KDM6B interacts with TFDP1 to activate P53 signaling in regulating mouse palatogenesis

  1. Tingwei Guo
  2. Xia Han
  3. Jinzhi He
  4. Jifan Feng
  5. Junjun Jing
  6. Eva Janečková
  7. Jie Lei
  8. Thach-Vu Ho
  9. Jian Xu
  10. Yang Chai  Is a corresponding author
  1. Center for Craniofacial Molecular Biology, Herman Ostrow School of Dentistry, University of Southern California, United States

Abstract

Epigenetic regulation plays extensive roles in diseases and development. Disruption of epigenetic regulation not only increases the risk of cancer, but can also cause various developmental defects. However, the question of how epigenetic changes lead to tissue-specific responses during neural crest fate determination and differentiation remains understudied. Using palatogenesis as a model, we reveal the functional significance of Kdm6b, an H3K27me3 demethylase, in regulating mouse embryonic development. Our study shows that Kdm6b plays an essential role in cranial neural crest development, and loss of Kdm6b disturbs P53 pathway-mediated activity, leading to complete cleft palate along with cell proliferation and differentiation defects in mice. Furthermore, activity of H3K27me3 on the promoter of Trp53 is antagonistically controlled by Kdm6b, and Ezh2 in cranial neural crest cells. More importantly, without Kdm6b, the transcription factor TFDP1, which normally binds to the promoter of Trp53, cannot activate Trp53 expression in palatal mesenchymal cells. Furthermore, the function of Kdm6b in activating Trp53 in these cells cannot be compensated for by the closely related histone demethylase Kdm6a. Collectively, our results highlight the important role of the epigenetic regulator KDM6B and how it specifically interacts with TFDP1 to achieve its functional specificity in regulating Trp53 expression, and further provide mechanistic insights into the epigenetic regulatory network during organogenesis.

Editor's evaluation

Using the mouse secondary palate as a model, this study reports original findings on the function of Kdm6b, a H3K27me3 demethylase, in the regulation of embryonic development. The authors show that Kdm6b plays an essential role in neural crest development, and that loss of Kdm6b perturbs the p53 pathway, leading to complete clefting of the secondary palate along with cell proliferation and differentiation defects.

https://doi.org/10.7554/eLife.74595.sa0

Introduction

Embryonic development is a highly complex self-assembly process during which precursor cells are coordinated to generate appropriate cell types and assemble them into well-defined structures, tissues, and organs (Shahbazi et al., 2016). During this process, precursor cells undergo extensive and rapid cell proliferation until they reach the point of exit from the cell cycle to differentiate into various cell lineages (Ruijtenberg and van den Heuvel, 2016; Miermont et al., 2019). How these precursor cells modulate expression of different genes and proceed through diverse proliferation and differentiation processes is a very complex and interesting question. Growing evidence shows that epigenetic regulation, which includes mechanisms such as DNA methylation, histone modifications, chromatin accessibility, and higher-order organization of chromatin, provides the ability to modify gene expression and associated protein production in a cell type-specific manner, thus playing an essential role in achieving signaling specificity and regulating cell fate during embryonic development (Hanna et al., 2018).

Among these various layers of epigenetic regulation, DNA methylation and histone methylation are the best characterized and known to be key regulators of diverse cellular events (Bannister and Kouzarides, 2011; Smith and Meissner, 2013; Molina-Serrano et al., 2019). For example, methylation of lysine 27 on histone H3 (H3K27me) by methyltransferases is a feature of heterochromatin that renders it inaccessible to transcription factors, thus maintaining transcriptional repression, across many species (Wiles and Selker, 2017). On the other hand, methylation of H3K4me3 found near the promoter region can couple with the NURF complex to increase chromatin accessibility for gene activation (Wysocka et al., 2006; Soares et al., 2017). Demethylation, which results from removing a methyl group, also plays important roles during development. For instance, demethylation of H3K4 is required for maintaining pluripotency in embryonic stem cells, and demethylases KDM6A and KDM6B are required for proper gene expression in mature T cells (Lessard and Crabtree, 2010; Jambhekar et al., 2019). These studies clearly show that failure to maintain epigenomic integrity can cause deleterious consequences for embryonic development and adult tissue homeostasis (Henckel et al., 2007; Kim et al., 2009; Kang et al., 2019).

Palatogenesis is a complex process known to be regulated by multiple genetic regulatory mechanisms, including several signaling pathways (BMP, SHH, WNT, FGF, and TGFβ) and different transcription factors (such as Msx1, Sox9, Lhx6/8, Dlx5, and Shox2) (Satokata and Maas, 1994; Yu et al., 2005; Chai and Maxson, 2006; Levi et al., 2006; Cobourne et al., 2009; Lee and Saint-Jeannet, 2011; Nakamura et al., 2011; Bush and Jiang, 2012; He and Chen, 2012; Parada and Chai, 2012; Xu et al., 2016; Reynolds et al., 2019). However, environmental effects can also contribute to orofacial defects, which lends further support to the notion that genetic factors are not sufficient to fully explain the etiology of many birth defects (Dixon et al., 2011; Roessler et al., 2012; Seelan et al., 2012). Furthermore, case studies have revealed that heterozygous mutation of a chromatin-remodeling factor, SATB2, and variation in DNA methylation can cause cleft palate in patients (Leoyklang et al., 2007; Chandrasekharan and Ramanathan, 2014; Young et al., 2021). These cases have drawn our attention to the function of epigenetic regulation in palatogenesis.

The contribution of cranial neural crest cells (CNCCs) is critical to palate mesenchyme formation. Recently, studies have begun to address the role of epigenetic regulation in neural crest cell fate determination during development. For instance, homozygous loss of Arid1a, a subunit of SWI/SNF chromatin remodeling complex, in neural crest cells results in lethality in mice, associated with severe defects in the heart and craniofacial bones (Chandler and Magnuson, 2016). In addition, both lysine methyltransferase Kmt2a and demethylase Kdm6a are essential for cardiac and neural crest development (Shpargel et al., 2017; Sen et al., 2020). However, how these epigenetic changes lead to tissue-specific response during neural crest fate determination remain to be elucidated.

In this study, using palatogenesis as a model we investigated the functional significance of the demethylase Kdm6b in regulating the fate of CNCCs during palatogenesis. We have discovered that loss of Kdm6b in cranial neural crest (CNC)-derived cells results in complete cleft palate along with soft palate muscle defects. We also found cell proliferation and differentiation defects of CNC-derived cells in Kdm6b mutant mice. More importantly, our study shows that the level of H3K27me3 on the promoter of Trp53 (also known as P53 in human) is antagonistically controlled by Kdm6b and Ezh2. Furthermore, without Kdm6b, the transcription factor TFDP1, which binds to the promoter of Trp53, cannot activate expression of Trp53 in palatal mesenchymal cells. More importantly, the function of Kdm6b in activating Trp53 in these cells cannot be compensated for by the closely related histone demethylase Kdm6a. Our study highlights the importance of epigenetic regulation on cell fate decision and its function in regulating activity of Trp53 in CNC-derived cells during organogenesis.

Results

Loss of Kdm6b in CNC-derived cells results in craniofacial malformations

Previous research has shown that the X-chromosome-linked H3K27 demethylase KDM6A is indispensable for neural crest cell differentiation and viability as it establishes appropriate chromatin structure (Schwarz et al., 2014; Shpargel et al., 2017). However, we do not yet have a comprehensive understanding of the roles of two other members of the KDM6 family, Kdm6b and Uty, in regulating CNCCs during craniofacial development. More importantly, we have yet to understand how demethylase achieves its functional specificity in regulating downstream target genes. In order to elucidate the functions of Kdm6b and Uty, we first evaluated the expression patterns of KDM6 family members in the palatal region (Figure 1—figure supplement 1A–F). We found that, of these, Kdm6b is more abundantly expressed than Kdm6a and Uty in both palate mesenchymal and epithelial cells, which indicated it might play a critical role in regulating palatogenesis.

To investigate the tissue-specific function of Kdm6b during craniofacial development, we generated Wnt1Cre;Kdm6bfl/fl and Krt14Cre;Kdm6bfl/fl mice to specifically target the deletion of Kdm6b in CNC-derived and epithelial cells, respectively. Loss of Kdm6b in CNC-derived cells resulted in complete cleft palate in Wnt1Cre;Kdm6bfl/fl mice (90% phenotype penetrance, N = 10) and postnatal lethality at newborn stage (100% phenotype penetrance, N = 10) without interrupting expression of other KDM6 family members ( Figure 1A and B, Figure 1—figure supplement 1A–N). To evaluate when Kdm6b was inactivated in the CNC-derived cells, we also investigated the expression of Kdm6b at E9.5, well prior to the formation of the palate primordium, and found that Kdm6b was efficiently inactivated in the CNC-derived cells at this stage (Figure 1—figure supplement 1O and P). Interestingly, loss of Kdm6b in epithelial cells did not lead to obvious defects in the craniofacial region in Krt14Cre;Kdm6bfl/fl mice (Figure 1—figure supplement 2A–H). These results emphasized that Kdm6b is specifically required in CNC-derived cells during palatogenesis. CT images also confirmed the complete cleft palate phenotype and revealed that the most severe defects in the palatal region of Wnt1Cre;Kdm6bfl/fl mice were hypoplastic palatine processes of the maxilla and palatine bones (Figure 1C and D). Except for a minor flattened skull, other CNC-derived bones did not show significant differences between control and Wnt1Cre;Kdm6bfl/fl mice (Figure 1—figure supplement 2I–N). To evaluate the phenotype in more detail, we performed histological analysis and found that although the palatal shelves were able to elevate, the maxilla and palatine bones, as well as the palate stromal mesenchyme and soft palate muscles, failed to grow towards the midline in Wnt1Cre;Kdm6bfl/fl mice (Figure 1G–R, Figure 1—figure supplement 3A–P). Furthermore, in the posterior soft palate region, Wnt1Cre;Kdm6bfl/fl mice also showed morphological defects related to the orientation of muscle fibers and the pterygoid plate (Figure 1—figure supplement 3A–X). However, since the soft palate forms subsequent to the hard palate, it is difficult to identify whether the soft palatal muscle phenotype is a primary defect or a consequence resulting from an anterior cleft. Therefore, we focused on the anterior hard palate for further investigation. Collectively, these data indicate that mesenchymal Kdm6b is indispensable for craniofacial development and plays an essential role during palatogenesis.

Figure 1 with 3 supplements see all
Loss of Kdm6b in cranial neural crest (CNC)-derived cells results in cleft palate.

(A, B) Whole-mount oral view shows complete cleft palate phenotype in Wnt1Cre;Kdm6bfl/fl mice. Arrowhead in (B) indicates the cleft palate. Scale bar: 2 mm. (C, D) CT imaging reveals that the palatine process of the maxilla and palatine bone (PB) is missing in Wnt1Cre;Kdm6bfl/fl mice. White arrowheads in (C) indicate the palatine process of maxilla (PPM) in control mice, and red arrows indicate the PB. White asterisk in (D) indicates the missing PPM in Wnt1Cre;Kdm6bfl/fl mice, and red asterisk in (D) indicates the missing PB in Kdm6b mutant mice. Scale bars: 1 mm. (E, F) Sagittal views of CT images demonstrate the locations of HE sections in (G–R). Red lines indicate the locations of sections. Yellow arrow in (E) indicates palatal shelf, and yellow asterisk in (F) indicates cleft. Scale bars: 1 mm. (G–R) Histological analysis of control and Wnt1Cre;Kdm6bfl/fl mice. (H, J, L, N, P, R) are magnified images of boxes in (G, I, K, M, O, Q), respectively. Asterisks in (M, O, Q) indicate cleft in Kdm6b mutant mice. Scale bar: 200 µm. Mes: mesenchyme.

Kdm6b is critical for proliferation and differentiation of CNC-derived palatal mesenchymal cells

During craniofacial development, CNCCs migrate ventrolaterally and populate the branchial arches to give rise to distinct mesenchymal structures in the head and neck, such as the palate. Failure of CNCCs to populate pharyngeal arches causes craniofacial defects (Noden, 1983; Noden, 1991; Trainor and Krumlauf, 2000; Cordero et al., 2011). To determine whether Kdm6b mutant CNCCs successfully populate the first pharyngeal arch, which gives rise to the palatal shelves, we generated tdTomato reporter mice and collected samples at E10.5. The results showed that CNCCs’ migration was not adversely affected in Kdm6b mutant mice (Figure 2—figure supplement 1A and B). Then, we evaluated the process of palatogenesis at different embryonic stages and found that the cleft palate phenotype emerged as early as E14.5 in Wnt1Cre;Kdm6bfl/fl mice (Figure 2—figure supplement 1C and D). These data established that Kdm6b is not essential for CNCCs entering the pharyngeal arch but is specifically required in regulating post-migratory CNC-derived cells during palatogenesis.

Because cell proliferation defects in CNC-derived cells frequently lead to craniofacial defects, we tested whether loss of Kdm6b can affect cell proliferation using EdU labeling. After 2 hr of EdU labeling, we found that the number of cells positively stained with EdU was significantly increased in the CNC-derived palatal mesenchyme in Wnt1Cre;Kdm6bfl/fl mice compared to controls (Figure 2A–C). In addition, after 48 hr of EdU labeling, we found that the number of Ki67 and EdU double-positive cells was significantly increased in Wnt1Cre;Kdm6bfl/fl mice (Figure 2D–H). These results indicated that loss of Kdm6b in CNC-derived cells resulted in more cells remaining in the cell cycle and actively proliferating, which further led to hyperproliferation of mesenchymal cells in the palatal region of Wnt1Cre;Kdm6bfl/fl mice. Meanwhile, palatal mesenchymal cells of Wnt1Cre;Kdm6bfl/fl mice showed more expression of β-galactosidase, which suggested increased cellular senescence in Wnt1Cre;Kdm6bfl/fl mice (Figure 2—figure supplement 1E–G). To evaluate cellular senescence in vivo, we stained Lamin B1 in EdU-labeled samples. After 48 hr of EdU labeling, we found that there was less expression of Lamin B1 in the palatal mesenchyme of Wnt1Cre;Kdm6bfl/fl mice. More importantly, fewer EdU+ cells expressing Lamin B1 were observed in Wnt1Cre;Kdm6bfl/fl mice (Figure 2—figure supplement 1H–L). These data suggested that hyperproliferation may cause increased cellular senescence in palatal mesenchymal cells of Wnt1Cre;Kdm6bfl/fl mice.

Figure 2 with 1 supplement see all
Kdm6b is critical for proliferation and differentiation of cranial neural crest (CNC)-derived palatal mesenchyme cells.

(A, B) Immunostaining of EdU at E13.5 after 2 hr of EdU labeling. Dotted lines indicate palatal shelf region. Dashed lines indicate the palatal region used for quantification in (C). Scale bar: 50 µm. (C) Quantification of EdU+ cells represented in (A, B). *p<0.05. (D–G) Co-localization of EdU and Ki67 at E13.5 after 48 hr of EdU labeling. Dotted lines indicate palatal shelf region. Dashed lines indicate the palatal region used for quantification in (H). (F, G) are magnified images of boxes in (D, E). Arrows in (F, G) indicate representative cells that are only EdU+, while arrowheads indicate representative cells that are positive for both EdU and Ki67. Scale bar: 50 µm. (H) Quantification of EdU and Ki67 double-positive cells represented in (D, E). *p<0.05. (I–L) Immunostaining of RUNX2 at indicated stages. Insets are higher-magnification images of boxes in (I–L). Asterisks in (J, L) indicate decreased RUNX2+ cells observed in Wnt1Cre;Kdm6bfl/fl mice. Scale bar: 50 µm. White dotted lines indicate the palatal region used for quantification in (Q). (M–P) Immunostaining of SP7 at indicated stages. Insets are higher-magnification images of boxes in (O, P). Asterisk in (P) indicates decreased SP7+ cells observed in Wnt1Cre;Kdm6bfl/fl mice. Scale bar: 50 µm. White dotted lines indicate the palatal region used for quantification in (R). (Q, R) Quantification results for RUNX2+ and SP7+ cells represented in (I–P). *p<0.05. (S–W) Osteogenic differentiation assay using Alizarin red S staining. (W) is the quantification result of Alizarin red S staining represented in (S, T). Scale bars: 2 mm in (S, T); 200 µm in (U, V). *p<0.05.

Typically, cell proliferation and differentiation are inversely correlated. Differentiation of precursor cells is generally associated with arrested proliferation and permanently exiting the cell cycle (Ruijtenberg and van den Heuvel, 2016). To test whether cell differentiation was affected in the CNC-derived palatal mesenchyme in Wnt1Cre;Kdm6bfl/fl mice, we examined the distribution of the early osteogenesis marker RUNX2 and the later osteogenesis marker SP7 in the palatal region from E13.5 to E15.5 (Figure 2I–R). There was a decrease in the number of RUNX2+ cells in the palatal mesenchyme at both E13.5 and E14.5 in Wnt1Cre;Kdm6bfl/fl mice in comparison to the control (Figure 2I–L and Q). In addition, SP7+ cells were also decreased in Wnt1Cre;Kdm6bfl/fl mice at both E14.5 and E15.5 (Figure 2M–P and R). Furthermore, when we induced osteogenic differentiation in palatal mesenchymal cells from E13.5 embryos for 3 weeks, we found that cells from Wnt1Cre;Kdm6bfl/fl mice showed much less calcium deposition than cells from control mice, indicating a reduction in osteogenic potential in cells from Kdm6b mutant mice (Figure 2S–W). These results indicated that Kdm6b was indispensable for maintaining normal proliferation and differentiation of CNC-derived cells.

Loss of Kdm6b in CNC-derived cells disturbs P53 pathway-mediated activity

In order to identify the downstream targets of Kdm6b in the palatal mesenchyme, we performed RNA-seq analysis of palatal tissue at E12.5. The results showed that more genes were downregulated than upregulated in the palatal mesenchyme in Wnt1Cre;Kdm6bfl/fl mice (Figure 3A), which is consistent with the function of Kdm6b in removing the repressive mark H3K27me3. We further used Ingenuity Pathway Analysis (IPA) and Gene Ontology (GO) analysis to analyze the pathways that were most disturbed in the palatal mesenchyme in Kdm6b mutant mice. Surprisingly, both analyses indicated that pathways involving Trp53 might be disturbed in the palatal mesenchyme in Wnt1Cre;Kdm6bfl/fl mice (Figure 3B and C).

P53 signaling pathway is disturbed in Wnt1Cre;Kdm6bfl/fl mice.

(A) Bulk RNA-seq of palatal tissues collected at E12.5 represented in heatmap. Differentially expressed genes were selected using p<0.05 and fold change <–1.2 or >1.2. (B) Top seven signaling pathways disturbed in Kdm6b mutant mice, identified by Ingenuity Pathway Analysis. Red box indicates the top upregulated pathway observed in Wnt1Cre;Kdm6bfl/fl sample. (C) Top 10 signaling pathways identified by Gene Ontology analysis using differentially expressed genes identified by bulk RNA-seq analysis. Red box indicates P53 signaling is one of the top 10 pathways. X-axis shows the percentage of genes hit against total number of pathways hit. (D–G) Expression of Trp53 at E13.5 using RNAscope in situ hybridization. Dotted lines in (D, F) indicate palatal shelf. (E, G) are magnified images of boxes in (D, F). Asterisk in (G) indicates decreased expression of Trp53 observed in Wnt1Cre;Kdm6bfl/fl mice. Scale bar: 50 µm. (H) RT-qPCR quantification of Trp53 in palatal tissues collected at E13.5. *p<0.05. (I–L) Immunostaining of γH2AX at E13.5. Dotted lines in (I, K) indicate palatal shelf and dashed lines indicate quantification area. (J, L) are magnified images of boxes in (I, K), respectively. Arrowheads in (J, L) indicate representative γH2AX+ cells. Scale bar: 50 µm. (M) Quantification of γH2AX+ cells represented in (I, K). *p<0.05. (N–Q) Co-localization of EdU and γH2AX at E13.5 after 2 hr of EdU labeling. Dotted lines in (N, P) indicate palatal shelf region, while dashed lines indicate the palatal region used for quantification in (R). (O, Q) are magnified images of boxes in (N, P), respectively. Red arrows in (O, Q) indicate representative EdU+ cells with less than five γH2AX foci, while white arrowheads indicate representative cells that are positive for EdU and with greater than five γH2AX foci. Scale bar: 50 µm. (R) Quantification of EdU+ cells with greater than five γH2AX foci represented in (N, P).*p<0.05.

The tumor suppressor P53 plays prominent roles in regulating DNA damage response, including arresting cell growth for DNA repair, directing cellular senescence, and activating apoptosis (Mijit et al., 2020). Mutation of Trp53 is a major cause of cancer development (Williams and Schumacher, 2016). Previous research has shown that some homozygous Trp53 mutant mice exhibit craniofacial defects, including cleft palate, while inappropriate activation of Trp53 during embryogenesis also causes developmental defects, including craniofacial abnormalities (Tateossian et al., 2015; Bowen et al., 2019). These results suggest that precise dosage of Trp53 is indispensable for craniofacial development. We analyzed the expression of Trp53 in our samples and found that it significantly decreased in the palatal region of the Kdm6b mutant mice (Figure 3D–H). These results indicate that Kdm6b plays an important role in regulating the P53 pathway in the CNC-derived mesenchyme during palatogenesis. To further evaluate the consequence of downregulated Trp53 in Wnt1Cre;Kdm6bfl/fl mice, we assessed DNA damage, which are tightly related to the function of Trp53, in our study. We found that DNA damage increased, as indicated by γH2AX expression, in the palatal mesenchyme in Wnt1Cre;Kdm6bfl/fl mice (Figure 3I–M). More importantly, we observed significantly increased γH2AX foci in the EdU+ cells of Wnt1Cre;Kdm6bfl/fl palatal mesenchyme (Figure 3N–R). These data indicated that actively proliferating cells in Wnt1Cre;Kdm6bfl/fl mice experienced more severe DNA damage compared to those in the control mice, which might be the result of replication stress caused by the hyperproliferation we observed in Wnt1Cre;Kdm6bfl/fl mice.

Altered Trp53 expression is responsible for the developmental defects in Wnt1Cre;Kdm6bfl/fl mice

To further test whether downregulated expression of Trp53 is a key factor in the developmental defects we observed in Wnt1Cre;Kdm6bfl/fl mice, we transfected palatal mesenchymal cells from control mice with siRNA to knock down Trp53. qPCR revealed that the expression of Trp53 was significantly decreased in the cells treated with siRNA after 3 days (Figure 4—figure supplement 1A). At the same time, the group transfected with siRNA for Trp53 showed a significant increase in EdU+ cells (Figure 4A–C). In addition, significantly increased γH2AX+ cells were also observed in the group transfected with siRNA for Trp53 (Figure 4D–F). These data suggested that downregulated expression of Trp53 in the palatal mesenchymal cells is a key factor that led to the hyperproliferation and increased DNA damage we observed in Wnt1Cre;Kdm6bfl/fl mice. Furthermore, expression of both Runx2 and Sp7 was also significantly reduced in the palatal mesenchymal cells transfected with siRNA for Trp53 (Figure 4—figure supplement 1B and C), which indicated that the downregulated expression of Trp53 in the palatal mesenchymal cells resulted in differentiation defects, which were also observed in Wnt1Cre;Kdm6bfl/fl mice.

Figure 4 with 1 supplement see all
Altered Trp53 expression is responsible for the developmental defects in Wnt1Cre;Kdm6bfl/fl mice.

(A–C) Cells collected from E13.5 palatal tissue are transfected with siRNA to knock down expression of Trp53. Cell proliferation is evaluated using EdU labeling 3 days after transfection. (A, B) show proliferation of cells assessed by EdU labeling. Difference in EdU+ cells between mock- and siRNA-transfected groups is quantified in (C). Scale bar: 100 µm. *p<0.05. (D–F) Cells collected from E13.5 palatal tissue are transfected with siRNA to knock down expression of Trp53. DNA damage is evaluated using γH2AX 3 days after transfection. (D, E) show γH2AX+ cells. Difference in γH2AX+ cells between mock- and siRNA-transfected groups is quantified in (F). Scale bar: 100 µm. *p<0.05. (G–R) Histological analysis of control and Wnt1Cre;Kdm6bfl/fl mice treated with Nutlin-3. (J–L, P–R) are magnified images of boxes in (G–I, M–O), respectively. Scale bar: 200 µm. Mes: mesenchyme; PPM: palatine process of maxilla; PB: palatine bone.

To further investigate the function of Trp53 in Wnt1Cre;Kdm6bfl/fl mice, we tried to increase P53 in Kdm6b mutant mice using available small molecules. Previous research showed that MDM2, a ubiquitin ligase, specifically targets P53 for degradation and there is increased P53 activity in Mdm2 mutant mice, which exhibit a range of developmental defects (Arya et al., 2010; Bowen and Attardi, 2019; Bowen et al., 2019). Nutlin-3, an MDM2 inhibitor that can specifically interrupt interaction between MDM2 and P53, increases P53 in mouse primary neural stem progenitor cells and rescues neurogenic deficits in Fmr1 KO mice (Li et al., 2016). We treated pregnant mice with Nutlin-3 at a dosage based on their body weight at E10.5, E12.5, and E14.5 of pregnancy and then collected samples at E16.5 for analysis. To assess the potential influence of the solvent used to dissolve Nutlin-3 (10% DMSO in corn oil), we also treated mice with 10% DMSO in corn oil at the same embryonic stages. None of the Kdm6b mutant mice were rescued after this treatment (N = 3) (Figure 4—figure supplement 1D and E). In contrast, Nutlin-3 treatment successfully rescued the cleft palate in three out of five Wnt1Cre;Kdm6bfl/fl mice (Figure 4G–R). The remaining two showed a normal hard palate, but presented with posterior soft palate defects. Western blot analysis showed that the protein level of P53 was successfully restored in the Nutlin-3-treated group (Figure 4—figure supplement 1F and G). This result further revealed that downregulation of Trp53 in Wnt1Cre;Kdm6bfl/fl mice plays an essential role in the palatal defects. The genetic interaction between Kdm6b and Trp53 is important for the development of post-migratory CNCCs.

Level of H3K27me3 is antagonistically regulated by Kdm6b and Ezh2 during palatogenesis

The lysine-specific demethylase KDM6B is able to activate gene expression via removing the H3K27me3 repressive mark (Jiang et al., 2013). To investigate whether Kdm6b regulates the expression of Trp53 through modifying the level of H3K27me3, we first examined the status of H3K27me3 in our samples and found that loss of Kdm6b in CNC-derived cells resulted in accumulation of H3K27me3 in the nucleus of CNC-derived palatal mesenchymal cells (Figure 5A–D). Furthermore, immunoblotting revealed that the level of H3K27me3 was increased in the palatal region of Kdm6b mutant mice (Figure 5E). Since the level of H3K27me3 can also be modified by the methyltransferases EZH1 and EZH2, we further evaluated whether expression of EZH1 and EZH2 was affected in the palatal region. We found no obvious differences in either the distribution of Ezh1+ cells or the EZH1 protein level between control and Kdm6b mutant mice (Figure 5F–J). Similarly, no dramatic changes were observed in either the distribution of EZH2+ cells or its protein level between control and Kdm6b mutant mice (Figure 5K–O). These results indicated that increased H3K27me3 in Wnt1Cre;Kdm6bfl/fl mice was mainly caused by loss of Kdm6b in CNC-derived cells. However, we did notice a broader contribution and stronger signal of EZH2 than EZH1 in the CNC-derived palatal mesenchyme. To investigate whether an increase of H3K27me3 in the CNC-derived cells caused the cleft phenotype we observed in Wnt1Cre;Kdm6bfl/fl mice, we generated Wnt1Cre;Kdm6bfl/fl;Ezh2fl/+ mice and assessed the level of H3K27me3 in this model. In Wnt1Cre;Kdm6bfl/fl;Ezh2fl/+ mice, we observed a rescue of the abnormal accumulation of H3K27me3 (Figure 5P–V). More importantly, haploinsufficiency of Ezh2 in this model successfully rescued the cleft palate phenotype observed in Wnt1Cre;Kdm6bfl/fl mice (Figure 6A–O) with 70% efficiency (N = 10). CT scanning showed that both the palatine processes of the maxilla and palatine bone were restored in the Wnt1Cre;Kdm6bfl/fl;Ezh2fl/+ mice (Figure 6A–C). Both bone and palatal mesenchymal tissue were rescued in Wnt1Cre;Kdm6bfl/fl;Ezh2fl/+ mice (Figure 6D–O). Furthermore, both EdU+ and RUNX2+ cells were restored to normal levels in Wnt1Cre;Kdm6bfl/fl;Ezh2fl/+ mice (Figure 7A–H). These results suggested that an antagonistic interaction between the histone demethylase KDM6B and methyltransferase EZH2 that modulates H3K27me3 is essential for palatogenesis.

Level of H3K27me3 is antagonistically regulated by Kdm6b and Ezh2 during palatogenesis.

(A–E) Contribution of H3K27me3 in the palatal shelf is evaluated using immunostaining and Western blot at E13.5. Dotted lines in (A, C) indicate palatal shelf region. (B, D) are magnified images of boxes in (A, C). Asterisk in (B) indicates no accumulation of H3K27me3 observed in control mice. Arrowheads in (D) indicate accumulation of H3K27me3 observed in Wnt1Cre;Kdm6bfl/fl mice. Scale bar: 50 µm. (F–J) Contribution of Ezh1 in the palatal shelf is evaluated using RNAscope in situ hybridization and Western blot at E13.5. Dotted lines in (F, H) indicate palatal shelf region. (G, I) are magnified images of boxes in (F, H). Arrowheads in (G, I) indicate representative Ezh1+ cells. Scale bar: 50 µm. (K–O) Contribution of EZH2 in the palatal shelf is evaluated using immunostaining and Western blot at E13.5. Dotted lines in (K, M) indicate palatal shelf region. (L, N) are magnified images of boxes in (K, M). Arrowheads in (L, N) indicate representative EZH2+ cells. Scale bar: 50 µm. (P–V) Contribution of H3K27me3 in the palatal shelf of control mice, Kdm6b mutant mice, and EZH2 haploinsufficient model is evaluated using immunostaining and Western blot at E13.5. Dotted lines in (P, R, T) indicate palatal shelf region. (Q, S, U) are magnified images of boxes in (P, R, T), respectively. Asterisks in (Q, U) indicate no accumulation of H3K27me3 observed in control (Q) and Wnt1Cre;Kdm6bfl/fl;Ezh2fl/+ mice (U). White arrowheads in (S) indicate accumulation of H3K27me3 observed in Wnt1Cre;Kdm6bfl/fl mice. Scale bar: 50 µm.

Haploinsufficiency of Ezh2 rescues cleft palate in Wnt1Cre;Kdm6bfl/fl;Ezh2fl/+ mice.

(A–C) CT images at PN0.5. White arrowheads in (A, C) indicate palatine process of maxilla (PPM) observed in control and Wnt1Cre;Kdm6bfl/fl;Ezh2fl/+ rescue model. Red arrows in (A, C) indicate palatine bone observed in control and Wnt1Cre;Kdm6bfl/fl;Ezh2fl/+ rescue model. White asterisk in (B) indicates missing palate PPM in Wnt1Cre;Kdm6bfl/fl mice, and red asterisk indicates missing palatine bone in Kdm6b mutant mice. Scale bar: 0.4 mm. (D–F) Coronal views of CT images at PN0.5. Asterisk in (E) indicates cleft palate observed in Wnt1Cre;Kdm6bfl/fl mice. Scale bar: 0.3 mm. (G–I) Whole-mount oral view at PN0.5. Arrowhead in (H) shows complete cleft palate observed in Wnt1Cre;Kdm6bfl/fl mice. Dashed lines in (G–I) indicate location of sections in (J–O). Scale bar: 2 mm. (J–O) Histological analysis of samples at PN0.5. Asterisk in (K, N) indicates cleft palate in Wnt1Cre;Kdm6bfl/fl mice. (M–O) are magnified images of boxes in (J–L), respectively. Dotted lines in (M–O) outline the bone structure. Scale bar: 200 µm. Mes: mesenchyme.

EdU+ and RUNX2+ cells are restored in Wnt1Cre;Kdm6bfl/fl;Ezh2fl/+ mice.

(A–C) Immunostaining of EdU at E13.5 after 2 hr of EdU labeling. Dotted lines indicate palatal shelf region. Dashed lines indicate the palatal region used for quantification in (D). Scale bar: 50 µm. (D) Quantification of EdU+ cells represented in (A–C). ANOVA is used for statistical analysis. *p<0.05. N.S: not significant. (E–G) Immunostaining of RUNX2 at E13.5. Dotted lines indicate palatal shelf region. Dashed lines indicate the palatal region used for quantification in (H). Scale bar: 50 µm. (H) Quantification of RUNX2+ cells represented in (E–G). ANOVA is used for statistical analysis. *p<0.05. N.S: not significant.

Kdm6b activates expression of Trp53 through removing H3K27me3 at the promoter of Trp53 and interacts with transcription factor TFDP1 in regulating P53 signaling pathway

Chromatin accessibility represents the degree to which chromatinized DNA is able to physically interact with nuclear macromolecules such as transcription factors for gene regulation (Klemm et al., 2019). The repressive mark H3K27me3 is usually associated with facultative heterochromatin and results in transcriptional repression due to decreased chromatin accessibility (Wiles and Selker, 2017; Möller et al., 2019; den Broeder et al., 2020). The methyltransferase EZH2 and demethylases KDM6A/KDM6B can regulate the methylation status of H3K27 to affect gene expression (Pediconi et al., 2019). To test whether Kdm6b and Ezh2 can regulate expression of Trp53 via H3K27me3, we first examined whether deposition of H3K27me3 changes at the promoter of Trp53 in our models using ChIP-qPCR. A primer set was designed at 1127 bp upstream of Trp53 exon 1, and the results showed that deposition of H3K27me3 significantly increased at the promoter of Trp53 in the palatal region of Kdm6b mutant mice, while this increase was dampened in the Ezh2 haploinsufficiency model (Figure 8A). Meanwhile, haplosufficiency of Ezh2 in Wnt1Cre;Kdm6bfl/fl;Ezh2fl/+ mice was able to restore the decreased expression of Trp53 observed in the CNC-derived palatal mesenchyme of Wnt1Cre;Kdm6bfl/fl mice (Figure 8B–H). These data suggested that Kdm6b and Ezh2 co-regulate expression of Trp53 through H3K27me3. To further reveal whether KDM6B regulates the expression of Trp53 directly, we performed ChIP-qPCR using KDM6B antibody and found that deposition of KDM6B significantly increased at the promoter of Trp53 in the palatal region (Figure 8I). In addition, to test whether KDM6B has a unique role in activating Trp53 during palatogenesis, we transfected palatal mesenchymal cells from Kdm6b mutant mice with either a plasmid overexpressing Kdm6b or a plasmid overexpressing another histone demethylase, Kdm6a. Increased expression of Trp53 could be detected only in the group transfected with Kdm6b-overexpressing plasmid (Figure 8J and K, Figure 8—figure supplement 1A and B). These results indicated that Kdm6b has an essential and unique role in activating Trp53 during palatogenesis.

Figure 8 with 1 supplement see all
Kdm6b regulates expression of Trp53 through H3K27me3 and interacts with transcription factor TFDP1 in the activation of Trp53.

(A) ChIP-qPCR shows H3K27me3 deposition at the promoter region of Trp53 in palatal tissues of control, Kdm6b mutant, and Ezh2 haploinsufficient mice. ANOVA is used for statistical analysis. *p<0.05. N.S: not significant. (B–G) Expression of Trp53 in the palatal region at E13.5 using RNAscope in situ hybridization. Dotted lines in (B–D) indicate palatal shelf. (E–G) are magnified images of boxes in (B–D), respectively. Asterisk in (F) indicates decreased expression of Trp53 observed in Wnt1Cre;Kdm6bfl/fl mice. Scale bar: 50 µm. (H) RT-qPCR analysis of Trp53 expression in the palatal region of control, Kdm6b mutant, and Ezh2 haploinsufficient mice. ANOVA is used for statistical analysis. *p<0.05. N.S: not significant. (I) ChIP-qPCR shows KDM6B deposition at the promoter region of Trp53 in the palatal tissue of control mice. *p<0.05. (J, K) RT-qPCR analysis of Trp53 expression in palatal mesenchymal cells transfected with Kdm6b- or Kdm6a-overexpressing plasmids. *p<0.05. N.S: not significant. (L) ATAC-seq analysis indicates that the promoter region of Trp53 is accessible for transcription factor TFDP1. (M) ChIP-qPCR using palatal tissue shows that binding of TFDP1 to the promoter of Trp53 decreases in the Kdm6b mutant mice. *p<0.05. N.S: not significant. (N, O) Co-localization of TFDP1, Kdm6b, and Trp53 at E13.5 using immunostaining and RNAscope in situ hybridization. Dotted lines in (N) indicate palatal shelf. (O) is a magnified image of the box in (N). Arrowheads in (O) indicate representative cells that are positive for TFDP1, Kdm6b, and Trp53. Scale bar: 50 µm in (N) and 5 µm in (O). (P) RT-qPCR quantification shows the expression of Tfdp1 in samples collected at E13.5. N.S: not significant. (Q) RT-qPCR analysis of Trp53 expression in palatal mesenchymal cells after Tfdp1 siRNA transfection. *p<0.05. (R, S) RT-qPCR analysis of Trp53 expression in palatal mesenchymal cells transfected with Tfdp1 overexpressing plasmid. (R) *p<0.05. N.S: not significant. (T) Co-immunoprecipitation (Co-IP) experiment using protein extract from palatal tissues indicates that KDM6B and TFDP1 are present in the same complex. Anti-KDM6B antibody was used for immunoprecipitation (IP). IgG served as negative control. IB: immunoblotting.

Figure 8—source data 1

Source data for Figure 8A.

https://cdn.elifesciences.org/articles/74595/elife-74595-fig8-data1-v2.xlsx
Figure 8—source data 2

Source data for Figure 8H.

https://cdn.elifesciences.org/articles/74595/elife-74595-fig8-data2-v2.xlsx
Figure 8—source data 3

Source data for Figure 8I.

https://cdn.elifesciences.org/articles/74595/elife-74595-fig8-data3-v2.xlsx
Figure 8—source data 4

Source data for Figure 8J.

https://cdn.elifesciences.org/articles/74595/elife-74595-fig8-data4-v2.xlsx
Figure 8—source data 5

Source data for Figure 8K.

https://cdn.elifesciences.org/articles/74595/elife-74595-fig8-data5-v2.xlsx
Figure 8—source data 6

Source data for Figure 8M.

https://cdn.elifesciences.org/articles/74595/elife-74595-fig8-data6-v2.xlsx
Figure 8—source data 7

Source data for Figure 8P.

https://cdn.elifesciences.org/articles/74595/elife-74595-fig8-data7-v2.xlsx
Figure 8—source data 8

Source data for Figure 8Q.

https://cdn.elifesciences.org/articles/74595/elife-74595-fig8-data8-v2.xlsx
Figure 8—source data 9

Source data for Figure 8R.

https://cdn.elifesciences.org/articles/74595/elife-74595-fig8-data9-v2.xlsx
Figure 8—source data 10

Source data for Figure 8S.

https://cdn.elifesciences.org/articles/74595/elife-74595-fig8-data10-v2.xlsx
Figure 8—source data 11

Source data for Figure 8T.

https://cdn.elifesciences.org/articles/74595/elife-74595-fig8-data11-v2.zip

As an H3K27me3 demethylase, KDM6B is important for the regulation of chromatin structure for gene expression. To target a specific sequence in genome, a histone demethylase needs to interact with DNA binding proteins such as transcription factors or IncRNAs (Dimitrova et al., 2015; Gurrion et al., 2017). To identify a transcription factor that can interact with KDM6B, we performed ATAC-seq analysis of palate tissue at E13.5. Through motif analysis, we found that the promoter region of Trp53 was accessible to members of the E2F transcription factor family (E2F4 and E2F6) and transcription factor TFDP1 (also known as Dp1), a binding partner of E2F family members (Figure 8L, Figure 8—figure supplement 1C). We noticed that this open region was also located at the Transcription start site (TSS) of Wrap53, which was previously reported to regulate endogenous Trp53 mRNA levels and P53 protein levels (Mahmoudi et al., 2009). To evaluate whether the decrease of Trp53 was caused by altered expression of Wrap53 in our model, we examined the expression of Wrap53 in the palatal region at E13.5. Almost no expression of Wrap53 was detected in the palatal region (Figure 8—figure supplement 1D and E). Previous research reported that inactivation of E2Fs resulted in milder phenotypes than those associated with loss of Tfdp1, which leads to early embryonic lethality (Kohn et al., 2003). This result suggested that TFDP1 may play a more critical role than E2Fs during embryonic development. A motif of TFDP1 was detected 1011 bp upstream of Trp53 exon 1, which is very close to the H3K27me3 deposition site, by ATAC-seq analysis. ChIP-qPCR using palate tissue at E13.5 also revealed that binding of TFDP1 to the promoter region of Trp53 decreased in the Kdm6b mutant mice (Figure 8M). Immunohistochemistry analysis showed that TFDP1+ cells were distributed in the palatal region and co-expressed with Trp53 and Kdm6b (Figure 8N and O). We also confirmed co-expression of Kdm6b and Tfdp1 in the palatal region (especially enriched in Pax9+, Aldh1a2+, and Twist1+ cells) using our previously published scRNA-seq data (Figure 8—figure supplement 1F; Han et al., 2021). Meanwhile, the expression level and distribution of TFDP1 were not affected in the palatal mesenchyme of Wnt1Cre;Kdm6bfl/fl mice (Figure 8P, Figure 8—figure supplement 1G and H). These data indicated that Tfdp1 is not a downstream target of Kdm6b. To further test whether Tfdp1 regulated expression of Trp53 in the palatal mesenchymal cells, we transfected palatal mesenchymal cells from control mice at E13.5 using siRNA to knock down Tfdp1. qPCR revealed that the expression of Trp53 was decreased in cells treated with siRNA for Tfdp1 (Figure 8Q, Figure 8—figure supplement 1I). This data further indicated that Trp53 is a direct downstream target of Tfdp1.

To reveal the function of Kdm6b-Tfdp1 interaction in the regulation of Trp53 during palatogenesis, we transfected palatal mesenchymal cells with Tfdp1-overexpressing plasmid and found that the expression of Trp53 increased in the cells from control mice but not in the cells from Wnt1Cre;Kdm6bfl/fl mice (Figure 8R and S, Figure 8—figure supplement 1J and K). This result suggested that Kdm6b is specifically required and plays an essential role in the activation of Trp53 through interaction with Tfdp1 during palatogenesis. We performed co-immunoprecipitation (Co-IP) experiments and found that KDM6B and TFDP1 were indeed involved in the same complex (Figure 8T). Collectively, these data suggested that KDM6B and TFDP1 work together to activate Trp53 expression in the palatal mesenchyme and play an important role in regulating palatogenesis (Figure 9).

Summary schematic drawing.

Discussion

The development of an organism from a single cell to multiple different cell types requires tightly regulated gene expression (Bruneau et al., 2019). Transcription factors, which are among the key regulators of this process, are intimately involved in cell fate commitment (Nelms and Labosky, 2010; Soldatov et al., 2019). However, a transcription factor by itself cannot act on densely packed DNA in chromatin form. Thus, transcription factors must work in coordination with epigenetic regulatory mechanisms such as histone modifications, DNA methylation, chromatin remodeling, and others to dynamically regulate chromatin states for gene expression (Wilson and Filipp, 2018; Gökbuget and Blelloch, 2019). Insults to the epigenetic landscape due to genetic, environmental, or metabolic factors can lead to diverse developmental defects and diseases (Hobbs et al., 2014; Zoghbi and Beaudet, 2016; Flavahan et al., 2017). Cleft palate comprises 30% of orofacial clefts and can result from genetic mutations, environmental effects, or a combination thereof (Seelan et al., 2012). Much progress has been made in taking inventory of the gene mutations associated with craniofacial defects in recent years, and growing evidence has shown that epigenetic regulation plays an important role during neural crest development. For example, haploinsufficiency of KDM6A in humans causes severe psychomotor developmental delay, global growth restriction, seizures and cleft palate (Lindgren et al., 2013). Furthermore, studies have shown that Kdm6a and Arid1a are both indispensable during neural crest development (Chandler and Magnuson, 2016; Shpargel et al., 2017). DNA methyltransferase3A (DNMT3A) plays a critical role in mediating the transition from neural tube to neural crest fate (Hu et al., 2012). Meanwhile, loss of Ezh2, a component of PRC2, in CNC-derived cells completely prevents craniofacial bone and cartilage formation (Schwarz et al., 2014). These studies have clearly shown that epigenetic regulation is crucial for neural crest development. In this study, we further demonstrate the important role of epigenetic regulation during the neural crest contribution to palate development using Wnt1Cre;Kdm6bfl/fl mice as a model. We show that the demethylase KDM6B is not only required for normal CNC-derived palatal mesenchymal cell proliferation, but also for maintaining cell differentiation.

Epigenetic regulators, transcription factors, and lineage-specific genes work together to achieve spatiotemporally restricted, tissue-specific gene regulation (Hu et al., 2014). In this study, we reveal that Kdm6b works with the transcription factor Tfdp1 to specifically regulate the expression of Trp53 during palatogenesis. The molecular mechanisms underlying the function of Trp53 in genomic stability and tumor suppression have been studied extensively. However, the role of Trp53 in regulating the development of CNC-derived cells still remains largely unclear, although several studies have been conducted recently on certain aspects of this topic. For instance, it has been shown that Trp53 is able to coordinate CNC cell growth and epithelial-mesenchymal transition/delamination processes by modulating cell cycle genes and proliferation (Rinon et al., 2011). It has also been established that both deletion and overexpression of Trp53 result in craniofacial defects (Tateossian et al., 2015; Bowen et al., 2019). Furthermore, nuclear stabilization of P53 protein in Tcof+/- induces neural crest cell progenitors to undergo cell cycle arrest and caspase3-mediated apoptosis in the neuroepithelium. Inhibition of P53 function successfully rescues the neurocristopathy in an animal model of Treacher Collins syndrome, which results from mutation in Tcof1 (Jones et al., 2008). These studies have clearly shown that appropriate function of Trp53 is indispensable in CNCCs. However, none of these studies have addressed upstream regulation of Trp53 in CNCCs.

Here, we show that proper function of Trp53 during the differentiation and proliferation of CNCCs is orchestrated by Kdm6b and Ezh2 through H3K27me3. Altering the balance between Ezh2 and Kdm6b can cause abnormal H3K27me3 function, which further affects the downstream transcription factor Trp53. In addition, we have detected spontaneous DNA damage in the developing palate and increased accumulation of DNA damage in the Wnt1Cre;Kdm6bfl/fl mice. These findings further demonstrate the critical function of Trp53 in protecting embryonic cells from DNA damage during development. Furthermore, the ability of cells to proliferate is limited by the length of the telomeres, which gradually shorten during each cell replication (Blagoev, 2009). Once the telomeres are too short for DNA replication, the result is cellular senescence, which induces an irreversible inability to proliferate (Bernadotte et al., 2016). In this study, we show that downregulated expression of Trp53 in Wnt1Cre;Kdm6bfl/fl mice results in hyperproliferation and increased DNA damage in the proliferative cells, which might further lead to increased cell senescence. Previous studies reported that few Trp53-/- mice exhibit cleft palate or other craniofacial abnormalities, which suggests that loss of Trp53 alone is not powerful enough to cause defects during craniofacial development (Rinon et al., 2011; Tateossian et al., 2015) and that there might be other factors that can compensate for the loss of function of Trp53. In our study, Wnt1Cre;Kdm6bfl/fl mice showed a cleft palate phenotype with high penetrance, and using MDM2 inhibitor Nutlin-3 we successfully rescued cleft palate in Kdm6b mutant mice. Our results suggested that in the Kdm6b mutant mice the decrease of Trp53 cannot be compensated for by other factors, and disturbed P53 signaling is the key factor causing cleft palate in Kdm6b mutant mice.

Previous research has shown that KDM3A functions as a cofactor of STAT3 to activate the JAK2-STAT3 signaling pathway (Kim et al., 2018) and that KDM2A coordinates with c-Fos in regulating COX-2 (Lu et al., 2015). Our study shows that KDM6B coordinates with the transcription factor TFDP1 to activate expression of Trp53 in CNCCs, and that Ezh2 and Kdm6b co-regulate H3K27 methylation status, which may affect the ability of TFDP1 to bind to the chromatin during palatogenesis. It has been reported that Tfdp1 is crucial for embryonic development and regulating Wnt/β-catenin signaling (Kohn et al., 2003; Kim et al., 2012). Interaction between KDM6B and TFDP1 discovered in this study further increases our knowledge of the coordination between epigenetic regulators and transcription factors during organogenesis. As environmental insults can adversely affect the function of epigenetic regulators, our findings provide a better understanding of the epigenetic regulation and transcription factors involved in regulating the fate of CNC cells and craniofacial development, which can provide important clues about human development, as well as potential therapeutic approaches for craniofacial birth defects.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Strain, strain background (Mus musculus)Kdm6bflox/floxManna et al., 2015, Jackson LaboratoryStock# 029615;RRID:IMSR_JAX:029615
Strain, strain background (M. musculus)Ezh2flox/floxJackson LaboratoryStock# 022616;
RRID:IMSR_JAX:022616
Strain, strain background (M. musculus)Trp53flox/floxJackson LaboratoryStock# 008462;RRID:IMSR_JAX:008462
Strain, strain background (M. musculus)Wnt1CreZhao et al., 2008
Strain, strain background (M. musculus)Krt14creJackson LaboratoryStock# 018964;RRID:IMSR_JAX:018964
Strain, strain background (M. musculus)ROSA26loxp-STOP-loxp-tdTomatoJackson LaboratoryStock# 007905;RRID:IMSR_JAX:007905
Sequence-based reagentMm-Kdm6a probeAdvanced Cell DiagnosticsCat# 456961
Sequence-based reagentMm-Kdm6b probeAdvanced Cell DiagnosticsCat# 477971
Sequence-based reagentMm-Kdm6b-01 probeAdvanced Cell DiagnosticsCat# 501231
Sequence-based reagentMm-Uty probeAdvanced Cell DiagnosticsCat# 451741
Sequence-based reagentMm-Trp53 probeAdvanced Cell DiagnosticsCat# 402331
Sequence-based reagentMm-Trp53-C2 probeAdvanced Cell DiagnosticsCat# 402331-C2
Sequence-based reagentMm-Ezh1 probeAdvanced Cell DiagnosticsCat# 415501
Sequence-based reagentMm-Wrap53 probeAdvanced Cell DiagnosticsCat# 1143201-C1
AntibodyMyosin heavy chain (MHC) (mouse monoclonal)DSHBCat# P13538(1:10)
AntibodyHistone H3 tri methyl K27 (rabbit monoclonal)Cell Signaling TechnologyCat# 9733s(1:200)(1:1000)
AntibodyPhospho-histone H2A.X (rabbit monoclonal)Cell Signaling TechnologyCat# 9718s(1:200)
AntibodyDP1 (rabbit monoclonal)AbcamCat# ab124678(1:100)(1:1000)
AntibodyEZH2 (rabbit monoclonal)Cell Signaling TechnologyCat# 5246s(1:200)(1:2000)
AntibodyRUNX2 (rabbit monoclonal)Cell Signaling TechnologyCat# 12556s(1:200)
AntibodySP7 (rabbit polyclonal)AbcamCat# ab22552(1:200)
AntibodyLamin B1 (rabbit monoclonal)Cell Signaling TechnologyCat# 17416s(1:100)
AntibodyAnti-mouse Alexa Fluor 488 (goat polyclonal)Life TechnologiesCat# A11001(1:200)
AntibodyAnti-mouse Alexa Fluor 568 (goat polyclonal)Life TechnologiesCat# A-11004(1:200)
AntibodyAnti-rat Alexa Fluor 488 (goat polyclonal)Life TechnologiesCat# A-11006(1:200)
AntibodyAnti-rabbit Alexa Fluor 488 (goat polyclonal)Life TechnologiesCat# A-11008(1:200)
AntibodyAnti-rabbit Alexa Fluor 568 (goat polyclonal)Life TechnologiesCat# A-11036(1:200)
AntibodyEZH1 (rabbit polyclonal)AbcamCat# ab189833(1:1000)
AntibodyKDM6A (rabbit polyclonal)AbcamCat# ab36938(1:1000)
AntibodyKDM6B (C-term) (rabbit polyclonal)AbCEPTACat# AP1022b(1:1000)
AntibodyKDM6B (N-term) (rabbit polyclonal)AbCEPTACat# AP1022a(1:1000)
AntibodyP53 (mouse monoclonal)Santa CruzCat# sc-126(1:1000)
AntibodyHistone H3 (rabbit monoclonal)Cell Signaling TechnologyCat# 4499s(1:1000)
Antibodyβ-Actin (mouse monoclonal)AbcamCat# Ab20272(1:2000)
AntibodyRabbit IgG HRP-conjugated antibody (goat polyclonal)R&D SystemCat# HAF008(1:2000)
AntibodyMouse IgG HRP-conjugated antibody (goat polyclonal)R&D SystemCat# HAF007(1:2000)
AntibodyHRP, mouse anti-rabbit IgG LCS (mouse monoclonal)IPKineCat# A25022(1:2000)
Commercial assay or kitGoat anti-mouse IgG Alexa Fluor 488 Tyramide SuperBoost KitThermo Fisher ScientificCat# B40912(1:200)
Commercial assay or kitRNAscope Multiplex Fluorescent Kit v2Advanced Cell DiagnosticsCat# 323110
Commercial assay or kitRNAscope 2.5 HD Assay – REDAdvanced Cell DiagnosticsCat# 322350
Commercial assay or kitTSA Plus Cyanine 3 SystemPerkinElmerCat# NEL744001KT
Commercial assay or kitTSA Plus Fluoresceine SystemPerkinElmerCat# NEL771B001KT
Commercial assay or kitRNeasy Micro KitQIAGENCat# 74004
Commercial assay or kitDAB Peroxidase (HRP) Substrate Kit (with nickel)Vector LaboratoriesRRID:AB_2336382;Cat# SK4100
Software, algorithmImageJNIHRRID:SCR_003070
Software, algorithmIngenuity Pathway AnalysisQIAGEN, IncRRID:SCR_008653
Software, algorithmGraphPad PrismGraphPad SoftwareRRID:SCR_002798
Software, algorithmSeuratSatija labRRID:SCR_016341
Software, algorithmCell Ranger10X Genomics, IncRRID:SCR_017344

Animals

To generate Wnt1Cre;Kdm6bfl/fl mice, we crossed Wnt1Cre;Kdm6bfl/+ mice with Kdm6bfl/fl mice (Zhao et al., 2008; Manna et al., 2015). Reporter mice used in this study were tdTomato conditional reporter (JAX#007905) (Madisen et al., 2010). Ezh2fl/fl and Trp53fl/fl mice were purchased from Jackson Laboratory (JAX#022616, #008462) (Marino et al., 2000; Shen et al., 2008). Genotyping was carried out as previously described (Zhao et al., 2008). Briefly, tail samples were lysed by using DirectPCR tail solution (Viagen 102T) with overnight incubation at 55°C. After heat inactivation at 85°C for 1 hr, PCR-based genotyping (GoTaq Green MasterMix, Promega, and C1000 Touch Cycler, Bio-Rad) was used to detect the genes. All mouse studies were conducted with protocols approved by the Department of Animal Resources and the Institutional Animal Care and Use Committee (IACUC) of the University of Southern California (Protocols 9320 and 20299).

MicroCT analysis

Request a detailed protocol

MicroCT was used to analyze the control, Kdm6b, and other mutant samples. Mouse samples were dissected and fixed in 4% paraformaldehyde overnight at 4°C followed by CT scanning (Scanco Medical µCT50 scanner) at the University of Southern California Molecular Imaging Center as previously described (Grosshans et al., 2006; Sugii et al., 2017). AVIZO 9.1.0 (Visualization Sciences Group) was used for visualization and 3D microCT reconstruction.

Alcian blue-Alizarin red staining

Request a detailed protocol

Mouse heads were dissected and fixed in 95% EtOH overnight at room temperature. Staining was performed as previously described (Rigueur and Lyons, 2014). Briefly, 95% EtOH was replaced with 100% acetone for 2 days and then samples were incubated in Alcian blue solution (80% EtOH, 20% glacial acetic acid, and 0.03% [w/v] Alcian blue 8GX [Sigma, A3157]) for 1–3 days. Samples were then de-stained with 70% EtOH and incubated in 95% EtOH overnight. After incubation, samples were pre-cleared with 1% KOH and then incubated in Alizarin red solution (0.005% [w/v] Alizarin red [Sigma, A5533] in 1% [w/v] KOH) for 2–5 days. After clearing samples with 1% KOH, they were stored in 100% glycerol until analysis.

Sample preparation for sectioning

Request a detailed protocol

Samples for paraffin sectioning were prepared using the standard protocol in our laboratory. Briefly, samples were fixed in 4% PFA and decalcified with 10% EDTA as needed. Then, the samples were dehydrated with serial ethanol solutions (50, 70, 80, 90, and 100%) at room temperature followed by xylene and then embedded in paraffin wax. Sections were cut to 6 µm on a microtome (Leica) and mounted on SuperFrost Plus slides (Fisher, 48311-703). Cryosectioning samples were fixed and decalcified the same way as samples prepared for paraffin sectioning. Sucrose (15 and 30%) was used to remove water from the samples before embedding them in OCT compound (Tissue-Tek, 4583). Cryosections were cut to 8 µm on a cryostat (Leica) and mounted on SuperFrost Plus slides (Fisher).

Histological analysis

Request a detailed protocol

Paraffin sections prepared as described above were used for histological analysis. Hematoxylin and eosin staining was performed using the standard protocol (Cardiff et al., 2014).

Immunofluorescence assay

Request a detailed protocol

Cryosections and paraffin sections prepared as described above were used for immunofluorescence assays. Sections were dried for 2 hr at 55℃. Paraffin sections were deparaffinized and rehydrated before antigen retrieval. Heat-mediated antigen retrieval was used to process sections (Vector, H-3300) and then samples were blocked for 1 hr in blocking buffer at room temperature (PerkinElmer, FP1020). Primary antibodies diluted in blocking buffer were incubated with samples overnight at 4℃. After washing with PBST (0.1% Tween20 in 1× PBS), samples were then incubated with secondary antibodies at room temperature for 2 hr. DAPI (Sigma, D9542) was used for nuclear staining. All images were acquired using Leica DMI 3000B and Keyence BZ-X710/810 microscopes. Detailed information about the primary and secondary antibodies is listed in Supplementary file 1.

EdU labeling

Request a detailed protocol

EdU solution was prepared at 10 mg/mL in PBS, and then pregnant mice at the desired stage were given an intraperitoneal injection (IP) based on their weight (0.1 mg of EdU/1 g of mouse). Embryos were collected after 2 hr or 48 hr and then prepared for sectioning as above. EdU signal was detected using Click-It EdU cell proliferation kit (Invitrogen, C10337), and images were acquired using Leica DMI 3000B and Keyence BZ-X710/810 microscopes.

RNAscope in situ hybridization

Request a detailed protocol

RNAscope in situ hybridization in this study was performed on cryosections using RNAscope 2.5HD Reagent Kit-RED assay (Advanced Cell Diagnostics, 322350) and RNAscope multiplex fluorescent v2 assay (Advanced Cell Diagnostics, 323100) according to the manufacturer’s protocol. RNAscope probes used in this study included Kdm6a, Kdm6b, Uty, and Trp53. Detailed information about the probes is listed in Supplementary file 2.

RNA-sequencing and analysis

Request a detailed protocol

Palate samples from control and Wnt1Cre;Kdm6bfl/fl mice were collected at E12.5 for RNA isolation with RNeasy Micro Kit (QIAGEN) according to the manufacturer’s protocol. The quality of RNA samples was determined using an Agilent 2100 Bioanalyzer, and all samples for sequencing had RNA integrity (RIN) numbers > 9. cDNA library preparation and sequencing were performed at the USC Molecular Genomics Core. Single-end reads with 75 cycles were performed on Illumina HiSeq 4000 equipment, and raw reads were trimmed and aligned using TopHat (version 2.0.8) with the mm10 genome. CPM was used to normalize the data, and differential expression was calculated by selecting transcripts that changed with p< 0.05.

RNA extraction and real-time qPCR

Request a detailed protocol

Palatal tissue used for RNA isolation was dissected at desired stages, and an RNeasy Plus Micro Kit (QIAGEN, 74034) was used to isolate the total RNA followed by cDNA synthesis using an iScript cDNA synthesis kit (Bio-Rad, 1708891). Real-time qPCR quantification was done on a Bio-Rad CFX96 Real-Time system using SsoFast EvaGreen Supermix (Bio-Rad, 1725201). Detailed information about the primers is listed in Supplementary file 3.

ChIP-qPCR

Request a detailed protocol

Palate tissue was dissected from control and Wnt1Cre;Kdm6bfl/fl mice at E13.5. Each replicate contained 60–80 mg tissue combined from multiple animals. Samples were prepared following the manufacturer’s protocol (Chromatrap, 500191). Briefly, tissue was cut into small pieces and then fixed with 1% formaldehyde at room temperature for 15 min, followed by incubating with 0.65 M glycine solution. Then, the sample was washed twice with PBS, resuspended in Hypotonic Buffer, and incubated at 4℃ for 10 min to obtain nuclei, which were then resuspended in Digestion Buffer. After chromatin was sheared to 100–500 bp fragments using Shearing Cocktail, 10 µg chromatin with H3K27me3 antibody (CST 9733s, 1:50), DP1 antibody (Abcam ab124678, 1:10), KDM6B antibody (Abcepta AP1022b, 1:10), or immunoglobulin G-negative control (2 µg) was added to Column Conditioning Buffer to make up the final volume of 1000 µL. Immunoprecipitation (IP) slurry was mixed thoroughly and incubated on an end-to-end rotor for 1 hr at 4℃. An equivalent amount of chromatin was set as an input. After 1 hr incubation, IP slurry was purified using Chromatrap spin column at room temperature and chromatin was eluted using ChIP-seq elution buffer. Chromatin sample and input were further incubated at 65℃ overnight to reverse cross-linking. DNA was purified with Chromatrap DNA purification column after proteinase K treatment. ChIP eluates, negative control, and input were assayed using real-time qPCR. Primers were designed using the promoter region of Trp53. Detailed information is available in Supplementary file 3.

Western blot and co-immunoprecipitation

Request a detailed protocol

For Western blot, palate tissue was dissected from control and Wnt1Cre;Kdm6bfl/fl mice at E13.5. The tissue sample was lysed using RIPA buffer (Cell Signaling, 9806) with protease inhibitor (Thermo Fisher Scientific, A32929) for 20 min on ice followed by centrifugation at 4℃ to remove tissue debris. Protein extracts were then mixed with sample buffer (Bio-Rad, 1610747) and boiled at 98℃ for 10 min. Then, denatured protein extract was separated in 4–15% precast polyacrylamide gel (Bio-Rad, 456-1084) and then transferred to 0.45 µm PVDF membrane. Transferred membrane was incubated with 5% milk for 1 hr at room temperature and incubated with primary antibody at 4℃ overnight. After washing with TBST, the membrane was incubated with secondary antibody for 2 hr at room temperature and signals were detected using SuperSignal West Femto (Thermo Fisher Scientific, 34094) and Azure 300 (Azure Biosystems).

For Co-IP, palate tissue was dissected from control mice at E13.5 and 60–80 mg tissue was combined as one sample for each replicate. After lysing using RIPA buffer, 60 µL of the protein extract was mixed with sample buffer and boiled at 98℃ to serve as input. The remaining protein extract was incubated with primary antibody at 4℃ overnight. Protein G beads from GE Healthcare (GE Healthcare, 10280243) were used to purify the target protein, and then the protein sample was analyzed using Western blot. Detailed information about the primary and secondary antibodies is listed in Supplementary file 4.

siRNA and plasmid transfection

Request a detailed protocol

Palatal tissue was dissected from control and Wnt1Cre;Kdm6bfl/fl mice at E13.5, then cut into small pieces using a scalpel. This minced tissue was then cultured in DMEM medium (Gibco, 2192449) containing 40% MSC FBS (Gibco, 2226685P) and 1% Pen Strep (Gibco, 2145477) at 37℃.

siRNA (QIAGEN) and plasmid (OriGene) transfection was performed following the manufacturer’s protocol (QIAGEN, 301704, and OriGene, TF81001). Briefly, siRNA was transfected into cells in 24-well plates at 10 nM for 3 days followed by qPCR and EdU proliferation assay. Plasmid was transfected into cells in 24-well plates using 1 µg/µL stock solution for 2 days followed by real-time qPCR. The primers designed for qPCR are listed in Supplementary file 3. The siRNA sequence and plasmid information are listed in Supplementary files 5 and 6.

ATAC-seq analysis

Request a detailed protocol

Palate tissue of E13.5 control mice was digested using TrypLE express enzyme (Thermo Fisher Scientific, 12605010) and incubated at 37℃ for 20 min with shaking at 600 rpm. Single-cell suspension was prepared according to the 10X Genomics sample preparation protocol and processed to generate ATAC-seq libraries according to a published protocol (Buenrostro et al., 2015). Sequencing was performed using the NextSeq 500 platform (Illumina), and ATAC-seq reads were aligned to the UCSC mm10 reference genome using BWA-MEN (Li, 2013). Then, ATAC-seq peaks were called by MACS2 and annotated. Known transcription factor biding motifs were analyzed by HOMER (Zhang et al., 2008; Heinz et al., 2010). Quality files for sequencing are listed in Supplementary file 7.

Cell differentiation assay

Request a detailed protocol

Palatal tissue was dissected from control and Wnt1Cre;Kdm6bfl/fl mice at E13.5 and cultured as previously described. Then, the differentiation assay was conducted according to the manufacturer’s protocol (Gibco, A1007201) (Chen et al., 2020). Briefly, mesenchymal cells were seeded into cell culture plates at the desired concentration followed by incubation at 36°C in a humidified atmosphere of 5% CO2 for the required time (a minimum of 2 hr and up to 4 days). Then, the growth medium was replaced by complete differentiation medium and cells were continuously incubated for 3 weeks under osteogenic conditions. After specific periods of cultivation, cells were stained using 2% Alizarin red S solution (PH 4.2) solution. Images were acquired using EPSON Scan and Keyence BZ-X710/810 microscopes. Quantification of the Alizarin red S staining was conducted according to the manufacturer’s protocol (ScienCell, 8678).

Senescence β-galactosidase staining

Request a detailed protocol

Palatal tissue was dissected from control mice at E13.5, then cut into small pieces using a scalpel. This minced tissue was then cultured in DMEM medium (Gibco, 2192449) containing 40% MSC FBS (Gibco, 2226685P) and 1% Pen Strep (Gibco, 2145477) at 37℃. A cell monolayer was stained using a senescence β-galactosidase staining kit (Cell Signaling, 9860) according to the manufacturer’s protocol. Images were acquired using a Keyence BZ-X710/810 microscope.

Nutlin-3 treatment

Request a detailed protocol

Nutlin-3 (Sigma, N6287) was dissolved in corn oil (Sigma, C8267) with 10% DMSO (Sigma, D2650) and given to pregnant mice on days 10.5, 12.5, and 14.5 of pregnancy at a dosage based on their weight (10 mg/kg) (Li et al., 2016). Then, the embryos were collected at E16.5 for analysis.

Statistics

Statistical analysis was completed using GraphPad Prism, and significance was assessed by independent two-tailed Student’s t-test or ANOVA. The chosen level of significance for all statistical tests in this study was p<0.05. Data is presented as mean ± SEM. N = 3 samples were analyzed for each experimental group unless otherwise stated.

Data availability

Sequencing data have been deposited in GEO under accession code GSE175383.

The following data sets were generated
    1. Guo T
    2. Han X
    3. He J
    4. Jing J
    5. Lei J
    6. T-V Ho
    7. Xu J
    8. Chai Y
    9. Fng J
    10. Janeekova E
    (2022) NCBI Gene Expression Omnibus
    ID GSE175383. KDM6B interacts with TFDP1 to activate P53 signalling in regulating mouse palatogenesis.
The following previously published data sets were used
    1. Han X
    2. Feng J
    3. Guo T
    4. Loh Y-H
    5. Yuan Y
    6. T-V Ho
    7. Cho CK
    8. Li J
    9. Jing J
    10. Janeckova E
    11. He J
    12. Pei F
    13. Bi J
    14. Song B
    15. Chai Y
    (2021) NCBI Gene Expression Omnibus
    ID GSE155928. Runx2-Twist1 interaction coordinates cranial neural crest guidance of soft palate myogenesis.

References

    1. Noden DM
    (1991)
    Cell movements and control of patterned tissue assembly during craniofacial development
    Journal of Craniofacial Genetics and Developmental Biology 11:192–213.

Decision letter

  1. Marianne E Bronner
    Senior and Reviewing Editor; California Institute of Technology, United States
  2. Eric Liao
    Reviewer; Harvard Medical School, United States

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Kdm6b confers Tfdp1 with the competence to activate p53 signalling in regulating palatogenesis" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by Marianne Bronner as the Senior Editor and Reviewing Editor. The following individual involved in review of your submission has agreed to reveal their identity: Eric Liao (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

The reviewers agree that this is an interesting paper with the wealth of high quality data/ However, there are several areas that need further experimentation and clarification.

1) The results as written do not fully address the stated major aim of the study, i.e. investigating "how epigenetic regulators coordinate with tissue-specific regulatory factors during morphogenesis of specific organs". This point is overemphasized in the manuscript but not fully developed so should be toned down. As case in point, the abstract overstates and goes well beyond the data presented.

2) The regulations of the p53 pathway requires additional and deeper investigations plus added discussion.

3) Additional quantitative analyses are required to support claims regarding increased numbers of apoptotic cells and decreased numbers of Runx2 positive cells in mutant palate mesenchyme.

Reviewer #1 (Recommendations for the authors):

Kdm6b encodes a histone demethylase that specifically demethylates H3K27me3, a repressive histone mark, and thus plays important roles in transcriptional activation of gene expression. Previous studies have shown that mice homozygous for germline loss of Kdm6b function die perinatally, but whether Kdm6b plays an important role in palate development has not been documented. In this manuscript, the showed that the palatal mesenchyme of the Wnt1-Cre;Kmd6bfl/fl embryos had a defect in osteogenic differentiation by analysis of Runx2 and Sp7 expression and by using an in vitro osteogenic differentiation assay. The defect in palatogenesis in the Wnt1-Cre;Kmd6bfl/fl embryos correlated with increased levels of H3K27me3 in the palatal mesenchyme cells. By genetically reducing the dosage of Ezh2, which is a major H3K27 methyltransferase, they show that Ezh2 heterozygosity reduced the H3K27me3 in the palatal mesenchyme and partly rescued cleft palate in the Wnt1-Cre;Kmd6bfl/fl;Ezh2+/- embryos, indicating that antagonistic actions of Ezh2 and Kdm6b in H3K27 methylation are important in regulating palatogenesis. They performed RNA-seq analysis and found that the Wnt1-Cre;Kmd6bfl/fl embryos exhibited significant down-regulation of expression of 382 genes and significant upregulation of 259 genes in the developing palatal shelves at E12.5. Ingenuity pathway analysis of the RNA-seq data showed that the "Role of BRCA1 in DNA Damage" pathway was most significantly upregulated, whereas Gene Ontology analysis showed that the "p53 pathway" was among the top 10 pathways affected. They validated that the Wnt1-Cre;Kmd6bfl/fl embryos had increased DNA damage (marked by H2AX foci) in proliferating cells in the palatal mesenchyme and reduced expression of p53 mRNAs. Remarkably, treating the Wnt1-Cre;Kmd6bfl/fl embryos in utero through maternal IP injection of Nutlin-3, a small molecule inhibitor of MDM2-p53 interactions that causes stabilization of p53 protein, was able to rescue the cleft palate defect, indicating that repression of p53 expression in the Wnt1-Cre;Kmd6bfl/fl embryos plays an important role in the cleft palate pathogenesis. The manuscript further shows that the Wnt1-Cre;Kmd6bfl/fl embryonic palate had increased H3K27me3 at the p53 gene promoter region by ChIP-qPCR analysis. They performed ATAC-seq analysis of E13.5 wildtype embryonic palate and found that an ATAC-seq peak in the p53 gene promoter region contained E2F transcription factor binding motif. They show that Tfdp1, a member of the E2F family, was expressed in the embryonic palate and its expression was unaltered in the Wnt1-Cre;Kmd6bfl/fl embryonic palate. The performed ChIP-qPCR, of which the result suggested decreased Tfdp1 occupancy at the p53 gene promoter in the Wnt1-Cre;Kmd6bfl/fl embryonic palate than in control samples. They show that siRNA knockdown of Tfdp1 in primary palatal mesenchyme cells resulted in decrease in p53 mRNA expression and that overexpression of exogenous Tfdp1 caused increased p53 mRNA expression in control but not the Wnt1-Cre;Kmd6bfl/fl embryonic palate cells. In addition, they performed co-immunoprecipitation analysis and found that Tfdp1 and Kdm6b proteins were pulled down together from embryonic palatal extracts. They concluded that these results collectively suggested that Kdm6b and Tfdp1 work together to activate p53 expression in the palatal mesenchyme and play an important role in regulating palatogenesis.

In addition, this study generated RNA-seq data from E12.5 control and the Wnt1-Cre;Kmd6bfl/fl mutant embryonic palatal shelves, as well as ATAC-seq data from E13.5 wildtype mouse embryonic palatal tissues, which will serve as useful resources for palate development research. However, as written, the results do not directly address the stated major aim for investigating "how epigenetic regulators coordinate with tissue-specific regulatory factors during morphogenesis of specific organs".

The title, " Kdm6b confers Tfdp1 with the competence to activate p53 signaling in regulating palatogenesis", is not an accurate summary of the findings presented. First, the manuscript reports that Kdm6b mutant mice exhibit 90% penetrance of cleft palate, but among thousands of p53-null mice that have been studied in the last two decades few p53-null mice have been shown to have a cleft palate (of many primary research papers on studies of p53-null mice, only Tateossian et al. 2015 reported cleft palate in 2 out of 10 p53 homozygous mutant mice in one particular genetic background). Thus, the decreased expression of p53 in the palatal mesenchyme alone could not account for the cause of the disruptions in palate development in the Kdm6b mutant mice even though Nutlin-3 treatment, presumably through stabilization of the p53 protein, was able to partly rescue palate morphogenesis in the Kdm6b mutant embryos. Second, no data has been shown to indicate that Tfdp1 mediated activation of p53 is required for palate development. Third, the data presented indicate that Kdm6b function in the neural crest cells resulted in reduction in the H3K27me3 repression mark in the palatal mesenchyme cells, including in the p53 gene promoter region near where Tfdp1 binds, but there is no evidence that Kdm6b function is specific for Tfdp1 binding nor whether Kdm6-mediated demethylation of H3K27me3 at the p53 gene region is mediated by interaction with Tfdp1. Tfdp1 could be one of many transcription factors that could only bind to their target sites after heterochromatin-like repressive marks are removed.

Many conclusions and/or interpretation of the results, particularly the statements in the Abstract, in the manuscript were inaccurate.

1. The third and fourth sentences (Lines 36 – 37) in the Abstract appear to suggest that the major aim of the study was to use palatogenesis as a model to answer the question "how epigenetic regulators coordinate with tissue-specific regulatory factors during morphogenesis of specific organs", but the study does not address this aim at all. No "coordination" between Kdm6b mediated epigenetic regulation with any "tissue-specific" regulatory factor is demonstrated. The data show that lack of Kdm6b-mediated H3K27me3 demethylation in the neural crest cells resulted in repression of many genes, including p53, in the developing palatal mesenchyme and that Tfdp1 binding and activation of the p53 gene promoter may require prior Kdm6b-mediated removal of the H3K27me3 repressive mark.

2. Most of the conclusions/statements written in the Abstract were not accurate or not sufficiently supported by the data:

2a. Lines 41 – 42 (and similar sentences in Lines 99 – 100, Line 310, ) state, "activity of H3K27me3 on the promoter of p53 is precisely controlled by Kdm6b, and Ezh2 in regulating p53 expression in cranial neural crest cells". The data presented indicate opposing effects of Kdm6b and Ezh2 on the total amount of H3K27me3 in the palatal tissues and on the levels of p53 mRNA, but does not address how "precisely" they control of H3K27me3 levels at any particular locus.

2b. Lines 42 – 44 state, "Kdm6b renders chromatin accessible to the transcription factor Tfdp1, which binds to the promoter of p53 along with Kdm6b to specifically activate p53 expression during palatogenesis". There is no data presented to support for the second half of the sentence (and similar statements in Lines 101 and 426). There is no data showing binding of Tfdp1 and Kdm6b together at the p53 gene promoter. There is no data showing that binding of Tfdp1 at the p53 promoter is required for the p53 gene expression in the palatal mesenchyme. There is no data showing Kdm6b activates p53 expression through direct interaction with Tfdp1. There is no data showing that Tfdp1 mediated activation of p53 specifically occurs in the palatal tissues but not in other tissues when there is cellular stress.

2c. Lines 44 – 46, "our results highlight the important role of the epigenetic regulator Kdm6b and how it cooperates with Tfdp1 to achieve its functional specificity in regulating p53 expression,…" (and similar sentence in Lines 494 – 495). While the data showed that overexpression of Tfdp1 resulted in increased p53 mRNA expression in cultured palatal mesenchyme cells from control but not Kdm6b mutant embryos and the decrease in p53 mRNAs in the Kdm6b mutant palatal mesenchyme correlated with reduced Tfdp1 binding at the p53 gene promoter, the study has not provided evidence for direct cooperation between Kdm6b and Tfdp1 at the p53 gene promoter. It is quite possible that another unidentified factor interacts with Kdm6b to target Kdm6b to the p53 gene region among many other chromatin regions to remove the H3K27me3 repression mark in the developing palatal mesenchyme and Tfdp1 is only able to bind to the p53 gene promoter after the K3K27me3 demethylation. The results, particularly the demonstration of Ntlin-3 mediated rescue of palatal morphogenesis in the Kdm6b mutants, may have broad implications beyond Kdm6b mutant mouse model such that pharmacologically induced stabilization of p53 may be applicable for therapeutic intervention in cases where genetic-environment interactions disrupt developmental or other cellular processes while also inhibiting stress-induced activation of p53.

3. Lines 133 – 135 state, " although the palatal shelves were able to elevate, the maxilla and palatine bones, as well as the palate stromal mesenchyme and soft palate muscles, failed to grow towards the midline in Wnt1-Cre;Kdm6bfl/fl mice". However, the histology data in Figure 2—figure supplement 4, panels C/D/G/H, clearly show a failure or delay in palatal shelf elevation at E14.0 and E14.5 in the mutant embryos. The manuscript needs to accurately describe the results and investigate whether delay in palatal shelf elevation was the likely cause of cleft palate in the mutant.

4. Some numbers used to describe the results were inaccurate. For examples, Lines 119 – 120, "Loss of Kdm6b in CNC-derived cells resulted in complete cleft palate in Wnt1-Cre;Kmd6bfl/fl mice (90% phenotype penetrance, N=7)."

5. Given that the manuscript is focused on the effect of Kdm6b on the regulation of p53 in palatogenesis, the authors should discuss why few p53-/- mice had cleft palate, but Nutlin3 treatment was able to rescue the cleft palate defect in Kdm6b mutant mice at high efficiency. The manuscript indicated N=5 for the Nutlin-3 rescue experiment but did not indicate whether all 5 mutants were rescued. In addition, whereas the p53 mRNAs were reduced by about 50% in the Kdm6b mutant palate, the western blot in Figure 4—figure supplement 5F appears to show that Nutlin-3 treatment resulted in similar or more p53 protein in the mutant sample than the control. It would be helpful to quantify the relative levels of p53 protein in the Nutlin-3 treated mutant and control samples, and to discuss, if that is the result, how Nutlin-3 could induce p53 protein more efficiently in the mutant than in the control samples.

Reviewer #2 (Recommendations for the authors):

I have just a few comments that would not change the interpretation of the data nor would I request additional experiments.

1. Figure 2 - Figure Supplement 4, (A-B). Wnt1-Cre;Kdm6bfl/fl mice are compared to Wnt1-Cre;Kdm6bfl/+ , but why aren't they compared to a wild type control instead?

2. Figure 2 - Figure Supplement 4, (E-H). The TUNEL stain data should probably be quantified by using TUNEL+ cell numbers collected across several different sections/samples. The difference between the control and the mutant, as seen in the 4 images provided, is large if expressed as a percentage but still limited to 3 cells or fewer (especially at day E13.5). Even at E14.5, while there are 3 TUNEL+ cells seen in the palatal shelves of the mutant and 0 in the control, multiple TUNEL signals can be seen just outside of the dotted region in the control but not in the mutant; either the control has more cell death right next to the palatal shelves, or the small difference between the two can be explained as a sampling error without more data points.

3. L181-185, figure 2I-P; Similarly, can the authors state the sample size and quantify the RUNX2+ and SP7+ cells for easier comparison?

4. L213-219, figure 3B-C; Can the authors elaborate more on the identification of p53 from the analysis of the RNA-seq? As "epithelial adherens junction signaling" in IPA also corroborate with "cadherin signaling pathway" in GO. Same for "regulation of Actin-based motility by Rho" in IPA and "cytoskeletal regulation by Rho GTPase" in GO. Is there any other highly confident targets that are worth exploring?

5. Figure 4 - Figure Supplement 5, (F). The p53 band intensity difference between the control and the mutant in the 10% DMSO group is slight; the loading control band (actin) also demonstrates a small difference in size that follows the same pattern. It would be better to see the p53 band difference quantified.

6. L323-334, figure 5F-N; In contrast to what the authors claim, Ezh1 seems to have a stronger signal than Ezh2. Besides, the expression pattern between Ezh1 and Ezh2 is quite different. Can the authors support the claim by using normalized quantification? Otherwise, can the authors elaborate why they chose to KO Ezh2 in the Wnt1-Cre; Kdm6bfl/fl model?

7. Paper shows that inhibition of p53 by siRNA results in increased CNCC proliferation and DNA damage and reduces markers of osteogenesis, as seen in the siRNA treated cells. But is it possible to treat an embryo with a p53 inhibitor to directly demonstrate that loss of p53 in this model produces cleft palate? For example, Jones et al. treated pregnant mice with pifithrin-α to inhibit p53 (Nat. Med. 2011).

8. The paper describes how the control cells were treated with Tfdp1 siRNA, which resulted in lower levels of p53. Why wasn't the same treatment applied to the Wnt1-Cre;Kdm6bfl/fl cells, especially if both the controls and the mutants were later treated with the Tfdp1 overexpression plasmid?

Reviewer #3 (Recommendations for the authors):

In the present study, Tingwei Guo et al. use the mouse secondary palate as a model to assess the function of Kdm6b, a H3K27me3 demethylase, in the regulation of embryonic development. Guo's study shows that Kdm6b plays an essential role in neural crest development, and that loss of Kdm6b perturbs p53 pathway-mediated activity, leading to complete clefting of the secondary palate along with cell proliferation and differentiation defects.

In addition, the study reveals that Kdm6b and Ezh2 control p53 expression in cranial neural crest cells and that Kdm6b renders chromatin accessible to the transcription factor TFDP1 to activate p53 expression during palatogenesis. Together, the findings presented in this manuscript highlight the important role of the epigenetic regulator KDM6B and how it cooperates with TFDP1 to achieve its functional specificity in controlling p53 expression, and further provide mechanistic insights into the epigenetic regulatory network during secondary palate organogenesis.

– Over the last years, it has been reported by multiple groups that among the various layers of epigenetic regulation, DNA methylation and histone methylation are key drivers of diverse cellular events and developmental processes. In addition, it has been demonstrated that demethylation also plays important roles during development. For instance, demethylation of H3K4 is required for maintaining pluripotency in embryonic stem cells, and the demethylases KDM6A and KDM6B are required for proper gene expression. Indeed, the concept that failure to maintain epigenomic integrity can cause deleterious consequences for embryonic development has been extensively explored by various groups and is not novel per se. In addition, both lysine methyltransferase Kmt2a and demethylase Kdm6a have been recently shown to be essential for cardiac and neural crest development. For example, Shpargel reported that mice carrying neural crest deletion of Kdm6a exhibit craniofacial defects, including cleft or arched palate, cardiac abnormalities, and postnatal growth retardation, modeling the clinical features of Kabuki syndrome (Shpargel et al. PNAS, 2017). In summary, roles of these demethylases in neural crest development are already known. However, how these epigenetic changes lead to tissue-specific responses during neural crest fate determination and differentiation remains poorly understood and understudied, making the current manuscript of interest and timely.

– Epigenetic regulation plays extensive roles in development and diseases and its disruption not only can cause multiple developmental defects, but also increases the risk of neoplastic transformation. However, our knowledge of how epigenetic regulators coordinate with tissue-specific regulatory factors to control tissue and organ morphogenesis is still rudimentary. Therefore, the present paper will be of interest to the craniofacial biology community and to the broader developmental biology community, as well as to all those devoted to the study of the epigenetic and transcriptional regulation of morphogenesis and organogenesis.

– The study is robust, detailed, and comprises a wealth of original results and data of high quality, illustrated through many elegant figures. There are only some points of concern that need to be addressed, mainly related to additional quantitative analyses that are required for some of the experiments discussed in the manuscript and the need for clarifications regarding the regulation of the p53 pathway.

– Nomenclature: protein names should be written in uppercase throughout the text and in the figures. In the current manuscript use of the nomenclature is not consistent. Often protein and gene names are not listed correctly. See: http://www.informatics.jax.org/mgihome/nomen/gene.shtml#ps

– Figure 1: In Wnt1-Cre;Kdm6bfl/fl embryos the tongue appears to be grossly dysmorphic and abnormally positioned (e.g. see panel Q). If this is only due to a technical artifact of the section, then a better image should be chosen for Figure 1Q. If the tongue is instead abnormally formed and enlarged, then this result should be discussed and well documented, given that withdrawal of the tongue from between the vertical palatal shelves is required for their seamless fusion and closure during embryonic development.

– Figure 2 – Figure Suppl 4E-H: The Authors state that there is increased cell death in the palatal mesenchyme of Wnt1-Cre;Kdm6bfl/fl mice compared to controls, based on TUNEL staining, as shown in the figure. It is difficult to be convinced of this result, given that the TUNEL-positive cells are extremely sparse in both the control and mutant shown in these panels. If the Authors are convinced of their finding based on multiple experiments, they should show sections that better illustrate the defect, as well as quantify the numbers of TUNEL-positive cells over multiple sections for both control and mutant samples. In addition, they should include a graph comprising the total numbers of cells that were counted, the percentage of TUNEL-positive cells in both control and mutant, the fold increase of TUNEL-positive cells in the mutant, and the statistical significance. Unless these quantitative experiments are included, the Authors should definitely delete "increased cell death" for the mutant from the summary model shown in Figure 8.

However, if the Authors were to prove the presence of an increase of cell death in the palatal mesenchyme of Wnt1-Cre;Kdm6bfl/fl mice by conducting additional experiments, this finding would somehow contradict the result – which is very convincing – that p53 expression is decreased in mutant tissue (shown currently in Figure 7). Interestingly, in the discussion the authors mention (line 517): "In this study we notice that downregulated expression of p53 in Wnt1-Cre;Kdm6bfl/fl mice results in hyperproliferation and increased DNA damage in the proliferative cells, which might further lead to cell senescence." Assessing cell senescence, instead of cell death, in the mutant compared to the control could be revealing.

– Figure 2I-L: The Authors state: "There was a decrease in the number of Runx2+ cells in the palatal mesenchyme at both E13.5 and E14.5 in Wnt1-184 Cre;Kdm6bfl/fl mice in comparison to the control" (Line 182-184). Either the Authors quantify the "number of cells" (see also comment above), or they could do a qRT-PCR to evaluate Runx2 mRNA in mutant versus control tissues.

– Figure 3B: It would be beneficial to separate the genes that are upregulated from the genes that are downregulated in Wnt1-Cre;Kdm6bfl/fl mice versus controls. The overall message would be much clearer.

– Figure 6: The rescue experiments are one of the main strengths of this study: beautiful and rigorous use of genetics to validate the pathways under analysis! However, it would add a great deal of strength to the rescue experiments to also evaluate the rescue at the cellular level (i.e. assess the rescue of proliferation and/or DNA damage, even if partial, or alternatively, examine whether expression of Sp7 or Runx2 is rescued by qRT-PCR).

– Figure 7, panel I and Figure 7 – Figure Suppl 6: Red box that is described as highlighting the p53 promoter is placed instead over the promoter of a different gene, Wrap53. Indeed, this red box contains the TSS of Wrap53 and not p53. The Wrap53 gene encodes an essential component of the telomerase holoenzyme complex, a ribonucleoprotein complex required for telomere synthesis. The encoded protein interacts with other components of active telomerase and with small Cajal body RNAs (scaRNAs), which are involved in modifying splicing RNAs. It was reported that Wrap53 also functions as a p53 antisense transcript, which regulates endogenous p53 mRNA levels and the levels of P53 protein by targeting the 5' untranslated region of p53 mRNA (Mahmoudi et al. Mol Cell 2009). Therefore, Wrap53 can indirectly alter p53 levels of expression and regulation. siRNA knockdown of Wrap53 results in significant decrease in p53 mRNA and suppression of P53 induction upon DNA damage. Conversely, overexpression of Wrap53 increases p53 mRNA and protein levels (Mahmoudi et al. Molecular Cell 2009; Farnebo et al., Cell Cycle 2009; Saldana-Meyer et al., Genes and Development 2013). These studies unequivocally demonstrated that Wrap53 regulates p53. As it turns out, the primers used for the ChIP-qPCR experiment shown in Figure 7 were now blasted and they correspond to the Wrap53 promoter and not to the p53 promoter. Given all the knowledge discussed above, the current statement: "factor Tfdp1 binds to the promoter of p53 along with Kdm6b to specifically activate the expression of this tumor suppressor gene" should be re-evaluated and clarified. This point should be better investigated. As the distance between the Wrap53 and p53 genes is only approximately 1kb, in additional experiments the Authors should use primers specific to the p53 promoter and to the Wrap53 promoter together with chromatin sonicated to an average size of 300-600 bp, which should provide a clear answer as to whether TFDP1 binds directly to the p53 promoter, or to the Wrap53 promoter, or to both. If TFDP1 does not bind directly to p53 but to the Wrap53 promoter instead, regulation of p53 transcription would then be indirect, via Wrap53. This finding would still be of interest. This reviewer believes that this point is important and should be adequately clarified and examined in further depth.

– Figure 7 – Figure Suppl 6: The Authors also analyze expression patterns of Kdm6b and its co-expression with Tfdp1 by scRNAseq. The scRNAseq data sets have already been described by the same Authors in a(nother) interesting paper that was recently published. There appears to be a striking enrichment of both Kdm6b and Tfdp1 genes in a particular cell subpopulation emerging from the scRNAseq datasets. It would be very interesting to know which cell type comprises this particular cluster? Also, the individual Panels in B should be enlarged to better visualize each cell clusters. It would also be very helpful to the reader to list the specific cell types that comprise each cluster in the Panels in B.

https://doi.org/10.7554/eLife.74595.sa1

Author response

Essential revisions:

The reviewers agree that this is an interesting paper with the wealth of high quality data/ However, there are several areas that need further experimentation and clarification.

1) The results as written do not fully address the stated major aim of the study, i.e. investigating "how epigenetic regulators coordinate with tissue-specific regulatory factors during morphogenesis of specific organs". This point is overemphasized in the manuscript but not fully developed so should be toned down. As case in point, the abstract overstates and goes well beyond the data presented.

We thank reviewers for this comment, and we have added two experiments addressing the reviewer’s concern. One is ChIP-qPCR, which showed KDM6B is deposited at the promoter region of Trp53. Another experiment we have added used Kdm6b- and Kdm6a-overexpressing (OE) plasmids to transfect cells from the palatal mesenchyme of Kdm6b mutant mice. Expression of Trp53 increased in the cells transfected with Kdm6b OE plasmid, but not in the cells transfected with Kdm6a OE plasmid. These results suggested that not all the histone demethylases can activate expression of Trp53 in the palatal mesenchyme and that Kdm6b is the critical and functional specific for activating expression of Trp53 during palatogenesis. We also adjusted our statement in the manuscript, which has been highlighted in the revised version.

2) The regulations of the p53 pathway requires additional and deeper investigations plus added discussion.

We thank the reviewers for this suggestion. As we mentioned above, we have added two experiments regarding regulation of Trp53. We have also added some discussion regarding the phenotypes of Trp53-/- mice per the reviewers’ suggestions, which is highlighted in the Discussion section.

3) Additional quantitative analyses are required to support claims regarding increased numbers of apoptotic cells and decreased numbers of Runx2 positive cells in mutant palate mesenchyme.

We have assessed cellular senescence instead of cell death according to the reviewer’s suggestion. Data has been added to Figure2—figure supplement 1E-G. Both numbers of RUNX2+ and SP7+ cells in Figure 2 have been quantified per the reviewers’ suggestion.

Reviewer #1 (Recommendations for the authors):

[…]

Many conclusions and/or interpretation of the results, particularly the statements in the Abstract, in the manuscript were inaccurate.

1. The third and fourth sentences (Lines 36 – 37) in the Abstract appear to suggest that the major aim of the study was to use palatogenesis as a model to answer the question "how epigenetic regulators coordinate with tissue-specific regulatory factors during morphogenesis of specific organs", but the study does not address this aim at all. No "coordination" between Kdm6b mediated epigenetic regulation with any "tissue-specific" regulatory factor is demonstrated. The data show that lack of Kdm6b-mediated H3K27me3 demethylation in the neural crest cells resulted in repression of many genes, including p53, in the developing palatal mesenchyme and that Tfdp1 binding and activation of the p53 gene promoter may require prior Kdm6b-mediated removal of the H3K27me3 repressive mark.

We thank the reviewer for this comment and have rewritten these sentences as follows: “However, the question of how epigenetic changes lead to tissue-specific responses during neural crest fate determination and differentiation remains understudied.”

2. Most of the conclusions/statements written in the Abstract were not accurate or not sufficiently supported by the data:

2a. Lines 41 – 42 (and similar sentences in Lines 99 – 100, Line 310, ) state, "activity of H3K27me3 on the promoter of p53 is precisely controlled by Kdm6b, and Ezh2 in regulating p53 expression in cranial neural crest cells". The data presented indicate opposing effects of Kdm6b and Ezh2 on the total amount of H3K27me3 in the palatal tissues and on the levels of p53 mRNA, but does not address how "precisely" they control of H3K27me3 levels at any particular locus.

We thank the reviewer for this comment and have rewritten these sentences as follows: “Furthermore, activity of H3K27me3 on the promoter of Trp53 is antagonistically controlled by Kdm6b and Ezh2 in cranial neural crest cells.”

2b. Lines 42 – 44 state, "Kdm6b renders chromatin accessible to the transcription factor Tfdp1, which binds to the promoter of p53 along with Kdm6b to specifically activate p53 expression during palatogenesis". There is no data presented to support for the second half of the sentence (and similar statements in Lines 101 and 426). There is no data showing binding of Tfdp1 and Kdm6b together at the p53 gene promoter. There is no data showing that binding of Tfdp1 at the p53 promoter is required for the p53 gene expression in the palatal mesenchyme. There is no data showing Kdm6b activates p53 expression through direct interaction with Tfdp1. There is no data showing that Tfdp1 mediated activation of p53 specifically occurs in the palatal tissues but not in other tissues when there is cellular stress.

We thank the reviewer for this comment and have rewritten these sentences as follows: “More importantly, without Kdm6b, the transcription factor TFDP1, which normally binds to the promoter of Trp53, cannot activate Trp53 expression in palatal mesenchymal cells. Furthermore, the function of Kdm6b in activating Trp53 in these cells cannot be compensated for by the closely related histone demethylase Kdm6a.”

2c. Lines 44 – 46, "our results highlight the important role of the epigenetic regulator Kdm6b and how it cooperates with Tfdp1 to achieve its functional specificity in regulating p53 expression,…" (and similar sentence in Lines 494 – 495). While the data showed that overexpression of Tfdp1 resulted in increased p53 mRNA expression in cultured palatal mesenchyme cells from control but not Kdm6b mutant embryos and the decrease in p53 mRNAs in the Kdm6b mutant palatal mesenchyme correlated with reduced Tfdp1 binding at the p53 gene promoter, the study has not provided evidence for direct cooperation between Kdm6b and Tfdp1 at the p53 gene promoter. It is quite possible that another unidentified factor interacts with Kdm6b to target Kdm6b to the p53 gene region among many other chromatin regions to remove the H3K27me3 repression mark in the developing palatal mesenchyme and Tfdp1 is only able to bind to the p53 gene promoter after the K3K27me3 demethylation. The results, particularly the demonstration of Ntlin-3 mediated rescue of palatal morphogenesis in the Kdm6b mutants, may have broad implications beyond Kdm6b mutant mouse model such that pharmacologically induced stabilization of p53 may be applicable for therapeutic intervention in cases where genetic-environment interactions disrupt developmental or other cellular processes while also inhibiting stress-induced activation of p53.

We thank the reviewer for this comment and have rewritten these sentences as follows: “Collectively, our results highlight the important role of the epigenetic regulator KDM6B and how it specifically interacts with TFDP1 to achieve its functional specificity in regulating Trp53 expression, and further provide mechanistic insights into the epigenetic regulatory network during organogenesis.”

3. Lines 133 – 135 state, " although the palatal shelves were able to elevate, the maxilla and palatine bones, as well as the palate stromal mesenchyme and soft palate muscles, failed to grow towards the midline in Wnt1-Cre;Kdm6bfl/fl mice". However, the histology data in Figure 2—figure supplement 4, panels C/D/G/H, clearly show a failure or delay in palatal shelf elevation at E14.0 and E14.5 in the mutant embryos. The manuscript needs to accurately describe the results and investigate whether delay in palatal shelf elevation was the likely cause of cleft palate in the mutant.

We thank the reviewer for this comment. In Wnt1Cre;Kdm6bfl/fl mice, the development of the palate is generally delayed compared to the control mice. There is some variation at E14.5, but in most of the samples we investigated, the palatal shelves were able to elevate in Kdm6b mutant mice, as shown in Figure2 L and N. All the samples at E15.5 are accurately represented in Figure2 P. Furthermore, all the Kdm6b mutant mice are accurately represented by Figure1 M-R at newborn stage. We have never observed palatal shelves oriented vertically beside the tongue from E15.5 to newborn stage (n>10). Therefore, we conclude there is no elevation defect in Kdm6b mutant mice.

4. Some numbers used to describe the results were inaccurate. For examples, Lines 119 – 120, "Loss of Kdm6b in CNC-derived cells resulted in complete cleft palate in Wnt1-Cre;Kmd6bfl/fl mice (90% phenotype penetrance, N=7)."

We thank reviewer for this comment and have edited accordingly.

5. Given that the manuscript is focused on the effect of Kdm6b on the regulation of p53 in palatogenesis, the authors should discuss why few p53-/- mice had cleft palate, but Nutlin3 treatment was able to rescue the cleft palate defect in Kdm6b mutant mice at high efficiency. The manuscript indicated N=5 for the Nutlin-3 rescue experiment but did not indicate whether all 5 mutants were rescued. In addition, whereas the p53 mRNAs were reduced by about 50% in the Kdm6b mutant palate, the western blot in Figure 4—figure supplement 5F appears to show that Nutlin-3 treatment resulted in similar or more p53 protein in the mutant sample than the control. It would be helpful to quantify the relative levels of p53 protein in the Nutlin-3 treated mutant and control samples, and to discuss, if that is the result, how Nutlin-3 could induce p53 protein more efficiently in the mutant than in the control samples.

We thank the reviewer for this comment regarding Trp53-/- mice, which show a low penetrance of cleft palate. Regarding this question, we think Kdm6b as an epigenetic regulator acts on more downstream targets than just Trp53. In Trp53-/- mice, there might be other factors that could compensate for the lost function of Trp53. However, in Kdm6b mutant mice, this compensation cannot occur. We have also added this point in our Discussion section. Nutlin-3 treatment successfully rescued the hard palate cleft in all observed Wnt1Cre;Kdm6bfl/fl mice (N = 5). Two of these five showed a posterior soft palate cleft. We appreciate the reviewer’s comments on this detail and its improvement to the discussion in our manuscript.

We have added quantification for the Western blot results in Figure 4—figure supplement 1G as per the reviewer’s suggestion. Regarding why Nutlin-3 is more efficient in the mutant than in the controls, we cannot offer a specific explanation but note that this phenomenon is very common with Nutlin-3 treatment. For example, in Li et al.’s study, they also observed Nutlin-3 treatment is more efficient in knockout cells compared to controls (Li et al. 2016).

Reviewer #2 (Recommendations for the authors):

I have just a few comments that would not change the interpretation of the data nor would I request additional experiments.

1. Figure 2 - Figure Supplement 4, (A-B). Wnt1-Cre;Kdm6bfl/fl mice are compared to Wnt1-Cre;Kdm6bfl/+ , but why aren't they compared to a wild type control instead?

We thank the reviewer for this comment. We compared Wnt1Cre;Kdm6bfl/+;Rosa26-CAG-tdTomato to a wild type control and no differences were observed between these two. When we collected samples for this experiment, we compared the Kdm6b mutant mice to littermate controls to avoid developmental variation between different litters. Ratio to get Wnt1Cre;Kdm6bfl/+ is higher than wild type. The picture we used in the figure is of Wnt1Cre;Kdm6bfl/+;Rosa26-CAG-tdTomato.

2. Figure 2 - Figure Supplement 4, (E-H). The TUNEL stain data should probably be quantified by using TUNEL+ cell numbers collected across several different sections/samples. The difference between the control and the mutant, as seen in the 4 images provided, is large if expressed as a percentage but still limited to 3 cells or fewer (especially at day E13.5). Even at E14.5, while there are 3 TUNEL+ cells seen in the palatal shelves of the mutant and 0 in the control, multiple TUNEL signals can be seen just outside of the dotted region in the control but not in the mutant; either the control has more cell death right next to the palatal shelves, or the small difference between the two can be explained as a sampling error without more data points.

We appreciate the reviewer’s comment. A similar comment was also brought up by another reviewer. We have assessed cell senescence instead of cell death using primary palatal cell according to another reviewer’s suggestion. The data has been added to Figure 2-figure supplement.

3. L181-185, figure 2I-P; Similarly, can the authors state the sample size and quantify the RUNX2+ and SP7+ cells for easier comparison?

We thank the reviewer for this comment. Both RUNX2+ and SP7+ cells have been quantified. The results have been added to Figure 2.

4. L213-219, figure 3B-C; Can the authors elaborate more on the identification of p53 from the analysis of the RNA-seq? As "epithelial adherens junction signaling" in IPA also corroborate with "cadherin signaling pathway" in GO. Same for "regulation of Actin-based motility by Rho" in IPA and "cytoskeletal regulation by Rho GTPase" in GO. Is there any other highly confident targets that are worth exploring?

We appreciate the reviewer’s comment. In GO we did notice Wnt signaling pathway, which also has some crosstalk with Cadherin signaling, ranking as the top signaling pathway. We assessed expression of Axin2 in our sample using in situ RNAscope hybridization, but didn’t observe a difference between control and Kdm6b mutant mice. We agree with the reviewer that there are more signaling pathways that are worthy to investigate, but are beyond the scope of this study.

5. Figure 4 - Figure Supplement 5, (F). The p53 band intensity difference between the control and the mutant in the 10% DMSO group is slight; the loading control band (actin) also demonstrates a small difference in size that follows the same pattern. It would be better to see the p53 band difference quantified.

We thank the reviewer for this comment. We added a quantified result using Image J based on the integrated density of the bands of 3 individual samples. The data has been added to Figure 4-figure supplement 1G.

6. L323-334, figure 5F-N; In contrast to what the authors claim, Ezh1 seems to have a stronger signal than Ezh2. Besides, the expression pattern between Ezh1 and Ezh2 is quite different. Can the authors support the claim by using normalized quantification? Otherwise, can the authors elaborate why they chose to KO Ezh2 in the Wnt1-Cre; Kdm6bfl/fl model?

We appreciate the reviewer’s comment regarding Ezh1. Our in vivo staining of Ezh1 was done using RNAscope, while that of EZH2 was immunostaining. It’s hard to quantitatively compare these two signals directly based on the staining. However, we agree with the reviewer that the expression patterns of Ezh1 and Ezh2 are different. Expression of Ezh1 is more enriched in the oral side of the palatal shelf, while expression of Ezh2 is more universal. However, from the Western blot we could see at the protein level, the expression of EZH2 is higher than EZH1 when we load the same amount of total protein (based on a beta-actin internal control). Furthermore, conventional knockout Ezh1 mice don’t have any abnormal phenotypes (http://www.informatics.jax.org/marker/MGI:1097695), while Wn1Cre;Ezh2fl/fl mice show severe craniofacial defects. These results suggest that Ezh2 may have a more important role in regulating CNCCs.

7. Paper shows that inhibition of p53 by siRNA results in increased CNCC proliferation and DNA damage and reduces markers of osteogenesis, as seen in the siRNA treated cells. But is it possible to treat an embryo with a p53 inhibitor to directly demonstrate that loss of p53 in this model produces cleft palate? For example, Jones et al. treated pregnant mice with pifithrin-α to inhibit p53 (Nat. Med. 2011).

We thank the reviewer for this comment. We did generate Wnt1Cre;Trp53fl/fl mice, and only about 20% of the mutant mice showed a cleft palate phenotype. This ratio is similar to what has been previously reported for Trp53-/- mice. Regarding this question, we think Kdm6b as an epigenetic regulator affects more downstream targets than just Trp53. In Trp53-/- mice, there might be other factors that could compensate for the lost function of Trp53. However, in Kdm6b mutant mice, this compensation cannot occur. We have also added this point to our Discussion section.

8. The paper describes how the control cells were treated with Tfdp1 siRNA, which resulted in lower levels of p53. Why wasn't the same treatment applied to the Wnt1-Cre;Kdm6bfl/fl cells, especially if both the controls and the mutants were later treated with the Tfdp1 overexpression plasmid?

We thank the reviewer for this comment. The purpose of treating control cells with Tfdp1 siRNA is to test whether expression of Trp53 is regulated by the transcription factor TFDP1, which is validated by ATAC-seq and ChIP-qPCR results. Therefore, we used cells from the palatal mesenchyme of control mice to test the function of TFDP1 in regulating Trp53. Expression of Trp53 is already reduced in the Kdm6b mutant mice, so we think using cells from control mice is more appropriate for this experimental purpose.

Reviewer #3 (Recommendations for the authors):

[…]

– Nomenclature: protein names should be written in uppercase throughout the text and in the figures. In the current manuscript use of the nomenclature is not consistent. Often protein and gene names are not listed correctly. See: http://www.informatics.jax.org/mgihome/nomen/gene.shtml#ps

We appreciate the reviewers’ comments on this issue. All the protein and gene names have been corrected in the text and figures.

– Figure 1: In Wnt1-Cre;Kdm6bfl/fl embryos the tongue appears to be grossly dysmorphic and abnormally positioned (e.g. see panel Q). If this is only due to a technical artifact of the section, then a better image should be chosen for Figure 1Q. If the tongue is instead abnormally formed and enlarged, then this result should be discussed and well documented, given that withdrawal of the tongue from between the vertical palatal shelves is required for their seamless fusion and closure during embryonic development.

We really appreciate the reviewer’s comment and carefully went through our samples to examine the tongue morphology, including CT images which provide a better view of the size of the tongue (Figure 1 E-F are representative CT images). We didn’t observe any obvious size differences between control and mutant mice. Better images have been used for Figure 1Q.

– Figure 2 – Figure Suppl 4E-H: The Authors state that there is increased cell death in the palatal mesenchyme of Wnt1-Cre;Kdm6bfl/fl mice compared to controls, based on TUNEL staining, as shown in the figure. It is difficult to be convinced of this result, given that the TUNEL-positive cells are extremely sparse in both the control and mutant shown in these panels. If the Authors are convinced of their finding based on multiple experiments, they should show sections that better illustrate the defect, as well as quantify the numbers of TUNEL-positive cells over multiple sections for both control and mutant samples. In addition, they should include a graph comprising the total numbers of cells that were counted, the percentage of TUNEL-positive cells in both control and mutant, the fold increase of TUNEL-positive cells in the mutant, and the statistical significance. Unless these quantitative experiments are included, the Authors should definitely delete "increased cell death" for the mutant from the summary model shown in Figure 8.

However, if the Authors were to prove the presence of an increase of cell death in the palatal mesenchyme of Wnt1-Cre;Kdm6bfl/fl mice by conducting additional experiments, this finding would somehow contradict the result – which is very convincing – that p53 expression is decreased in mutant tissue (shown currently in Figure 7). Interestingly, in the discussion the authors mention (line 517): "In this study we notice that downregulated expression of p53 in Wnt1-Cre;Kdm6bfl/fl mice results in hyperproliferation and increased DNA damage in the proliferative cells, which might further lead to cell senescence." Assessing cell senescence, instead of cell death, in the mutant compared to the control could be revealing.

We really appreciate this suggestion. We have assessed cell senescence instead of cell death using primary palatal cell culture. Cells from Wnt1Cre;Kdm6bfl/fl mice showed increased cellular senescence compared to the cells from control mice. The data has been added to the Figure 2—figure supplement 1E-G.

– Figure 2I-L: The Authors state: "There was a decrease in the number of Runx2+ cells in the palatal mesenchyme at both E13.5 and E14.5 in Wnt1-184 Cre;Kdm6bfl/fl mice in comparison to the control" (Line 182-184). Either the Authors quantify the "number of cells" (see also comment above), or they could do a qRT-PCR to evaluate Runx2 mRNA in mutant versus control tissues.

We agree with the reviewer and have quantified both RUNX2+ and SP7+ cells in our samples. Quantification results have been added to Figure 2.

– Figure 3B: It would be beneficial to separate the genes that are upregulated from the genes that are downregulated in Wnt1-Cre;Kdm6bfl/fl mice versus controls. The overall message would be much clearer.

We thank the reviewer for this suggestion. We analyzed signaling pathways using only upregulated or downregulated genes according to this suggestion. However, after comparing the results, we think including both upregulated and downregulated genes provides better results, as both positive and negative regulators of signaling pathways are represented in the analysis.

– Figure 6: The rescue experiments are one of the main strengths of this study: beautiful and rigorous use of genetics to validate the pathways under analysis! However, it would add a great deal of strength to the rescue experiments to also evaluate the rescue at the cellular level (i.e. assess the rescue of proliferation and/or DNA damage, even if partial, or alternatively, examine whether expression of Sp7 or Runx2 is rescued by qRT-PCR).

We agree with the reviewer and appreciate this suggestion. We have assessed cell proliferation using EdU and RUNX2+ cells in our rescue model. The data has been added to Figure 7.

– Figure 7, panel I and Figure 7 – Figure Suppl 6: Red box that is described as highlighting the p53 promoter is placed instead over the promoter of a different gene, Wrap53. Indeed, this red box contains the TSS of Wrap53 and not p53. The Wrap53 gene encodes an essential component of the telomerase holoenzyme complex, a ribonucleoprotein complex required for telomere synthesis. The encoded protein interacts with other components of active telomerase and with small Cajal body RNAs (scaRNAs), which are involved in modifying splicing RNAs. It was reported that Wrap53 also functions as a p53 antisense transcript, which regulates endogenous p53 mRNA levels and the levels of P53 protein by targeting the 5' untranslated region of p53 mRNA (Mahmoudi et al. Mol Cell 2009). Therefore, Wrap53 can indirectly alter p53 levels of expression and regulation. siRNA knockdown of Wrap53 results in significant decrease in p53 mRNA and suppression of P53 induction upon DNA damage. Conversely, overexpression of Wrap53 increases p53 mRNA and protein levels (Mahmoudi et al. Molecular Cell 2009; Farnebo et al., Cell Cycle 2009; Saldana-Meyer et al., Genes and Development 2013). These studies unequivocally demonstrated that Wrap53 regulates p53. As it turns out, the primers used for the ChIP-qPCR experiment shown in Figure 7 were now blasted and they correspond to the Wrap53 promoter and not to the p53 promoter. Given all the knowledge discussed above, the current statement: "factor Tfdp1 binds to the promoter of p53 along with Kdm6b to specifically activate the expression of this tumor suppressor gene" should be re-evaluated and clarified. This point should be better investigated. As the distance between the Wrap53 and p53 genes is only approximately 1kb, in additional experiments the Authors should use primers specific to the p53 promoter and to the Wrap53 promoter together with chromatin sonicated to an average size of 300-600 bp, which should provide a clear answer as to whether TFDP1 binds directly to the p53 promoter, or to the Wrap53 promoter, or to both. If TFDP1 does not bind directly to p53 but to the Wrap53 promoter instead, regulation of p53 transcription would then be indirect, via Wrap53. This finding would still be of interest. This reviewer believes that this point is important and should be adequately clarified and examined in further depth.

We agree with the reviewer’s comment about Wrap53. We checked our RNAseq results, which show a very low expression level of Wrap53. To further confirm this result, we performed in situ RNAscope staining of our sample. The result is shown below. As our data clearly shows that Trp53 is widely expressed in palatal mesenchyme (Figure 3), it is very unlikely that Wrap53 is involved in the regulation of Trp53 during palatogenesis.

– Figure 7 – Figure Suppl 6: The Authors also analyze expression patterns of Kdm6b and its co-expression with Tfdp1 by scRNAseq. The scRNAseq data sets have already been described by the same Authors in a(nother) interesting paper that was recently published. There appears to be a striking enrichment of both Kdm6b and Tfdp1 genes in a particular cell subpopulation emerging from the scRNAseq datasets. It would be very interesting to know which cell type comprises this particular cluster? Also, the individual Panels in B should be enlarged to better visualize each cell clusters. It would also be very helpful to the reader to list the specific cell types that comprise each cluster in the Panels in B.

We thank the reviewer for this suggestion. We have extracted palatal mesenchyme clusters from published data and identified these clusters with enrichment of both Kdm6b and Tfdp1 expression. ScRNAseq analysis shows that cells co-expressing Kdm6b and Tfdp1 are enriched in 3 clusters (Pax9+, Aldh1a2+, and Twist1+ cells). This result has been added to Figure 8—figure supplement 1 and the size of the images has also been adjusted.

https://doi.org/10.7554/eLife.74595.sa2

Article and author information

Author details

  1. Tingwei Guo

    Center for Craniofacial Molecular Biology, Herman Ostrow School of Dentistry, University of Southern California, Los Angeles, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4089-1867
  2. Xia Han

    Center for Craniofacial Molecular Biology, Herman Ostrow School of Dentistry, University of Southern California, Los Angeles, United States
    Contribution
    Data curation, Formal analysis, Methodology, Validation, Writing - review and editing
    Competing interests
    No competing interests declared
  3. Jinzhi He

    Center for Craniofacial Molecular Biology, Herman Ostrow School of Dentistry, University of Southern California, Los Angeles, United States
    Contribution
    Data curation, Methodology, Validation, Writing - review and editing
    Competing interests
    No competing interests declared
  4. Jifan Feng

    Center for Craniofacial Molecular Biology, Herman Ostrow School of Dentistry, University of Southern California, Los Angeles, United States
    Contribution
    Formal analysis, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9944-2604
  5. Junjun Jing

    Center for Craniofacial Molecular Biology, Herman Ostrow School of Dentistry, University of Southern California, Los Angeles, United States
    Contribution
    Formal analysis, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5745-5207
  6. Eva Janečková

    Center for Craniofacial Molecular Biology, Herman Ostrow School of Dentistry, University of Southern California, Los Angeles, United States
    Contribution
    Writing - review and editing
    Competing interests
    No competing interests declared
  7. Jie Lei

    Center for Craniofacial Molecular Biology, Herman Ostrow School of Dentistry, University of Southern California, Los Angeles, United States
    Contribution
    Writing - review and editing
    Competing interests
    No competing interests declared
  8. Thach-Vu Ho

    Center for Craniofacial Molecular Biology, Herman Ostrow School of Dentistry, University of Southern California, Los Angeles, United States
    Contribution
    Data curation, Formal analysis
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6293-4739
  9. Jian Xu

    Center for Craniofacial Molecular Biology, Herman Ostrow School of Dentistry, University of Southern California, Los Angeles, United States
    Contribution
    Conceptualization, Writing - review and editing
    Competing interests
    Reviewing editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8162-889X
  10. Yang Chai

    Center for Craniofacial Molecular Biology, Herman Ostrow School of Dentistry, University of Southern California, Los Angeles, United States
    Contribution
    Conceptualization, Data curation, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing - original draft, Writing - review and editing
    For correspondence
    ychai@usc.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2477-7247

Funding

National Institutes of Health (R01 DE012711)

  • Yang Chai

National Institutes of Health (R01 DE022503)

  • Yang Chai

National Institutes of Health (U01 DE028729)

  • Yang Chai

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Bridget Samuels and Linda Hattemer for critical reading of the manuscript and also acknowledge USC Libraries Bioinformatics Service for their assistance with data analysis. We also thank the USC Office of Research and the Norris Medical Library for the bioinformatics software and computing resources.

Ethics

All mouse studies were conducted with protocols approved by the Department of Animal Resources and the Institutional Animal Care and Use Committee (IACUC) of the University of Southern California (Protocols 9320 and 20299).

Senior and Reviewing Editor

  1. Marianne E Bronner, California Institute of Technology, United States

Reviewer

  1. Eric Liao, Harvard Medical School, United States

Version history

  1. Received: October 9, 2021
  2. Preprint posted: October 14, 2021 (view preprint)
  3. Accepted: February 24, 2022
  4. Accepted Manuscript published: February 25, 2022 (version 1)
  5. Version of Record published: April 13, 2022 (version 2)

Copyright

© 2022, Guo et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,509
    Page views
  • 252
    Downloads
  • 6
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Tingwei Guo
  2. Xia Han
  3. Jinzhi He
  4. Jifan Feng
  5. Junjun Jing
  6. Eva Janečková
  7. Jie Lei
  8. Thach-Vu Ho
  9. Jian Xu
  10. Yang Chai
(2022)
KDM6B interacts with TFDP1 to activate P53 signaling in regulating mouse palatogenesis
eLife 11:e74595.
https://doi.org/10.7554/eLife.74595

Further reading

    1. Computational and Systems Biology
    2. Developmental Biology
    Mohamad Ibrahim Cheikh, Joel Tchoufag ... Konstantin Doubrovinski
    Research Article

    In order to understand morphogenesis, it is necessary to know the material properties or forces shaping the living tissue. In spite of this need, very few in vivo measurements are currently available. Here, using the early Drosophila embryo as a model, we describe a novel cantilever-based technique which allows for the simultaneous quantification of applied force and tissue displacement in a living embryo. By analyzing data from a series of experiments in which embryonic epithelium is subjected to developmentally relevant perturbations, we conclude that the response to applied force is adiabatic and is dominated by elastic forces and geometric constraints, or system size effects. Crucially, computational modeling of the experimental data indicated that the apical surface of the epithelium must be softer than the basal surface, a result which we confirmed experimentally. Further, we used the combination of experimental data and comprehensive computational model to estimate the elastic modulus of the apical surface and set a lower bound on the elastic modulus of the basal surface. More generally, our investigations revealed important general features that we believe should be more widely addressed when quantitatively modeling tissue mechanics in any system. Specifically, different compartments of the same cell can have very different mechanical properties; when they do, they can contribute differently to different mechanical stimuli and cannot be merely averaged together. Additionally, tissue geometry can play a substantial role in mechanical response, and cannot be neglected.

    1. Developmental Biology
    2. Evolutionary Biology
    Kwi Shan Seah, Vinodkumar Saranathan
    Research Article

    The study of color patterns in the animal integument is a fundamental question in biology, with many lepidopteran species being exemplary models in this endeavor due to their relative simplicity and elegance. While significant advances have been made in unraveling the cellular and molecular basis of lepidopteran pigmentary coloration, the morphogenesis of wing scale nanostructures involved in structural color production is not well understood. Contemporary research on this topic largely focuses on a few nymphalid model taxa (e.g., Bicyclus, Heliconius), despite an overwhelming diversity in the hierarchical nanostructural organization of lepidopteran wing scales. Here, we present a time-resolved, comparative developmental study of hierarchical scale nanostructures in Parides eurimedes and five other papilionid species. Our results uphold the putative conserved role of F-actin bundles in acting as spacers between developing ridges, as previously documented in several nymphalid species. Interestingly, while ridges are developing in P. eurimedes, plasma membrane manifests irregular mesh-like crossribs characteristic of Papilionidae, which delineate the accretion of cuticle into rows of planar disks in between ridges. Once the ridges have grown, disintegrating F-actin bundles appear to reorganize into a network that supports the invagination of plasma membrane underlying the disks, subsequently forming an extruded honeycomb lattice. Our results uncover a previously undocumented role for F-actin in the morphogenesis of complex wing scale nanostructures, likely specific to Papilionidae.