The Polycomb group protein Ring1 regulates dorsoventral patterning of the mouse telencephalon

Dorsal-ventral patterning of the mammalian telencephalon is fundamental to the formation of distinct functional regions including the neocortex and ganglionic eminence. While Bone morphogenetic protein (BMP), Wnt, and Sonic hedgehog (Shh) signaling are known to determine regional identity along the dorsoventral axis, how the region-specific expression of these morphogens is established remains unclear. Here we show that the Polycomb group (PcG) protein Ring1 contributes to the ventralization of the mouse telencephalon. Deletion of Ring1b or both Ring1a and Ring1b in neuroepithelial cells induces ectopic expression of dorsal genes, including those for BMP and Wnt ligands, as well as attenuated expression of the gene for Shh, a key morphogen for ventralization, in the ventral telencephalon. We observe PcG protein–mediated trimethylation of histone 3 at lysine-27 and binding of Ring1B at BMP and Wnt ligand genes specifically in the ventral region. Furthermore, forced activation of BMP or Wnt signaling represses Shh expression. Our results thus indicate that PcG proteins suppress BMP and Wnt signaling in a region-specific manner and thereby allow proper Shh expression and development of the ventral telencephalon.

The authors clearly show that the genetic loss of Ring1b causes significant developmental phenotypes in the telencephalon (in this report, the brain is already much smaller by E11 and is accompanied by increased cell death). However, it is not very surprising that deletion of a Polycomb factor very early in the neuroepithelium would result in profound effects. From our understanding of Ring1 as part of the polycomb-repressive complex, it is already clear that Ring1b deletion would lead to de-repression of many important genes. In this paper, Ring1b KO caused depression of BMP and Wnt ligands, which likely contributed to the mis-expression of dorsal genes in ventral brain regions (and possibly also the loss of ventral gene expression). The authors have focused on genes related to dorsoventral patterning, but the data do not show that Ring1b "regulates" such patterning. Rather, the data simply show that Ring1 proteins are required for proper gene expression, including that related to dorsoventral patterning. Overall, these results seem fairly incremental to our knowledge of Ring1 function in neural development as well as its role as a transcriptional regulator.
Other comments: 1. Page 9: The subheading, "Ring1 promotes Shh expression and activates the Shh signaling pathway…" is an over-statement of the results. While Ring1b deletion causes loss of Shh expression, this result does not indicate that Ring1b "promotes" Shh expression. Indeed, the loss of Shh expression is likely quite indirect, as the authors show that Ring1b KO causes de-repression of BMP and Wnt ligands, and pharmacological activation of BMP/Wnt signaling reduced Shh expression in cultured cells. 2. In the abstract, the authors write, "Our results indicate that PcG suppresses BMP and Wnt in a region-specific manner so that Shh can be expressed properly…" The ChIP data do not strongly support this conclusion. The authors performed ChIP-qPCR analysis of E9 telencephalic tissues (not region-specific), and found that Ring1b and H3K27me3 are found at BMP/Wnt genes. This suggests that BMP/Wnt genes are targeted by PcG proteins throughout the brain. Do PcG proteins only target BMP and Wnt genes in ventral brain? If so, how is this regional-specificity achieved? 3. Does Ring1b deletion lead to loss of H3K27me3 and Ring1b at BMP and Wnt genes? This would be a much better control for these experiments. The authors should also consider performing H3K27me3 (and H2AK119ub) ChIP-seq, to analyze the chromatin changes that result from the early loss of Ring1b. This would also help distinguish which gene expression changes are more directly related to loss of Ring1b-mediated repression, from those that are indirectly related.

Reviewers' comments:
Reviewer #1 (Remarks to the Author): In this manuscript, Eto et al. study the role of the Polycomb group proteins Ring1A/B in the early developing mouse telencephalon using a conditional knockout approach. Deletion of Ring1A/B resulted in dorsalized expression of various transcription factors in neural progenitor cells of the ventral telencephalon, indicating that Polycomb group proteins are not only involved in the maintenance of the A-P axis during development, but also contribute to the regulation of dorsoventral patterning during the development of the early telencephalon. Transcriptome analysis revealed that Ring1B suppresses several major signaling pathways, including Wnt and BMP pathways, resulting in reduced Shh expression. Eto et al. went on to show that H3K27me3 and Ring1b are found at the promoters of a subset of BMP and Wnt pathway genes, providing a mechanism for the action of Ring1 proteins in early neural progenitor cells.
Overall, the study is well executed and the manuscript well written. The results contribute to a better understanding of the regulation of early telencephalon development and in particular, the role of Polycomb group-mediated regulation in this process. I find the study interesting and suitable for publication in Nature Communications.
We thank the reviewer for his/her strong support on our study and constructive criticisms, which we address in detail below.
Nevertheless, I have some points that should be taken into consideration: 1) The authors mention that the Sox1-Cre is expressed from <E8.5 and in combination with Ring1b F/F induces dorsalization of the early telencphalon (p. 6). In contrast, the Foxg1-IRES-Cre, expressed from E9.0, does not lead to this phenotype (data not shown, discussed on p. 13). I think would be helpful to include the data. Moreover, it would be important to know when the Ring1 proteins are lost. Currently, data is presented for E10, but since half a day appears to make a difference, it would be interesting to know when Ring1 action is required. Indeed, already at the very early neuroepithelial stage?
As suggested by the reviewer, we now include the data of Ring1B deletion with the use of the Foxg1-IRES-Cre mice (new Supplementary Fig. 10). We found that the proportion of the dorsal (Pax6 + and Neurog1 + ) and ventral (Nkx2.1 + and Ascl1 + ) telencephalic regions was not significantly different between these Ring1B-depleted mice and control mice at E11 (new Supplementary Fig. 10), indicating that the dorsalization phenotype was not found when the Ring1B gene was deleted by Foxg1-IRES-Cre. These new data are now mentioned in our revised discussion (page 13, lines 415-420).
Moreover, as suggested, we examined the time of reduction of Ring1B protein when the Ring1B gene was deleted by Sox1-Cre (new Supplementary Fig. 1a-d, h, i, l, m). We found that Ring1B proteins and H2AK119ub were already reduced in the telencephalon at E9 when the Ring1B gene was deleted by Sox1-Cre in control mice as well as in Ring1A KO mice (new Supplementary Fig. 1h-o). These results suggest a role of Ring1B action at the early developmental stage in establishing the dorsoventral patterning of telencephalon. These new data are now mentioned in our revised results (page 5, lines 131-144).
2) In addition, in Figure S1A it appears as if there was a gradient with Ring1B staining showing a greater reduction in the ventral region of the telencephalon. Is this a representative image?
No, this apparent gradient is not representative. We thank the reviewer for pointing this out. We thus replaced the images with more representative ones (new Supplementary Fig. 1a, b). Moreover, we quantified the level of Ring1B immunohistofluorescence signals along the dorsoventral axis of telencephalon (new Supplementary Fig. 1c, d). The Ring1B signal intensities appeared quite even along this axis in control telencephalon as well as in Ring1A KO telencephalon. These new data are now mentioned in our revised results and discussion (page 5, lines 136-138; page 13, lines 433-435).
3) The authors state (on p. 13) that they did not detect premature neurogenesis in the KO embryos (data not shown). It would be important to show the data as this is an alternative potential mechanism leading to a reduced brain size.
We agree with the reviewer that this is an important point. We now include the results regarding the effect of Ring1B deletion on the extent of neurogenesis at E9, when telencephalic NPCs start neurogenesis in BL6/J mice (new Supplementary Fig. 9). At this stage, a small number of cells were found to become positive for the neuronal marker βIII-tubulin in Ring1A-deleted telencephalon, in particular in the ventrocaudal part. Importantly, we found that Ring1B deletion did not increase βIII-tubulin + cells in Ring1A-deleted telencephalon, indicating that it does not cause premature neurogenesis in the early-stage telencephalon. These new data are now described in our revised discussion (page 13, lines 411-414).
4) In addition to Ring1b, H3K27me3 is strongly enriched at several Wnt and BMP pathway genes ( Figure 8B). Given that the classical model of consecutive PcG action (PRC2 => H3K27me3 => PRC1) has been repeatedly challenged, it would be interesting to known whether H3K27me3 levels are affected at Wnt and BMP genes upon Ring1 KO.
We thank the reviewer for raising this excellent point with regard to the effect of Ring1/PRC1 on H3K27me3. We thus investigated the levels of H3K27me3 at the Bmp4, Bmp7, Wnt7b and Wnt8b loci as well as those at the β-actin or Gapdh loci (negative controls) and the Hoxa1 or Hoxd3 loci (positive controls) by ChIP-qPCR in control and Ring1B-deleted telencephalon at E10 and, unexpectedly, found that Ring1B deletion did not significantly reduce the levels of H3K27me3 at these loci except for the Bmp7 locus (new Supplementary Fig. 7). This is a very important observation given that Ring1B deletion under the same condition was sufficient for the increase of their gene expression (new Fig. 4f, g, 5a-d, h-k ). These results suggest that H3K27me3 deposition alone is not sufficient for suppression of these BMP and Wnt gene loci while Ring1B is required for their suppression. This is consistent with a previously-proposed notion that PRC1 rather than PRC2 is responsible for gene suppression (Fursova et al., 2019; Blackledge et al., 2020). These new data are now described in our revised results (page 11, lines 340-344) and highlighted in discussion (page 14, lines 446-457).

5) While the Control and Ring1B
KO experiments are appropriately quantified, the IHC for Ring1A and Ring1A/B KO lacks quantification (Figures 1, 2, 3, 6). The quantification should be added.
According to the suggestion, we added quantification of the immunohistofluorescence signals for Nkx2.1, Pax6 and Neurog1 proteins (new Fig. 2e, f, k, 3g). We did not quantify the signals for Ascl1 protein and Shh mRNA in new Fig. 3f and 6e because they were under detectable levels in Ring1A/B dKO telencephalon. 6) Regarding the statistical analysis of experimental vs. control conditions, throughout the paper, the authors present data as Mean +/-SEM. Would the SD (standard deviation) not be more appropriate? Moreover, for several experiments (Figure 1, 2, 3, 4, 7), a paired t-test is applied, which I consider not justified as the animals are not subject to repeated measuring.
According to the suggestion, we changed SEM to SD throughout the paper.
As regards the statistical analyses, we used paired t test in several experiments because of the following reasons. In new Fig. 1f and 1h, the size of telencephalon was compared between control and Ring1B-deleted embryos among littermates, and we analyzed four litters. Given that the conditions between littermates are more similar than those between nonlittermates (with respect to environmental conditions and developmental age), we employed paired t test. In the experiments of new Fig. 2, 3, 4 and 7, the data were compared between control and Ring1B-deleted embryos among littermates, and we analyzed several litters. Moreover, the experiments such as immunostaining, chromatin immunoprecipitation and RT-qPCR analyses were performed separately for each litter. Again, given that the conditions between littermates are expected to be more similar than those between nonlittermates, we employed paired t test. 7) Regarding the gene expression analysis, an FDR of 0.15 was chosen. This appears very high considering that 15% of genes might be false positives. What was the rational for this cut-off?
Even though we chose the cut-off of 0.15 for FDR, p-values of individual genes were sufficiently low, with the highest p-value being 0.0173 among the genes with a FDR of <0.15 in new Supplementary Table 1. This is based on the definition of FDR by which the information of all genes is taken into account to evaluate a trend of many genes simultaneously (which is good for some analyses such as the pathway analysis we carried out in Fig. 4). Indeed, a number of papers chose an FDR of <0.15 (or a higher cut-off level) (e.g. a list of references).
Moreover, importantly, we confirmed the changes in expression of genes (related to Shh, BMP and Wnt signaling pathways by Ring1B deletion) picked up by this cut-off of FDR in the RNA-seq analysis with other means such as RT-qPCR (new Fig. 4f, g, 6b), in situ hybridization (new Fig. 5a-k) and immunohistochemical (Fig. 5l-n) analyses.
analysis of H3K27me3 in NPCs derived from DM, CTX, and V regions of the mouse telencephalon), and DRA010296 (CUT&Tag analysis of Ring1B in NPCs derived from DM, CTX, and V regions of the mouse telencephalon). The data will become open upon publication of this paper. 9) In Figure 4F, the 'Relative mRNA amount' of Wnt and Bmp pathway genes was determined by RT-qPCR. The KO conditions lacks error bars. Or was this KO set to a value of 1. This would be rather unconventional.
We agree with the reviewer and thus set the controls to a value of 1 in new Fig. 4f, g (instead of KO conditions). Figure 5A and B, the in situ hybridization signal is very difficult to see. Could the presentation be improved, possibly by showing even higher magnification zooms?

10) In
We appreciate the reviewer for pointing this out. To improve the quality of the images, we performed RNAscope®, a highly sensitive method for in situ hybridization (new Fig. 5a-k). We could detect clearer signals of Wnt8b, Axin2 and Bmp4 mRNA in mouse telencephalons at E10 and confirmed a ventral expansion of these mRNA by Ring1B deletion in the Ring1A KO background. Images at a higher magnification are also shown (new Fig. 5b, c, f, i, j). These new data are now mentioned in our revised results (page 8-9, lines 252-267). 11) There is a typo in line 385 ('independent').
We thank the reviewer for catching this typo and correct it.
We really appreciate the constructive comments from this reviewer. With the substantial amount of new data incorporated, we feel that our manuscript is significantly improved. --Reviewer #2 (Remarks to the Author): In this manuscript, Eto et al have investigated the role of the Polycomb group (PcG) protein Ring1 in mouse cortical development. The authors deleted Ring1b in the early neuroepithelium with the Sox1-Cre transgene, and by embryonic day 11 (E11), the telencephalon was significantly smaller. Apoptosis was significantly increased in Ring1b knockout (KO) mice, as well as the Ring1a/b double KO (dKO) mice. Because the brain was already significantly smaller by E11, the authors then focused most studies to E10-11. At E10, expression of ventral genes Nkx2.1 and Gsx2 was decreased in Ring1 KO mice. Ring1b KO mice also had increased expression of dorsal genes such as Emx1 in ventral brain regions. RNA-seq analysis of E11 Ring1b KO ventral brain revealed 953 upregulated genes, including those related to BMP and Wnt signaling. RNA-seq analysis also revealed 238 down-regulated genes including those related to Shh signaling. Addition of Shh agonist (SAG) to E10 Ring1b KO cultures caused upregulation of Shhresponsive genes such as Gli1 and Ptch1. Addition of BMP and Wnt activators to these cultures suppressed Shh expression. Blocking Shh signaling with cyclopamine did not affect BMP or Wnt ligands. ChIP-qPCR assays showed increased levels of H3K27me3 (a mark of polycomb activity) at Bmp4, Bmp7, Wnt 7b and Wnt8b. ChIP-qPCR showed Ring1B protein at Bmp4 and Wnt8b.
The authors clearly show that the genetic loss of Ring1b causes significant developmental phenotypes in the telencephalon (in this report, the brain is already much smaller by E11 and is accompanied by increased cell death). However, it is not very surprising that deletion of a Polycomb factor very early in the neuroepithelium would result in profound effects. From our understanding of Ring1 as part of the polycomb-repressive complex, it is already clear that Ring1b deletion would lead to de-repression of many important genes. In this paper, Ring1b KO caused depression of BMP and Wnt ligands, which likely contributed to the mis-expression of dorsal genes in ventral brain regions (and possibly also the loss of ventral gene expression). The authors have focused on genes related to dorsoventral patterning, but the data do not show that Ring1b "regulates" such patterning. Rather, the data simply show that Ring1 proteins are required for proper gene expression, including that related to dorsoventral patterning. Overall, these results seem fairly incremental to our knowledge of Ring1 function in neural development as well as its role as a transcriptional regulator.
We thank the reviewer for constructive criticisms, which we address in detail below.
Although it has been reported that Ring1/PcG regulates many developmental genes, this study provides the very first example to show that Ring1/PcG regulates morphogen expression in an area-specific manner.
Morphogenetic signals and their downstream transcription factors determine regional identity along the D-V axis in the developing central nervous system. Mutual inhibition between dorsal and ventral transcription factors plays a pivotal role in segregation and maintenance of regional identity, but the mechanisms that underlie the initial regional confinement of morphogen expression have remained largely unknown. We have now found that Ring1/PRC1 plays an essential role in establishment of the spatial expression patterns of morphogens along the D-V axis and in consequent regionalization of the telencephalon at the early stage of mouse development. Our new results (new Fig. 9 and Supplementary Fig. 8) also indicate that Ring1B and H3K27me3 are associated with genes encoding dorsal morphogens-such as Bmp4, Bmp7,