods, Immunocytochemistry (ICC)) to compare the PAX6 protein level in the wild type and the Rybp −/− lines (Fig. 3c,d). PAX6 protein was detected as early as d3 both in the wild type and the Rybp −/− EBs at a very low level (Fig. 3a,b). As expected, at later time points (d7, d10 and d14) the staining intensity of PAX6 was stronger in the atRA treated EBs (Fig. 3b-d) compared to the untreated EBs (Fig. 3a,c,d) in both cell lines. Results were further confirmed by quantification of the PAX6 protein signals by ImageJ (Fig. 3c,d).
These results showed that even without atRA treatment PAX6 was present in an excess in the absence of Rybp in comparison to the wild type EBs, and this difference was more pronounced after atRA treatment.

The localization of RYBP and PAX6 proteins dynamically change during in vitro neural differentiation.
To reveal the spatiotemporal distribution of PAX6 and RYBP in comparison to each other, we first subjected the wild type ES cells to in vitro neural differentiation (Materials and methods, in vitro neural differentiation) and analyzed the localization of the two proteins at d0, 3, 7, 10, and 14 of in vitro neural differentiation by ICC (Materials and methods, Immunocytochemistry (ICC)). Samples were examined using an Olympus LSM confocal microscope (Fig. 4a,b). ImageJ software was used to quantify the obtained fluorescent signals (Fig. 4c,d).
The most RYBP positive cells could be observed in ES (d0) cell stage and at early differentiation stages (d3) (Fig. 4a). After the atRA treatment, the amount of RYBP decreased (d7) and it appeared mostly in the outer layer of the EBs, where endoderm forms. (Fig. 4a). By d10 and 14, less sparsely spread RYBP positive signals were visible in the neural cultures (Fig. 4a). As expected, the most PAX6 positive signals appeared in cells at d7, after atRA treatment, and very few PAX6 signals were present at d10 and 14 in the wild type cultures (Fig. 4a). By subjecting Rybp −/− cells to in vitro neural differentiation as well, PAX6 appeared to be localized abundantly at d7, d10 and d14 in comparison to the wild type cells. (Fig. 4b,d). As expected, there was no detectable RYBP signal in the Rybp −/− cells at any time point of in vitro neural differentiation. Notably, the amount of PAX6 was higher after the atRA treatment (d7) in the Rybp −/− cells and PAX6 did not decrease by the end of differentiation (d10 and d14) (Fig. 4d). www.nature.com/scientificreports/ These results indicated that in differentiating cells, where there is no RYBP, more PAX6 protein appeared and that the two proteins are not or rarely co-localized.
RYBP and PAX6 are co-expressed at progenitor stage of in vitro neural differentiation. Previous experiments suggested that RYBP and PAX6 expressed complementary in most of the cells during neural differentiation. Therefore, co-expression was analyzed in an in vitro neural differentiation model system by FACS analysis. Wild type ES cells were exposed to atRA (Materials and methods, in vitro neural differentiation) for different times (d4, 6, 7, 8 and 14), or left unstimulated and subsequently samples were single (data not shown) or double labeled (Fig. 5) with RYBP and PAX6 antibodies then analyzed with flow cytometry analysis (Materials and methods, Flow Cytometry).
The analyses showed that the ratio of RYBP + /PAX6 + double positive cells increased up to 6th day of atRA treatment (Fig. 5b) then decreased after d7 and remained at similar level till d14 of differentiation. (Fig. 5b). These results showed that the most RYBP + /PAX6 + cells appeared at the progenitor cell stage and that the number of RYBP + /PAX6 + cells decreased by differentiation.
RYBP represses Pax6 expression through its P1 promoter. Since previous results suggested that RYBP might repress Pax6 gene expression at developmental stage specific manner, we next examined whether RYBP can indeed regulate Pax6 expression through its promoter using the luciferase reporter assay system. For this assay, we utilized a 2014 bp long P1 promoter region (-335 to 1698 from TSS) (Materials and methods, Cloning of Pax6 P1 promoter) of Pax6 that encompassed the entire exon 1 (Fig. 6a) and included a predicted 1127 bp long CpG island that covers 55.9% of the promoter region (Fig. 6b). www.nature.com/scientificreports/ To evaluate the regulatory potential of RYBP at the Pax6 P1 promoter, HEK293 cells were co-transfected with the Pax6 P1 promoter construct in combination with different dosages of the Rybp cDNA containing construct 35 (1 µg, 5 µg, 10 µg and 20 µg) (Materials and Methods, Luciferase reporter assay). 40 h after transfection, the protein cell lysates were harvested, processed and luciferase signals were measured. Our results showed over 2-folds repression of the Pax6 P1 promoter by RYBP in comparison to the control cell lysates (Fig. 6c). The fold decrease was proportional to RYBP dosage, and the luciferase levels reached almost the base level when 20 µg of RYBP was applied (Fig. 6c). Since RYBP influenced a sharp decrease in the luciferase levels of the Pax6 P1 promoter, we performed transcription factor binding site (TFBS) analysis of the Pax6 P1 promoter using the TRANSFAC tool (https:// genex plain. com/ trans fac). This revealed various known binding partners of RYBP that could synergistically function in regulating the expression of Pax6 24 (Fig. 6d). This included several E2 transcription factor (E2F) binding sites (E2F1, E2F2, E2F3 and E2F4), a Yin-yang-1 transcription factor (YY1) binding site and a Myelocytomatosis oncogene/MYC associated factor X (MYC/MAX) consensus sites (Fig. 6d). The promoter region also contained RXRα and PAX6 binding sites. This is in accordance with previously published data stating that RA signaling plays a crucial role in the regulation of Pax6 36,37 (Fig. 6d). These all together suggested that RYBP could potentially recruit repressor complexes via its binding partners and could repress Pax6 gene expression through its P1 promoter.
RYBP represses Pax6 gene expression in a polycomb dependent manner. Since RYBP is described for its regulatory functions as a core member of the ncPRC1s, we examined whether RYBP could repress the Pax6 P1 promoter by a polycomb dependent mechanism. To test this hypothesis, HEK293 cells were co-transfected with the Pax6 P1 promoter containing luciferase construct (Materials and Methods, Luciferase reporter assay) and constructs expressing RYBP (1 µg, 5 µg, 10 µg and 20 µg) and the catalytic ubiquitin ligase subunit of the ncPRC1 complex, RING1 (1 µg and 10 µg) 38 . We found that the RYBP repression was more pro- www.nature.com/scientificreports/ nounced when it was co-transfected with the RING1 as compared to RYBP alone (Fig. 7a). To check whether any members of the PRCs can potentially bind to the Pax6 P1 promoter, we analyzed existing ChIP-seq data from neural progenitor cells (NPCs) 39 using UCSC genome browser. Our analysis displayed binding peaks of RNF2, Polycomb group ring finger 2 (PCGF2), SUZ12 Polycomb repressive complex 2 subunit (SUZ12) and Enhancer of Zeste 2 polycomb repressive complex 2 subunit (EZH2) at the P1 promoter region of Pax6. The binding peaks of RNF2 and PCGF2 showed that ncPRC1 complexes could bind to the Pax6 P1 promoter in NPCs and potentially regulate Pax6 expression. However, it is likely that the PRC2 complex does not regulate Pax6 directly, because SUZ12 and EZH2 did not show significant binding (Fig. 7b). The methylation state of the Pax6 genomic locus in NPCs revealed that, RNF2 peaks coincided with H3K27me3 and H3K9me3 marks showing that the methylation of H3 can potentially play a role in the repression of Pax6 (Fig. 7c). In ES cells and NPCs the binding peaks of RNF2, CBX factors CBX2, CBX7 and CBX8, which are canonical PRC1 members, revealed a possible dynamic interplay among the PRC1 and ncPRC1 members in maintaining the regulatory status of the Pax6 genomic locus during differentiation (Fig S3a). Notably, the binding peaks of RNF2, CBX2 and CBX7 factors were present and colocalized at the whole genomic locus suggesting the poised state of the locus. In NPCs, the binding peaks of the same factors were more condensed and specific near the promoter region exhibiting the bivalent state of the promoter, which can be regulated by the presence of specific transcription factors (Fig S3b). www.nature.com/scientificreports/ The peak density heatmap showed the global occupancy of ncPRC1s core complex members RNF2 and PCGF2 at the CpG island in NPCs indicating the affinity of the PRC1s binding at the CpG islands. RNF2 binding was more condensed to the position of the CpG islands than PCGF2 suggesting that the position of the CpG islands most likely plays a central role in the repressive activity of ncPRC1s at the Pax6 P1 promoter, which also includes RYBP (Fig. 7d). These results demonstrated and confirmed further that the RYBP mediated Pax6 gene repression of its P1 promoter is polycomb dependent.

Discussion
In this study we unrevealed that the ncPRC1 core member RYBP is important for proper expression of the genes of the RA signaling pathway, and of their targets, such as Pax6 during in vitro neural differentiation of mouse ES cells (Figs. 1 and 2). Pax6 is a RA responsive gene and in the Sey mice where Pax6 mRNA is expressed at decreased level, the lens anlage cannot respond to exogenous RA after E9.0 40 confirming that Pax6 expression was sensitive to RA signaling and that optimal dosage of RA and signaling through RA was required for the development of the embryo proper. RA is a key inductive signal for neural differentiation 41 , optimal RA concentration itself is crucial for appropriate lineage commitment. High concentration of RA induce neural differentiation and low concentration of RA can induce cardiomyocyte differentiation 42 . Based on our current and past studies 23,27 we think that Rybp is important in balancing the proper RA level in cells during differentiation. Thus, Rybp might be important for the early steps of differentiation at the time of lineage commitment and tissue formation. Several studies have already shown that Rybp is indeed important for lineage commitment. For example, when Rybp was depleted from early hematopoietic stem cells and progenitor cells the correct B-1-to-B-2 B-cell progenitor ratio was changed 26 . However, the molecular mechanism of these fate decision making signaling events is still not understood. We have shown previously that the Rybp −/− ES cells form less mature neurons, astrocytes, and oligodendrocytes during in vitro neural differentiation while they form neural progenitors abundantly 23 (Fig S4  and S5). In this study we have shown that Pax6 expression is elevated during the entire time course of differentiation (at d7, d10 and d14) in the Rybp −/− cells while it was silenced properly in the wild type neural cultures. High level of Pax6 interferes with neural differentiation and can trigger an abundant but developmentally not fully competent neural progenitor pool formation. The Rybp heterozygous mutation in mice often led to the overgrowth of midbrain, exencephaly and disorganized, incompletely differentiated neurocortex and the causatives of this phenotype were not clarified yet 22 . The overgrowth of tissues may result from cells with hyperproliferative capacity or even a boost of progenitors, which are trapped at an early phase of differentiation and developmentally compromised. An elevated RA signaling or Pax6 level can drive accelerated progenitor formation contributing to the heterozygous phenotype and it is reasonable to postulate that increased mRNA expression level of key RA pathway members can alone facilitate Pax6 expression in the Rybp null neural cultures.
By monitoring the relative gene expression changes we have found that the expression of the RA pathway members altered during neural differentiation of Rybp −/− ES cells (Fig. 2). These alterations could indirectly lead to more robust expression of several genes, which are regulated by RA, such as Pax6 and many RA pathway members like Stra6, Crabp1, Cyp26a1 43,44 . The RA inducible trans-membrane receptor Stra6, is responsible for the retinol uptake thus high level of Stra6 (Fig. 2a) in the Rybp −/− indicates the possibility of higher retinol uptake of the cells. The elevated expression of Stra6 alone can cause alterations of RA pathway 45 . Expression of the retinol converting enzyme Rdh10 (Fig. 2d) showed similar kinetics as Stra6, which suggested that an increased amount of RDH10 was required to convert the excess retinol into retinal in the Rybp −/− cells. There was elevation in the expression levels of retinal converting enzyme Aldh1a1 (Fig. 2f) and Aldh1a2 (Fig. 2g) as well indicating that the conversion of retinal into RA was also possibly enhanced in the Rybp −/− cells. After atRA treatment the RA inducible Crabp1 expression (Fig. 2h) was also upregulated in the Rybp −/− cells, suggesting that more RA transported into the nucleus of Rybp −/− cells. Finally, another inducible RA pathway member, Cyp26a1 also showed elevated expression upon atRA treatment in the Rybp −/− cells as compared to the wild type cells. All of these effects together can also trigger the accelerated early neural processes and the largely obscured terminal differentiation of neural lineages in the lack of Rybp 23 .
However, Rybp depletion can also cause alteration in PAX6 levels independent to atRA treatment as well (Fig. 3a). Our results showed that the level of PAX6 is elevated (Fig. 3a,c) in the Rybp −/− neural cultures even when cells are not yet exposed to atRA. After atRA treatment, the PAX6 protein level remained elevated in the Rybp −/− cells (Fig. 4b,c) even at d10 and d14 of in vitro neural differentiation suggesting that RYBP is required for the repression of Pax6 during neurogenesis (Fig. 5a,b) and the high PAX6 protein level could contribute to the accelerated neural progenitor pool formation that hindered the formation of matured, terminally differentiated neural cultures.
RYBP is a member of the ncPRC1s, which form different complexes based on its PCGF subunit (PCGF1, 2, 3, 4, 5 or 6) they contain 11,17 . The different PCGFs subunits interact with different proteins and regulate different targeting mechanisms and repressor activity. Luciferase reporter assays using Pax6 P1 promoter (Fig. 6a) confirmed that RYBP could repress the Pax6 expression (Fig. 6c). E2F and YY1 transcription factors are previously shown to associate with RYBP for the regulation of genes 46 . YY1 is also widely considered to be the DNA binding interactor of ncPRC1s although it is not described to be purified as part of any ncPRC1 complex 17,24 . The presence of TFBSs of several interactors of RYBP and ncPRC1 complexes suggested a possible polycomb mediated repression of the Pax6 P1 promoter (Fig. 6d). Co-transfection of HEK293 cells with both RYBP and its core polycomb interactor RING1 further promoted the repression of the Pax6 P1 promoter (Fig. 7a) suggesting that RYBP regulates Pax6 in a polycomb dependent manner. Polycomb mediated regulation of genes are widely connected to the histone modifications of the given genomic locus 47,48 . The presence of H3K119ub1 and H3K27me3 marks at the gene regulatory regions are both produced and recognized by the PRC1 and PRC2 complexes, respectively [49][50][51] 55 of selected histone methylation marks H3K27me3 and H3K9me3 along with RNF2 binding peaks show co-occupancy at the Pax6 promoter in NPCs ( Fig. 7c and Fig S3). Pattern of histone methylation marks can predict the methylation states of the CpG islands present in the regulatory regions of genes 56 . Integrating RNF2 binding peaks and the global position of the CpG islands revealed that RNF2 binding peaks are also condensed at the CpG islands in NPCs (Fig. 7d). The same analysis for PRC2 members SUZ12 and EZH2 exhibited no significant binding but based on the H3K27me3 which is a PRC2 specific methylation mark, it is possible that PRC2 might also function in the repression of Pax6 (Fig. 7b). Taken together our results showed that RYBP can repress Pax6 expression through the Pax6 P1 promoter and that this repression was further enhanced when RING1 was also present suggesting that RYBP regulates the Pax6 locus in a polycomb dependent way (Fig. 8). These observations not only contribute to understanding the phenotypic changes of the Rybp −/−22,23 , but also to understand the development of neurodegenerative disease conditions when impairment occurs in adult neurogenesis due to abundant progenitor formation 57 .

Materials and methods
Cell lines and culture condition. Mouse (129SV/Ola) R1 58 (mentioned as Rybp +/+ or wild type) and D11 22 (mentioned as Rybp −/− or Rybp null mutant) ES cells were cultured as we reported in Henry et al. 2020 27 . In vitro neural differentiation. ES cells were induced to differentiate into neuronal lineages as previously described, with some modifications 23,34 . For single-cell suspensions, the cells were dissociated from monolayer culture (day 0) with 0.25% trypsin-EDTA (Trypsin-EDTA (0.5%), Gibco, Cat.No 15400-054). The 8 days old EBs were dissociated and plated and after two days the medium was changed to 1: . Neuroectoderm formation was induced by all-trans retinoic acid (atRA) treatment between day 4 and 8 of differentiation. At day 8 the EBs were collected, trypsinized to single cell suspension, then seeded onto surface coated tissue culture dishes and differentiated further for 6 more days (Fig S2).

Embryoid body (EB) formation and atRA treatment.
Mouse ES cells were harvested as single cell suspension using 0.25% trypsin-EDTA, then seeded at a density of 4.5 × 10 5 cells/ml in ES medium without LIF into 100 mm diameter bacteriological dishes where cell attachment was prevented. ES cells were kept in suspension for 14 days to form embryoid bodies (EBs). Medium was changed on every second day. EBs were kept in the presence (RA+) and absence (RA-) of 5 μM atRA from the 4th day (d4) of EB formation until the 14th day (d14). The EBs were harvested for ICC analysis on different time points: day 3, 7, 10 and 14 (d3, d7, d10 and d14) (Fig S2).
Quantitative real-time PCR (qRT-PCR). RNA was isolated from the cell cultures during the time course of in vitro neural differentiation at the designated time points (d0, d3, d7, d10 and d14) using Gene Jet RNA www.nature.com/scientificreports/ 5 µg; pcDNA3.1-Rybp-HA: 1 µg, 5 µg, 10 µg and 20 µg; pcDNA3.1-Ring1-FLAG: 1 µg and 10 µg) using CaPO 4 transfection method 60,61 . CaPO 4 transfection were performed as follows: 8 h before transfection the cell culture medium was changed, the vectors were diluted in TE (Tris-EDTA buffer solution, Sigma, Cat.No 93283) buffer, then 2.5 M CaCl 2 and 2 × HBS (Hepes Buffered Saline, Sigma, Cat.No H3375) were added in dropwise to the solution while the mix was bubbled to providing oxygenation, finally this solution was added to the medium. 16 h after transfection, the cell culture medium was changed. 48 h after transfection the cells were washed with PBS and lysed with 1 × Passive lysis buffer provided by the Luciferase reporter assay kit (Dual Luciferase Reporter Assay System, Promega, Cat.No E1500). 20 µg of lysate from each point was mixed with 100 µl Luciferase Assay Reagent II substrate (provided by the kit) and the induced luciferase activity was recorded immediately according to the manufacturer's instructions. The luciferase activity was recorded with Perkin Elmer TopCount NXT Luminometer. All activities were measured in triplicates.

Statistical analysis.
All experiments were repeated three times from three independent biological samples (by defrosting independent vial of cells) and technical repeats were used as triplicates (e.g. three petri dishes/ sampeling) at each examined timepoints. Experiments were evaluated with Microsoft Excel by using Student T-test type 3. Means are standard deviation. Values of p < 0.05 were accepted as significant (*p < 0.05; **p < 0.01; ***p < 0.001).