Osr2-Cre (Chen et al., 2009), and Mll4 ^fl/fl (Lee et al., 2013) lines were used for generating an Mll4 conditional knockout (cKO) model. ROSA ^mT/mG (hereafter referred to as mTmG) reporter line (Muzumdar et al., 2007) was used to label palate mesenchyme cells expressing eGFP by crossing with Osr2-Cre line. Mice were maintained by intercrossing or by crossing with C57BL/6J inbred mice. All animal procedures were approved by the Institutional Animal Care and Use Committee at the University at Buffalo.

The heads of mice aged 1, 3, 6, and 10 weeks were fixed in 10% formalin overnight and subsequently washed with PBS. Micro-CT scans were performed using a SCANCO µCT 100 (SCANCO Medical, Switzerland) (voxel size, 10 µm). The scanned images were calibrated to milligrams of hydroxyapatite per cubic centimeter (mgHA/cm³) for accurate density measurement. Each image was manually adjusted for rotation and alignment along each axis using the FIJI software (Schindelin et al., 2012). To eliminate background noise, areas with measurements below 200.03 mgHA/cm³ were removed. Each image was then cropped to focus on the specific region of interest within the maxillary jaw. The Fijiyama plugin was used to manually register the images, aligning them for further analysis (Fernandez and Moisy, 2021). Finally, the 3Dscript plugin was used for 3-dimensional (3D) rendering (Schmid et al., 2019).

Mice were euthanized at predetermined stages. Head tissue specimens were fixed in 4% paraformaldehyde (PFA), dehydrated through an ethanol series, embedded in paraffin, and sectioned at 5–7 μm thickness, as described previously (Lee et al., 2022). Specimens harvested at E18.5, postnatal day (P) 0.5, and 1, 3, 6, and 10 weeks were decalcified in 14% ethylenediaminetetraacetic acid (EDTA) for 1, 3, 6, and 10 weeks, respectively. For histological analysis, sections were stained with Azan’s trichrome staining, as previously described (Lee et al., 2022), and were imaged on an Axioscope compound upright light microscope (for 20X and 40X magnifications) (Zeiss, Germany). An SZX16 stereomicroscope (Olympus, Japan) was used for 5X magnification images, as well as gross images. For immunofluorescence staining, antigen retrieval was performed in an antigen retrieval buffer, using a pressure cooker for 15 min at low pressure, followed by washing in PBS, blocking at room temperature in a blocking buffer (2% goat serum, 5% BSA, 1% Triton-X 100, 0.1% Tween-20 in PBS) for 1 h, and incubation with primary antibodies, diluted in blocking buffer, overnight at 4°C. On the second day, the sections were washed in PBS, incubated with secondary antibodies, diluted in blocking buffer, at room temperature for 1 h, washed in PBS, and mounted with Vectashield mounting media with DAPI (Vector Laboratories, USA; H-1200). The primary antibodies used were: homemade guinea pig anti- MLL4 (1:300) (Huisman et al., 2021), rabbit monoclonal anti-SOX9 (1:300; Abcam, United Kingdom; ab185230), mouse monoclonal anti-COL2A1 (1:50, Santa Cruz, USA; sc-52658), and rabbit monoclonal anti-RUNX2 (1:300; Cell Signaling, USA; #12556). Immunofluorescence images were acquired using a THUNDER Imager fluorescence microscope (Leica Microsystems, Germany). To eliminate non-specific signals from auto-fluorescence, we acquired additional images usng a non-stained CFP (cyan) filter and subtracted these signals following an established scientific method (Van de Lest et al., 1995).

The DeadEnd™ Fluorometric TUNEL System (Promega, USA; G3250) was used to assess apoptosis in paraffin sections, according to the manufacturer’s protocol.

Mouse heads, harvested at predetermined stages, were fixed in 100% ethanol for 3 days, stained in Alcian Blue solution (30 mg Alcian Blue dissolved in 20 mL glacial acetic acid plus 80 mL 95% ethanol) for 2 days, re-fixed in 100% ethanol for 24 h, cleared in 2% KOH for 3 h, and counterstained in Alizarin Red solution (50 mg Alizarin Red dissolved in 1 L of 1% KOH) for 24 h, followed by tissue clearing in 1% KOH/20% glycerol for 3 days. Images were acquired using a standard stereomicroscope (Amscope, United States).

Total RNA was extracted from micro-dissected midpalatal suture tissue samples using the RNeasy Micro Plus Kit (Qiagen, Germany; 74034). Four biological replicates per genotype (control and Mll4-cKO), were used for sequencing, with one sample in each genotype pooled from three individuals. Library construction and high-throughput sequencing were performed by Novogene company (China) using the Illumina NovaSeq 6000 platform. Sequencing reads were analyzed using Salmon (Patro et al., 2017) to quantify gene expression at the transcript level. The resulting outputs from Salmon were analyzed using DESeq2 (Love et al., 2014) to obtain gene-level expression data and identify differentially expressed genes (DEGs) between Mll4-cKO and littermate control samples. DEGs were defined by a log2 fold change (log2FC) ≥ 0.4 and an adjusted p-value ≤0.001. Volcano plots were generated using the EnhancedVolcano package (Blighe et al., 2024), and gene ontology (GO) analysis was performed with clusterProfiler (Wu et al., 2021). Raw expression data results are included in the Supplementary Material. The original RNA sequencing (RNA-seq) data files have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus (NCBI GEO) database under the accession number PRJNA1148613.

The ChIP-sequencing (ChIP-seq) datasets for UTX in cranial neural crest cells (GSE103849) were re-analyzed using a previously described data processing workflow (Shpargel et al., 2017). Briefly, sequencing reads were aligned to the mouse mm10 reference genome using Bowtie2 (Langmead and Salzberg, 2012) and converted to BAM files using SAMtools (Li et al., 2009). Reads mapping to the X and Y chromosomes were excluded. Filtered BAM files from biological replicates were merged and subjected to peak calling with MACS (Zhang et al., 2008). Peaks were annotated using the ChIPseeker R package (Yu et al., 2015). For visualization in Integrative Genomics Viewer (IGV), the filtered BAM files were further processed using BEDTools (Quinlan and Hall, 2010) and converted to BigWig format using the UCSC command-line tool bedGraphToBigWig (Kent et al., 2002). The intersection of Mll4-cKO DEGs with UTX target genes, along with functional enrichment analysis, was performed in R using clusterProfiler (Wu et al., 2021).

All statistical analyses were performed using Prism 10 software (GraphPad Software, United States). Pairwise differential expression was analyzed using Student’s t-test. Statistical significance was defined as p-values ≤0.05 (*), ≤0.01 (**), and ≤0.001 (***).

For the purpose of investigating the role of MLL4 specifically during palatogenesis, we used the Cre/loxP system driven by Osr2-Cre. Osr2-Cre is first activated at E11, when palatogenesis is initiated, and remains active in the palate mesenchyme cells (Lan et al., 2001; Lan et al., 2004). The Osr2-Cre;mTmG line allows us to confirm the activation of Osr2-Cre by detecting GFP signals using fluorescence microscopy. When the bilateral palatal shelves are actively undergoing morphogenesis, for example, at E13.5, Osr2-Cre was activated in the developing palatal shelves, which occupies the center of the developing face (Figures 1A, B). Thus, we were able to generate a novel mouse model with palatogenesis-specific deletion of Mll4 in the palate mesenchyme cells (Mll4 ^fl/fl ;Osr2-Cre, or Mll4-cKO). Surprisingly, compared to the previously reported neural crest-specific Mll4 deletion (Mll4 ^fl/fl ;Wnt1-Cre and Mll4 ^fl/fl ;Sox10-Cre) mice, which exhibited cleft palate and neonatal lethality (Shpargel et al., 2020), our palatogenesis-specific Mll4-cKO mice did not show cleft palate or lethality at birth and could grow up into adult mice that did not show much noticeable changes when grossly examined for exterior appearance at 10 weeks (Figures 1C, G). However, close examination of the orofacial skeleton dimensions by using micro-CT analysis revealed that the facial angle, length, and width in the maxilla showed mild but significant decreases in the Mll4-cKO mice (Figures 1D–F, H–J, N), which were signs of mild midfacial hypoplasia. Additional examination of the intra-oral aspect of the palate revealed several minor differences between the control and Mll4-cKO mice (Figures 1K, L). Compared to controls, the midpalatal suture appeared more widened in the Mll4-cKO palate (Figures 1K′, L′, arrowheads), with minor changes observed in the rugae patterning (Figures 1K′, L′, arrows). To test the possibility that these mild midfacial hypoplasia and palate phenotypes that we observed in the Mll4-cKO mice may have resulted from delay in their overall development and growth, we measured the body weight both at 10 weeks and at birth but did not find any significant difference (Figure 1M). We also compared the body weight between male and female mice at these two stages but found no significant difference (data not shown). Our initial observations suggested that, separate from its early role in neural crest cell differentiation, MLL4 may have other unknown functions in postnatal palate development and growth.

Thus, to gain better understanding of the midfacial and palate phenotypes observed from gross examination of the Mll4-cKO mice (Figure 1), we further examined the palate skeleton at 10 weeks using our micro-CT results. By examining the virtual 3D models that were created by reconstructing the planar X-ray images, we found that the overall palate width showed narrowing in the Mll4-cKO mice, which was consistent with the decrease in the facial width that we observed earlier. Anatomical landmarks for assessing the palate width, such as the incisive foramens, palatal processes of maxilla, and palatal processes of palatine, all showed constricted appearances in the virtual 3D model of the Mll4-cKO mice (Figures 2A, B). Using virtual slicing of the 3D models, we were able to further generate digital coronal/frontal cross sections of the palate. These virtual cross sections enabled us to perform comprehensive quantitative analysis of the overall horizontal dimension of the palate along its anteroposterior axis by precisely measuring the palate width at four representative section levels: the incisive foramen (in slice 1), the palatal processes of maxilla (in slice 2), the transpalatal suture (in slice 3), and the palatal processes of palatine (in slice 4) (Figures 2C, D). At all four section levels along the anteroposterior axis, the palate width showed significant decrease in the Mll4-cKO mice, compared to the control mice (Figures 2E1–E4). These results suggested that MLL4 plays an important role in transverse growth of the palate during postnatal palatogenesis.

Thus, to address the question of whether MLL4 is actually required for proper growth of the palate during postnatal palatogenesis, we conducted a time-course quantitative analysis of the horizontal dimension of the palate skeleton using micro-CT-generated virtual coronal/frontal cross sections at 1, 3, and 6 weeks. For this approach, we used the same four representative coronal/frontal cross section levels along the anteroposterior axis of the palate, which we used at 10 weeks. At 1 week, we found that palate growth was not significantly changed in the Mll4-cKO mice at the incisive foramen level (Figure 2E1). In contrast, at the rest of the levels, the first (M1), second (M2), and third molar (M3) levels, palate growth showed minor, but significant decrease (Figures 2E2–E4). Later, at 3, 6, and 10 weeks, palate growth showed significant decrease at all four levels along the anteroposterior axis in the Mll4-cKO mice, compared to the control mice (Figures 2E1–E4). These results suggested that MLL4 is indeed required for proper transverse growth of the palate during postnatal palatogenesis.

We next asked the question of how MLL4 promotes palate growth. Based on our finding that the Mll4-cKO mice showed significant narrowing of the palate during postnatal palatogenesis (Figure 2), we reasoned that the key biological events for palate growth that depend on MLL4 function may occur at younger ages, possibly as early as, or even prior to, birth. Thus, to identify the primary developmental mechanism underlying the role of MLL4 in palate growth, we examined the palate anatomy in the control and Mll4-cKO mice around birth, searching for the earliest signs of disruption in the Mll4-cKO palate. By first using skeletal staining of the developing cranium and palate tissues, we found that the overall skeletal pattern was similar between the control and Mll4-cKO mice (Figures 3A, B). However, where we had initially observed midline defects in the rugae pattern and a widened midpalatal suture in the Mll4-cKO mice (Figure 1), we found that the width of the midpalatal suture was significantly wider in the Mll4-cKO mice, compared to the controls, prior to birth (at E18.5) (Figures 3A, B, G). In parallel, histological examination of the palate, by using trichrome staining of the coronal/frontal paraffin sections, confirmed that the width of the midpalatal suture, situated between the bilateral bony palatal processes, was significantly increased in the Mll4-cKO mice, compared to the control mice (Figures 3C, D, C′, D′, G). To determine the cellular mechanisms underlying the wider midpalatal suture in the Mll4-cKO mice, we performed both cell proliferation, using EdU incorporation assays, and cell apoptosis, using TUNEL staining. While rarely any cells in the midpalatal suture were EdU-positive for cell proliferation in both the control and Mll4-cKO mice (data not shown), TUNEL-positive cell apoptosis showed minor but significant increase in the Mll4-cKO midpalatal suture, compared to controls, at E18.5 (Figures 3E, F, E′, F′, H). Our findings that the earliest signs of disrupted biological events were specifically discovered in the midpalatal suture of the perinatal Mll4-cKO palate suggested that MLL4 is essential for the establishment of the midpalatal suture around birth.

To gain insight into the molecular mechanism underlying the histological and cellular alterations observed in the perinatal Mll4-cKO palate tissue, we micro-dissected the midpalatal suture tissues and performed RNA-seq analysis on control and Mll4-cKO mice at birth (P0.5). Our results identified a total of 117 DEGs between the control and Mll4-cKO midpalatal suture tissues, with 40 genes upregulated and 77 genes downregulated in the Mll4-cKO group (adjusted p-value ≤0.001, log2FC ≥ 0.4) (Figure 4A). Notably, among the DEGs were genes associated with cartilage development, or chondrogenesis, including extracellular matrix genes Col2a1, Col9a1, Col9a2, Acan (Aggrecan), and Comp (Cartilage oligomeric matrix protein), as well as Mia/Cdrap1 (Melanoma inhibitory activity) and Snorc (Small novel rich in cartilage). GO analysis identified the top 10 most enriched biological processes, along with a list of downregulated genes associated with each process (Figures 4B–D). This analysis suggested that cartilage development, endochondral bone morphogenesis, and chondrocyte differentiation are among the highly enriched biological processes potentially regulated by MLL4.

In parallel, we selected several representative markers of osteochondral progenitor cell differentiation during chondrogenesis and endochondral ossification for immunofluorescence staining on paraffin sections of the perinatal palate (at P0.5) in control and Mll4-cKO mice. Among these markers, COL2A1 was selected from the DEG list (Figure 4) to validate our RNA-seq results. Additionally, RUNX2 and SOX9 were chosen as critical transcription factors regulating osteochondral progenitor cell differentiation: RUNX2 marks osteoblast differentiation, while SOX9 marks chondrocyte differentiation, providing insights into the developmental processes occurring in the midpalatal suture. We also verified the expression of MLL4 in the perinatal Mll4-cKO palate using immunofluorescence staining. Our results revealed that the expression of COL2A1 and SOX9, both early markers of chondroblast differentiation, was markedly reduced in the Mll4-cKO midpalatal suture (Figure 5B1, D2, E), compared to controls (Figures 5A1, C2, E). In contrast, the expression of RUNX2, an osteogenic differentiation marker, was similar between the control and Mll4-cKO midpalatal suture (Figure 5A2, B2, E). These findings suggest that the primary molecular events dependent on MLL4 function in the perinatal midpalatal suture are associated with chondrogenic cell differentiation, cartilage development, and potentially endochondral ossification. Additionally, the markedly reduced intensity of MLL4 in the perinatal Mll4-cKO palate confirms the effective deletion of the Mll4 gene in the Mll4-cKO palate (Figures 5C1, D1).

Finally, to verify whether the biological processes that were identified from our RNA-seq analysis were actually impacted as a result of Mll4 deficiency during postnatal palatogenesis, we performed histological examination of the midpalatal suture at birth and beyond (up to 1 week), using trichrome staining of the paraffin sections (Figures 6A–D). Interestingly, after birth (at P0.75), a newly formed layer of cartilage tissue capped the medial surface of each of the bilateral palatal processes, flanking the midpalatal suture in the control mice (Figure 6A, arrowheads). In contrast, the newborn Mll4-cKO mice lacked these midpalatal suture-flanking cartilage tissue layers on the medial surfaces of the bilateral palatal processes (Figure 6B). By histologically monitoring the postnatal development of the midpalatal suture, we found that, at 1 week in the control mice, the midpalatal suture cartilage layers showed uniform enlargement of the consisting chondrocytes (Figure 6C, arrowheads). In contrast, the 1-week old Mll4-cKO mice just started to develop the midpalatal suture cartilage layers, which were only partially formed (Figure 6D, arrowheads). Our findings suggested that this defective phenotype observed in the Mll4-cKO midpalatal suture cartilage was due to significant retardation of chondrogenic differentiation as a result of Mll4 deficiency in the developing palate. Thus, we further performed additional time-course examination of the midpalatal suture development at 3, 6, and 10 weeks. Interestingly, at all three stages (3, 6, and 10 weeks), the bilateral cartilage layers in both the control and Mll4-cKO midpalatal sutures were shown to be adhered to each other in the midline (Figures 6E–J). However, the interface between the bilateral cartilage layers was less straight in the Mll4-cKO suture, compared to the control suture (Figures 6E–J). Also at all three stages, the overall width of the midpalatal suture cartilage was markedly increased in the Mll4-cKO mice (brackets in Figures 6F, H, J) than in the control mice (brackets in Figures 6E, G, I). Importantly, compared to the horizontally organized chondrocyte “columns” in the control group, the Mll4-cKO suture chondrocytes appeared disorganized and were under-differentiated, especially at 10 weeks (Figures 6I, J, I′, J′). Taken together, these results suggest that Mll4 deficiency results in retardation in the developmental of the midpalatal suture during postnatal palatogenesis. And by integrating our findings in the developing midpalatal suture (Figure 6) and in the growing palate skeleton (Figure 2), our data suggest that Mll4 is associated with midpalatal suture development and palate growth during postnatal palatogenesis.

In this study, we generated for the first time a unique mouse model demonstrating that MLL4 plays an essential role in determining palate width (Figure 1). By using Osr2-Cre to delete Mll4 tissue-specifically in the palate mesenchyme from the onset of palatogenesis (at E11) (Lan et al., 2001; Lan et al., 2004), we examined the intrinsic role of MLL4 during the active process of palatogenesis (Figure 2). Unlike previously reported neural crest-specific Mll4 deletion models driven by Wnt1-Cre or Sox10-Cre, which exhibit cleft palate and neonatal lethality (Shpargel et al., 2020), our Mll4-cKO mice did not show cleft palate or neonatal lethality. This allowed us to extend our examinations beyond prenatal stages. Importantly, the narrow palate phenotype observed in our Mll4-cKO mice closely resembles a common craniofacial phenotype of Kabuki syndrome: the high-arched, narrow palate, reported in 72% (64/89) of individuals with Kabuki syndrome (Matsumoto and Niikawa, 2003; Porntaveetus et al., 2018). This prevalence is twice as high as the 35% (68/196) frequency of cleft lip/palate individuals in Kabuki syndrome (Matsumoto and Niikawa, 2003), which was studied using the aforementioned neural crest-specific Mll4 deletion models (Shpargel et al., 2020). Our findings further emphasize the clinical relevance of the narrow palate phenotype, especially when accompanied by midfacial hypoplasia, which poses potential risks for various dental and orofacial complications, such as crowded or impacted teeth, breathing difficulties, and further midfacial underdevelopment. Taken together, our unique Mll4-cKO model not only highlights a significant craniofacial problem in Kabuki syndrome but also sheds light on an understudied yet critical developmental process that warrants urgent attention. These results underscore the multifaceted role of MLL4 in craniofacial development and disease.

Here, we establish MLL4 as a crucial factor in midpalatal suture development, as evidenced by the significant developmental delays observed in the Mll4-cKO mice beginning in the perinatal stages (Figure 3) and persisting throughout postnatal palatogenesis into adulthood (Figure 6). Whole gene expression profiling of midpalatal suture tissues revealed that Mll4 is required for the timely expression of a large number of genes related to chondrogenesis, including Col2a1 (Bell et al., 1997; Lefebvre et al., 1997; Ng et al., 1997), Col9a1 (Genzer and Bridgewater, 2007; Zhang et al., 2003; Zhou et al., 1998), Col9a2 (Bernard et al., 2003), Acan (Sekiya et al., 2000; Han and Lefebvre, 2008), Comp (Liu et al., 2007), Mia/Cdrap1 (Xie et al., 1999), and Snorc (Jaiswal et al., 2020) (Figure 4). Interestingly, Sox9 did not appear in our DEG list, possibly due to 1) variability introduced during tissue micro-dissection, given its small size of the midpalatal suture and the potential for contamination from adjacent tissues, or 2) the relatively low steady-state expression level of Sox9 compared to other chondrogenic genes (e.g., RPKM values: Sox9 = 37.6 vs. Col2a1 = 787.8 and Acan = 308.2), which may obscure detectable differences. Nevertheless, immunofluorescence analysis confirmed that SOX9 expression was markedly downregulated in the mesenchyme cells in the Mll4-cKO midpalatal suture (Figure 5), suggesting that MLL4 may indeed regulate Sox9 expression.

Notably, the seven downregulated chondrogenic genes in the Mll4-cKO midpalatal suture mesenchyme are all known SOX9 targets, highlighting its role as a master regulator of chondrogenesis (Lefebvre et al., 2019). Recent studies suggest that SOX9 functions as a pioneer factor, recruiting epigenetic co-factors such as MLL4 to epigenetically prime and remodel chromatin to a transcriptionally accessible state during cell fate determination (Yang et al., 2023). Therefore, it is reasonable to speculate that Mll4 deficiency may affect midpalatal suture development through two mechanisms: 1) lack of MLL4 as a transcriptional co-activator of SOX9 and 2) downregulation of SOX9 itself. Both mechanisms would impair SOX9’s ability to activate downstream chondrogenic genes, as observed in the Mll4-cKO midpalatal suture. Furthermore, the known auto-regulatory function of SOX9 in chondrocytes (Mead et al., 2013) could exacerbate the chondrogenesis defects in our model.

To further explore the potential mechanism by which MLL4 regulates chondrogenic gene expression, we re-analyzed published UTX ChIP-seq datasets in O9-1 cranial neural crest cells (Shpargel et al., 2017), as MLL4 is known to coordinate with UTX to regulate target gene expression (Cho et al., 2007; Froimchuk et al., 2017). Integrating these datasets with our Mll4-cKO DEGs (Figure 4), we found that approximately 56% (66/117) of the DEGs overlapped with UTX target genes (Figure 7A), implicating them as potential direct transcriptional targets of MLL4. Functional classification of these genes revealed significant enrichment for skeletal system-related biological processes, including cartilage development (Figure 7B). Intriguingly, these potential MLL4 direct target genes included Comp and Snorc, as well as other genes associated with cartilage development and endochondral bone morphogenesis, such as Matn3, Prkg2, Scx, and Trpv4 (Figure 7C; Supplementary Figure S1). Although Sox9 did not appear in our DEG list, immunofluorescence results clearly indicated reduced SOX9 expression in the Mll4-cKO samples (Figure 5). Incorporating Sox9 as a “potential” DEG in our analysis revealed that it is targeted by UTX (Figures 7A, C), strongly suggesting that Sox9 is a key MLL4 target. Together, these results indicate that MLL4 activates a subset of chondrogenic genes, potentially through the regulation of Sox9, in collaboration with UTX. This proposed molecular mechanism is summarized schematically in Figure 7D.

Another factor contributing to gene expression changes in Mll4-deficient midpalaltal sutures is increased cell death. While Mll4-cKO mice exhibited a statistically significant increase in TUNEL-positive cells, the percentages of TUNEL-positive nuclei in both control and Mll4-cKO mice were relatively low (Figure 3), indicating that apoptosis is rare in this context. Thus, the widened midpalatal suture phenotype in Mll4-cKO mice is likely attributed to deficient development due to impaired chondrogenic gene expression (Figure 4). Analysis of DEGs for overlap with the Hallmark Apoptosis gene set (Liberzon et al., 2015) identified two downregulated genes, Igfbp6 and Lgals3, but no upregulated genes. SOX9, which is known to regulate apoptosis in neural crest cells (Cheung et al., 2005), may contribute to the observed increase in TUNEL-positive cells (Figures 3E–G). However, given the low percentage of apoptotic cells, increased apoptotic cells likely reflects an indirect consequence of developmental retardation in the midpalatal suture, rather than being the primary cause of the phenotype.

The chondrogenic retardation observed in our model aligns with previously reported defects in the cartilaginous primordia of the cranial base in neural crest-specific Mll4 deletion mice and in the growth plates of long bones in heterozygous Mll4 loss-of-function mice (Fahrner et al., 2019; Shpargel et al., 2020). Also, common downregulated genes were identified when comparing our DEG list with an RNA-seq dataset from earlier palatal shelves (at E14.25) in the neural crest-specific Mll4 deletion mice (Shpargel et al., 2020), which include Acan, Col2a1, Col9a1, Col9a2, Col9a3, and Mia. These findings suggest a potential consensus regulatory mechanism downstream of MLL4 in different states of cell differentiation at different stages of palatogenesis. Among these genes, Acan and Col2a1 have been identified as causal genes for cleft palate (Xu et al., 2021), highlighting a shared genetic basis for two distinct phenotypes—narrow palate and cleft palate—manifesting under different molecular contexts at their onset. This is consistent with previous studies showing that high-arched, narrow palate and cleft palate may share common genetic causes (Conley et al., 2016; Tabler et al., 2013). However, it remains unclear why bilateral palatal shelves fail to fuse only in the neural crest-specific Mll4 deletion mice (resulting in cleft palate) but not in our palate mesenchyme-specific Mll4-cKO mice. This may reflect an earlier requirement for MLL4 during the migration and differentiation of neural crest cells prior to their incorporation into the palate mesenchyme, potentially explaining the cleft palate phenotype in neural crest-specific Mll4 mutants.

For future studies, it will be important to investigate whether the H3K4 methyltransferase activity of MLL4 is specifically required for its role in postnatal palatogenesis. Recent findings (Xie et al., 2023) suggest that MLL4 has both H3K4 methyltransferase-dependent and -independent roles in development and differentiation. Notably, while H3K4 methylation appears largely dispensable for enhancer activation during embryonic stem cell differentiation, it is crucial for lineage-selective processes in early embryonic development. Given that MLL4 predominantly mediates the deposition of H3K4 mono- and di-methylation (H3K4me1/2) (Lee et al., 2013; Ang et al., 2016; Wang et al., 2017), its enhancer-related functions may be particularly relevant during postnatal palatogenesis. Exploring these distinct mechanisms in the context of enhancer priming and tissue-specific development may provide valuable insights into the regulatory functions of MLL4.

The midpalatal suture has long been regarded as a critical skeletal structure that is anatomically situated at the very center of the craniofacial complex, playing a significant role in the development and growth of the palate and the maxillary jaw (Latham, 1971; Li et al., 2015; Rice, 2008). Therefore, it is intriguing to discover that the two major phenotypes observed in the Mll4-cKO mice were a growth deficit in the palate and developmental retardation of the midpalatal suture. While the causal relationship between midpalatal suture developmental retardation and lateral growth deficit remains to be tested in future studies, our findings invite the possibility that MLL4 regulates growth of the palate and the midface by modulating midpalatal suture. This highlights a functional link between midpalatal suture development and palate growth and supports the potential central role of the midpalatal suture as a growth site driving craniofacial development.

In addition to serving as a growth site, the midpalatal suture functions as a mechanosensitive structure that responds to environmental forces, such as suckling, mastication, and skeletal growth (Roth et al., 2022; Li et al., 2016). For example, in orthodontic dentistry, rapid palatal expansion procedures apply mechanical force to the maxillary jaw, widening the midpalatal suture and inducing tensile forces that promote new bone formation, thereby increasing maxillary width. These responses have been partially attributed to Piezo2-positive chondrogenic mesenchyme cells in the midpalatal suture (Gao et al., 2022). Therefore, it would be interesting to test whether Mll4 deficiency impacts the mechanosensory properties of the midpalatal suture, possibly through Piezo2-positive chondrogenic mesenchyme cells. This can be explored by simulating mechanical stress, such as performing a palatal expansion procedure in Mll4-cKO mice (Hou et al., 2007; Katebi et al., 2012). Such studies would provide valuable insights into the role of MLL4 in craniofacial development and the biological responses of the midpalatal suture to mechanical cues.