PRDM16 functions as a co-repressor in the BMP pathway to suppress neural stem cell proliferation

Reviewer #1 (Public review):

Summary:

This manuscript describes the role of PRDM16 in modulating BMP response during choroid plexus (ChP) development. The authors combine PRDM16 knockout mice and cultured PRDM16 KO primary neural stem cells (NSCs) to determine the interactions between BMP signaling and PRDM16 in ChP differentiation.

They show PRDM16 KO affects ChP development in vivo and BMP4 response in vitro. They determine genes regulated by BMP and PRDM16 by ChIP-seq or CUT&TAG for PRDM16, pSMAD1/5/8, and SMAD4. They then measure gene activity in primary NSCs through H3K4me3 and find more genes are co-repressed than co-activated by BMP signaling and PRDM16. They focus on the 31 genes found to be co-repressed by BMP and PRDM16. Wnt7b is in this set and the authors then provide evidence that PRDM16 and BMP signaling together repress Wnt activity in the developing choroid plexus.

Strengths:

Understanding context-dependent responses to cell signals during development is an important problem. The authors use a powerful combination of in vivo and in vitro systems to dissect how PRDM16 may modulate BMP response in early brain development.

Main weaknesses of the experimental setup:

(1) Because the authors state that primary NSCs cultured in vitro lose endogenous Prdm16 expression, they drive expression by a constitutive promoter. However, this means the expression levels are very different from endogenous levels (as explicitly shown in Supplementary Figure 2B) and the effect of many transcription factors is strongly dose-dependent, likely creating differences between the PRDM16-dependent transcriptional response in the in vitro system and in vivo.

(2) It seems that the authors compare Prdm16_KO cells to Prdm16 WT cells overexpressing flag_Prdm16. Aside from the possible expression of endogenous Prdm16, other cell differences may have arisen between these cell lines. A properly controlled experiment would compare Prdm16_KO ctrl (possibly infected with a control vector without Prdm16) to Prdm16_KO_E (i.e. the Prdm16_KO cells with and without Prdm16 overexpression.)

Other experimental weaknesses that make the evidence less convincing:

(1) The authors show in Figure 2E that Ttr is not upregulated by BMP4 in PRDM16_KO NSCs. Does this appear inconsistent with the presence of Ttr expression in the PRDM16_KO brain in Figure1C?

(2) Figure 3: The authors use H3K4me3 to measure gene activity. This is however, very indirect, with bulk RNA-seq providing the most direct readout and polymerase binding (ChIP-seq) another more direct readout. Transcription can be regulated without expected changes in histone methylation, see e.g. papers from Josh Brickman. They verify their H3K4me3 predictions with qPCR for a select number of genes, all related to the kinetochore, but it is not clear why these genes were picked, and one could worry whether these are representative.

(3) Line 256: The overlap of 31 genes between 184 BMP-repressed genes and 240 PRDM16-repressed genes seems quite small.

(4) The Wnt7b H3K4me3 track in Fig. 3G is not discussed in the text but it shows H3K4me3 high in _KO and low in _E regardless of BMP4. This seems to contradict the heatmap of H3K4me3 in Figure 3E which shows H3K4me3 high in _E no BMP4 and low in _E BMP4 while omitting _KO no BMP4. Meanwhile CDKN1A, the other gene shown in 3G, is missing from 3E.

(5) The authors use PRDM16 CUT&TAG on dissected dorsal midline tissues to determine if their 31 identified PRDM16-BMP4 co-repressed genes are regulated directly by PRDM16 in vivo. By manual inspection, they find that "most" of these show a PRDM16 peak. How many is most? If using the same parameters for determining peaks, how many genes in an appropriately chosen negative control set of genes would show peaks? Can the authors rigorously establish the statistical significance of this observation? And why wasn't the same experiment performed on the NSCs in which the other experiments are done so one can directly compare the results? Instead, as far as I could tell, there is only ChIP-qPCR for two genes in NSCs in Supplementary Figure 4D.

(6) In comparing RNA in situ between WT and PRDM16 KO in Figure 7, the authors state they use the Wnt2b signal to identify the border between CH and neocortex. However, the Wnt2b signal is shown in grey and it is impossible for this reviewer to see clear Wnt2b expression or where the boundaries are in Figure 7A. The authors also do not show where they placed the boundaries in their analysis. Furthermore, Figure 7B only shows insets for one of the regions being compared making it difficult to see differences from the other region. Finally, the authors do not show an example of their spot segmentation to judge whether their spot counting is reliable. Overall, this makes it difficult to judge whether the quantification in Figure 7C can be trusted.

(7) The correlation between mKi67 and Axin2 in Figure 7 is interesting but does not convincingly show that Wnt downstream of PRDM16 and BMP is responsible for the increased proliferation in PRDM16 mutants.

Weaknesses of the presentation:

Overall, the manuscript is not easy to read. This can cause confusion.

Reviewer #2 (Public review):

Summary:

This article investigates the role of PRDM16 in regulating cell proliferation and differentiation during choroid plexus (ChP) development in mice. The study finds that PRDM16 acts as a corepressor in the BMP signaling pathway, which is crucial for ChP formation.

The key findings of the study are:
(1) PRDM16 promotes cell cycle exit in neural epithelial cells at the ChP primordium.
(2) PRDM16 and BMP signaling work together to induce neural stem cell (NSC) quiescence in vitro.
(3) BMP signaling and PRDM16 cooperatively repress proliferation genes.
(4) PRDM16 assists genomic binding of SMAD4 and pSMAD1/5/8.
(5) Genes co-regulated by SMADs and PRDM16 in NSCs are repressed in the developing ChP.
(6) PRDM16 represses Wnt7b and Wnt activity in the developing ChP.
(7) Levels of Wnt activity correlate with cell proliferation in the developing ChP and CH.

In summary, this study identifies PRDM16 as a key regulator of the balance between BMP and Wnt signaling during ChP development. PRDM16 facilitates the repressive function of BMP signaling on cell proliferation while simultaneously suppressing Wnt signaling. This interplay between signaling pathways and PRDM16 is essential for the proper specification and differentiation of ChP epithelial cells. This study provides new insights into the molecular mechanisms governing ChP development and may have implications for understanding the pathogenesis of ChP tumors and other related diseases.

Strengths:

(1) Combining in vitro and in vivo experiments to provide a comprehensive understanding of PRDM16 function in ChP development.

(2) Uses of a variety of techniques, including immunostaining, RNA in situ hybridization, RT-qPCR, CUT&Tag, ChIP-seq, and SCRINSHOT.

(3) Identifying a novel role for PRDM16 in regulating the balance between BMP and Wnt signaling.

(4) Providing a mechanistic explanation for how PRDM16 enhances the repressive function of BMP signaling. The identification of SMAD palindromic motifs as preferred binding sites for the SMAD/PRDM16 complex suggests a specific mechanism for PRDM16-mediated gene repression.

(5) Highlighting the potential clinical relevance of PRDM16 in the context of ChP tumors and other related diseases. By demonstrating the crucial role of PRDM16 in controlling ChP development, the study suggests that dysregulation of PRDM16 may contribute to the pathogenesis of these conditions.

Weaknesses:

(1) Limited investigation of the mechanism controlling PRDM16 protein stability and nuclear localization in vivo. The study observed that PRDM16 protein became nearly undetectable in NSCs cultured in vitro, despite high mRNA levels. While the authors speculate that post-translational modifications might regulate PRDM16 in NSCs similar to brown adipocytes, further investigation is needed to confirm this and understand the precise mechanism controlling PRDM16 protein levels in vivo.

(2) Reliance on overexpression of PRDM16 in NSC cultures. To study PRDM16 function in vitro, the authors used a lentiviral construct to constitutively express PRDM16 in NSCs. While this approach allowed them to overcome the issue of low PRDM16 protein levels in vitro, it is important to consider that overexpressing PRDM16 may not fully recapitulate its physiological role in regulating gene expression and cell behavior.

(3) Lack of direct evidence for AP1 as the co-factor responsible for SMAD relocation in the absence of PRDM16. While the study identified the AP1 motif as enriched in SMAD binding sites in Prdm16 knockout cells, they only provided ChIP-qPCR validation for c-FOS binding at two specific loci (Wnt7b and Id3). Further investigation is needed to confirm the direct interaction between AP1 and SMAD proteins in the absence of PRDM16 and to rule out other potential co-factors.

Reviewer #3 (Public review):

Summary:

Bone morphogenetic protein (BMP) signaling instructs multiple processes during development including cell proliferation and differentiation. The authors set out to understand the role of PRDM16 in these various functions of BMP signaling. They find that PRDM16 and BMP co-operate to repress stem cell proliferation by regulating the genomic distribution of BMP pathway transcription factors. They additionally show that PRDM16 impacts choroid plexus epithelial cell specification. The authors provide evidence for a regulatory circuit (constituting of BMP, PRDM16, and Wnt) that influences stem cell proliferation/differentiation.

Strengths:

I find the topics studied by the authors in this study of general interest to the field, the experiments well-controlled and the analysis in the paper sound.

Weaknesses:

I have no major scientific concerns. I have some minor recommendations that will help improve the paper (regarding the discussion).

Author response:

Public Reviews:

Reviewer #1 (Public review):

Summary:

This manuscript describes the role of PRDM16 in modulating BMP response during choroid plexus (ChP) development. The authors combine PRDM16 knockout mice and cultured PRDM16 KO primary neural stem cells (NSCs) to determine the interactions between BMP signaling and PRDM16 in ChP differentiation.

They show PRDM16 KO affects ChP development in vivo and BMP4 response in vitro. They determine genes regulated by BMP and PRDM16 by ChIP-seq or CUT&TAG for PRDM16, pSMAD1/5/8, and SMAD4. They then measure gene activity in primary NSCs through H3K4me3 and find more genes are co-repressed than co-activated by BMP signaling and PRDM16. They focus on the 31 genes found to be co-repressed by BMP and PRDM16. Wnt7b is in this set and the authors then provide evidence that PRDM16 and BMP signaling together repress Wnt activity in the developing choroid plexus.

Strengths:

Understanding context-dependent responses to cell signals during development is an important problem. The authors use a powerful combination of in vivo and in vitro systems to dissect how PRDM16 may modulate BMP response in early brain development.

Main weaknesses of the experimental setup:

(1) Because the authors state that primary NSCs cultured in vitro lose endogenous Prdm16 expression, they drive expression by a constitutive promoter. However, this means the expression levels are very different from endogenous levels (as explicitly shown in Supplementary Figure 2B) and the effect of many transcription factors is strongly dose-dependent, likely creating differences between the PRDM16-dependent transcriptional response in the in vitro system and in vivo.

We acknowledge that our in vitro experiments may not ideally replicate the in vivo situation, a common limitation of such experiments, our primary aim was to explore the molecular relationship between PRDM16 and BMP signaling in gene regulation. Such molecular investigations are challenging to conduct using in vivo tissues. In vitro NSCs treated with BMP4 has been used a model to investigate NSC proliferation and quiescence, drawing on previous studies (e.g., Helena Mira, 2010; Marlen Knobloch, 2017). Crucially, to ensure the relevance of our in vitro findings to the in vivo context, we confirmed that cultured cells could indeed be induced into quiescence by BMP4, and this induction necessitated the presence of PRDM16. Furthermore, upon identifying target genes co-regulated by PRDM16 and SMADs, we validated PRDM16's regulatory role on a subset of these genes in the developing Choroid Plexus (ChP) (Fig. 7 and Suppl.Fig7-8). Only by combining evidence from both in vitro and in vivo experiments could we confidently conclude that PRDM16 serves as an essential co-factor for BMP signaling in restricting NSC proliferation.

(2) It seems that the authors compare Prdm16_KO cells to Prdm16 WT cells overexpressing flag_Prdm16. Aside from the possible expression of endogenous Prdm16, other cell differences may have arisen between these cell lines. A properly controlled experiment would compare Prdm16_KO ctrl (possibly infected with a control vector without Prdm16) to Prdm16_KO_E (i.e. the Prdm16_KO cells with and without Prdm16 overexpression.)

We agree that Prdm16 KO cells carrying the Prdm16-expressing vector would be a good comparison with those with KO_vector. However, despite more than 10 attempts with various optimization conditions, we were unable to establish a viable cell line after infecting Prdm16 KO cells with the Prdm16-expressing vector. The overall survival rate for primary NSCs after viral infection is low, and we observed that KO cells were particularly sensitive to infection treatment when the viral vector was large (the Prdm16 ORF is more than 3kb).

As an alternative oo assess vector effects, we instead included two other control cell lines, wt and KO cells infected with the 3xNLS_Flag-tag viral vector, and presented the results in supplementary Fig 2. When we compared the responses of the four lines — wt, KO, wt infected with the Flag vector, KO infected with the Flag vector — to the addition and removal of BMP4, we confirmed that the viral infection itself has no significant impacts on the responses of these cells to these treatments regarding changes in cell proliferation and Ttr induction.

Given that wt cells and the KO cells, with or without viral backbone infection behave quite similarly in terms of cell proliferation, we speculate that even if we were successful in obtaining a cell line with Prdm16-expressing vector in the KO cells, it may not exhibit substantial differences compared to wt cells infected with Prdm16-expressing vector.

Other experimental weaknesses that make the evidence less convincing:

(1) The authors show in Figure 2E that Ttr is not upregulated by BMP4 in PRDM16_KO NSCs. Does this appear inconsistent with the presence of Ttr expression in the PRDM16_KO brain in Figure1C?

The reviwer’s point is that there was no significant increase in Ttr expression in Prdm16_KO cells after BMP4 treatment (Fig. 2E), but there remained residule Ttr mRNA signals in the Prdm16 mutant ChP (Fig. 1C). We think the difference lies in the measuable level of Ttr expression between that induced by BMP4 in NSC culture and that in the ChP. This is based on our immunostaining expreriment in which we tried to detect Ttr using a Ttr antibody. This antibody could not detect the Ttr protein in BMP4-treated Prdm16_expressing NSCs but clearly showed Ttr signal in the wt ChP. This means that although Ttr expression can be significantly increased by BMP4 in vitro to a level measurable by RT-qPCR, its absolute quantity even in the Prdm16_expressing condition is much lower compared to that in vivo. Our results in Fig 1C and Fig 2E, as well as Fig 7B, all consistently showed that Prdm16 depletion significantly reduced Ttr expression in in vitro and in vivo.

(2) Figure 3: The authors use H3K4me3 to measure gene activity. This is however, very indirect, with bulk RNA-seq providing the most direct readout and polymerase binding (ChIP-seq) another more direct readout. Transcription can be regulated without expected changes in histone methylation, see e.g. papers from Josh Brickman. They verify their H3K4me3 predictions with qPCR for a select number of genes, all related to the kinetochore, but it is not clear why these genes were picked, and one could worry whether these are representative.

H3K4me3 has widely been used as an indicator of active transcription and is a mark for cell identity genes. And it has been demonstrated that H3K4me3 has a direct function in regulating transciption at the step of RNApolII pausing release. As stated in the text, there are advantages and disadvantages of using H3K4me3 compared to using RNA-seq. RNA-seq profiles all gene products, which are affected by transcription and RNA stability and turnover. In contrast, H3K4me3 levels at gene promoter reflects transcriptional activity. In our case, we aimed to identify differential gene expression between proliferation and quiescence states. The transition between these two states is fast and dynamic. RNA-seq may not be able to identify functionally relevant genes but more likely produces false positive and negative results. Therefore, we chose H3K4me3 profiling.

We agree that transcription may change without histone methylation changes. This may cause an under-estimation of the number of changed genes between the conditions.

We validated 7 out of 31 genes (Wnt7b, Id3, Mybl2, Spc24, Spc25, Ndc80 and Nuf2). We chose these genes based on two critira: 1) their function is implicated in cell proliferation and cell-cycle regulation based on gene ontology analysis; 2) their gene products are detectable in the developing ChP based on the scRNA-seq data. Three of these genes (Wnt7b, Id3, Mybl2) are not related to the kinetochore. We now clarify this description in the revised text.

(3) Line 256: The overlap of 31 genes between 184 BMP-repressed genes and 240 PRDM16-repressed genes seems quite small.

This indicates that in addition to co-repressing cell-cycle genes, BMP and PRDM16 have independent fucntions. For example, it was reported that BMP regulates neuronal and astrocyte differentiation (Katada, S. 2021), while our previous work demonstrated that Prdm16 controls temporal identity of NSCs (He, L. 2021).

(4) The Wnt7b H3K4me3 track in Fig. 3G is not discussed in the text but it shows H3K4me3 high in _KO and low in _E regardless of BMP4. This seems to contradict the heatmap of H3K4me3 in Figure 3E which shows H3K4me3 high in _E no BMP4 and low in _E BMP4 while omitting _KO no BMP4. Meanwhile CDKN1A, the other gene shown in 3G, is missing from 3E.

The track in Fig 3G shows the absolute signal of H3K4me3 after mapping the sequencing reads to the genome and normaliz them to library size. Compare the signal in Prdm16_E with BMP4 and that in Prdm16_E without BMP4, the one with BMP4 has a lower peak. The same trend can be seen for the pair of Prdm16_KO cells with or without BMP4. The heatmap in Fig. 3E shows the relative level of H3K4me3 in three conditions. The Prdm16_E cells with BMP4 has the lowest level, while the other two conditions (Prdm16_KO with BMP4 and Prdm16_E without BMP4) display a higher level. These two graphs show a consistent trend of H3K4me3 changes at the Wnt7b promoter across these conditions.

(5) The authors use PRDM16 CUT&TAG on dissected dorsal midline tissues to determine if their 31 identified PRDM16-BMP4 co-repressed genes are regulated directly by PRDM16 in vivo. By manual inspection, they find that "most" of these show a PRDM16 peak. How many is most? If using the same parameters for determining peaks, how many genes in an appropriately chosen negative control set of genes would show peaks? Can the authors rigorously establish the statistical significance of this observation? And why wasn't the same experiment performed on the NSCs in which the other experiments are done so one can directly compare the results? Instead, as far as I could tell, there is only ChIP-qPCR for two genes in NSCs in Supplementary Figure 4D.

In our text, we indicated the genes containing PRDM16 binding peaks in the figures and described them as “Text in black in Fig. 6A and Supplementary Fig. 5A”. We will add the precise number “25 of these genes” in the main text to clarify it. To define a negative control set of genes, we will use BMP-only repressed 184-31 =153 genes (excluding PRDM16-BMP4 co-repressed), and of these 153 genes, we will determine how many have PRDM16 peaks in the E12.5 ChP data, say X. Then we will use binomial test to calculate p-value binom_test(25, 31, X/153, alternative=“greater).

We are confused with the second part of the comment “And why wasn't the same experiment performed on the NSCs in which the other experiments are done so one can directly compare the results? Instead, as far as I could tell, there is only ChIP-qPCR for two genes in NSCs in Supplementary Figure 4D.” If the reviewer meant why we didn’t sequence the material from sequential-ChIP or validate more taget genes, the reason is the limitation of the material. Sequential ChIP requires a large quantity of the antibodies, and yields little material barely sufficient for a few qPCR after the second round of IP. This yielded amount was far below the minimum required for library construction. The PRDM16 antibody was a gift, and the quantity we have was very limited. We made a lot of efforts to optimize all available commercial antibodies in ChIP and Cut&Tag, but none of them worked.

(6) In comparing RNA in situ between WT and PRDM16 KO in Figure 7, the authors state they use the Wnt2b signal to identify the border between CH and neocortex. However, the Wnt2b signal is shown in grey and it is impossible for this reviewer to see clear Wnt2b expression or where the boundaries are in Figure 7A. The authors also do not show where they placed the boundaries in their analysis. Furthermore, Figure 7B only shows insets for one of the regions being compared making it difficult to see differences from the other region. Finally, the authors do not show an example of their spot segmentation to judge whether their spot counting is reliable. Overall, this makes it difficult to judge whether the quantification in Figure 7C can be trusted.

To address these questions, in the revised manuscript we will include an individal channel of Wnt2b and mark the boundaries. We will also provide full-view images and examples of spot segmentation in supplementary figures as space limitation in the main figures.

(7) The correlation between mKi67 and Axin2 in Figure 7 is interesting but does not convincingly show that Wnt downstream of PRDM16 and BMP is responsible for the increased proliferation in PRDM16 mutants.

We agree that this result (the correlation between mKi67 and Axin2) alone only suggests that Wnt signaling is related to the proliferation defect in the Prdm16 mutant, and does not necessarily mean that Wnt is downstream of PRDM16 and BMP. Our concolusion is backed up by two additional lines of evidences: the Cut&Tag data in which PRDM16 binds to regulatory regions of Wnt7b and Wnt3a; BMP and PRDM16 co-repress Wnt7b in vitro.

An ideal result is that down-regulating Wnt signaling in Prdm16 mutant can rescue Prdm16 mutant phenotype. Such an experiment is technically challenging. Wnt plays diverse and essential roles in NSC regulation, and one would need to use a celltype-and stage-specific tool to down-regulate Wnt in the background of Prdm16 mutation. Moreover, Wnt genes are not the only targets regulated by PRDM16 in these cells, and downregulating Wnt may not be sufficient to rescue the phenotype.

Weaknesses of the presentation:

Overall, the manuscript is not easy to read. This can cause confusion.

We will revise the text to improve the clarity.

Reviewer #2 (Public review):

Summary:

This article investigates the role of PRDM16 in regulating cell proliferation and differentiation during choroid plexus (ChP) development in mice. The study finds that PRDM16 acts as a corepressor in the BMP signaling pathway, which is crucial for ChP formation.

The key findings of the study are:

(1) PRDM16 promotes cell cycle exit in neural epithelial cells at the ChP primordium.

(2) PRDM16 and BMP signaling work together to induce neural stem cell (NSC) quiescence in vitro.

(3) BMP signaling and PRDM16 cooperatively repress proliferation genes.

(4) PRDM16 assists genomic binding of SMAD4 and pSMAD1/5/8.

(5) Genes co-regulated by SMADs and PRDM16 in NSCs are repressed in the developing ChP.

(6) PRDM16 represses Wnt7b and Wnt activity in the developing ChP.

(7) Levels of Wnt activity correlate with cell proliferation in the developing ChP and CH.

In summary, this study identifies PRDM16 as a key regulator of the balance between BMP and Wnt signaling during ChP development. PRDM16 facilitates the repressive function of BMP signaling on cell proliferation while simultaneously suppressing Wnt signaling. This interplay between signaling pathways and PRDM16 is essential for the proper specification and differentiation of ChP epithelial cells. This study provides new insights into the molecular mechanisms governing ChP development and may have implications for understanding the pathogenesis of ChP tumors and other related diseases.

Strengths:

(1) Combining in vitro and in vivo experiments to provide a comprehensive understanding of PRDM16 function in ChP development.

(2) Uses of a variety of techniques, including immunostaining, RNA in situ hybridization, RT-qPCR, CUT&Tag, ChIP-seq, and SCRINSHOT.

(3) Identifying a novel role for PRDM16 in regulating the balance between BMP and Wnt signaling.

(4) Providing a mechanistic explanation for how PRDM16 enhances the repressive function of BMP signaling. The identification of SMAD palindromic motifs as preferred binding sites for the SMAD/PRDM16 complex suggests a specific mechanism for PRDM16-mediated gene repression.

(5) Highlighting the potential clinical relevance of PRDM16 in the context of ChP tumors and other related diseases. By demonstrating the crucial role of PRDM16 in controlling ChP development, the study suggests that dysregulation of PRDM16 may contribute to the pathogenesis of these conditions.

Weaknesses:

(1) Limited investigation of the mechanism controlling PRDM16 protein stability and nuclear localization in vivo. The study observed that PRDM16 protein became nearly undetectable in NSCs cultured in vitro, despite high mRNA levels. While the authors speculate that post-translational modifications might regulate PRDM16 in NSCs similar to brown adipocytes, further investigation is needed to confirm this and understand the precise mechanism controlling PRDM16 protein levels in vivo.

While mechansims controlling PRDM16 protein stability and nuclear localization in the developing brain are interesting, the scope of this paper is revealing the function of PRDM16 in the choroid plexus and its interaction with BMP signaling. We will be happy to pursuit this direction in our next study.

(2) Reliance on overexpression of PRDM16 in NSC cultures. To study PRDM16 function in vitro, the authors used a lentiviral construct to constitutively express PRDM16 in NSCs. While this approach allowed them to overcome the issue of low PRDM16 protein levels in vitro, it is important to consider that overexpressing PRDM16 may not fully recapitulate its physiological role in regulating gene expression and cell behavior.

As stated above, we acknowledge that findings from cultured NSCs may not directly apply to ChP cells in vivo. We are cautious with our statements. The cell culture work was aimed to identify potential mechanisms by which PRDM16 and SMADs interact to regulate gene expression and target genes co-regulated by these factors. We expect that not all targets from cell culture are regulated by PRDM16 and SMADs in the ChP, so we validated expression changes of several target genes in the developing ChP and now included the new data in Fig. 7 and Supplementary Fig. 7. Out of the 31 genes identified from cultured cells, four cell cycle regulators including Wnt7b, Id3, Spc24/25/nuf2 and Mybl2, showed de-repression in Prdm16 mutant ChP. These genes can be relevant downstream genes in the ChP, and other target genes may be cortical NSC-specific or less dependent on Prdm16 in vivo.

(3) Lack of direct evidence for AP1 as the co-factor responsible for SMAD relocation in the absence of PRDM16. While the study identified the AP1 motif as enriched in SMAD binding sites in Prdm16 knockout cells, they only provided ChIP-qPCR validation for c-FOS binding at two specific loci (Wnt7b and Id3). Further investigation is needed to confirm the direct interaction between AP1 and SMAD proteins in the absence of PRDM16 and to rule out other potential co-factors.

We agree that the finding of the AP1 motif enriched at the PRDM16 and SMAD co-binding regions in Prdm16 KO cells can only indirectly suggest AP1 as a co-factor for SMAD relocation. That’s why we used ChIP-qPCR to examine the presence of C-fos at these sites. Although we only validated two targets, the result confirms that C-fos binds to the sites only in the Prdm16 KO cells but not Prdm16_expressing cells, suggesting AP1 is a co-factor. We results cannot rule out the presence of other co-factors.

Reviewer #3 (Public review):

Summary:

Bone morphogenetic protein (BMP) signaling instructs multiple processes during development including cell proliferation and differentiation. The authors set out to understand the role of PRDM16 in these various functions of BMP signaling. They find that PRDM16 and BMP co-operate to repress stem cell proliferation by regulating the genomic distribution of BMP pathway transcription factors. They additionally show that PRDM16 impacts choroid plexus epithelial cell specification. The authors provide evidence for a regulatory circuit (constituting of BMP, PRDM16, and Wnt) that influences stem cell proliferation/differentiation.

Strengths:

I find the topics studied by the authors in this study of general interest to the field, the experiments well-controlled and the analysis in the paper sound.

Weaknesses:

I have no major scientific concerns. I have some minor recommendations that will help improve the paper (regarding the discussion).

We will revise the discussion according the suggestions.