Deficiency of Fam20b-Catalyzed Glycosaminoglycan Chain Synthesis in Neural Crest Leads to Cleft Palate

Cleft palate is one of the most common birth defects. Previous studies revealed that multiple factors, including impaired intracellular or intercellular signals, and incoordination of oral organs led to cleft palate, but were little concerned about the contribution of the extracellular matrix (ECM) during palatogenesis. Proteoglycans (PGs) are one of the important macromolecules in the ECM. They exert biological functions through one or more glycosaminoglycan (GAG) chains attached to core proteins. The family with sequence similarity 20 member b (Fam20b) are newly identified kinase-phosphorylating xylose residues that promote the correct assembly of the tetrasaccharide linkage region by creating a premise for GAG chain elongation. In this study, we explored the function of GAG chains in palate development through Wnt1-Cre; Fam20bf/f mice, which exhibited complete cleft palate, malformed tongue, and micrognathia. In contrast, Osr2-Cre; Fam20bf/f mice, in which Fam20b was deleted only in palatal mesenchyme, showed no abnormality, suggesting that failed palatal elevation in Wnt1-Cre; Fam20bf/f mice was secondary to micrognathia. In addition, the reduced GAG chains promoted the apoptosis of palatal cells, primarily resulting in reduced cell density and decreased palatal volume. The suppressed BMP signaling and reduced mineralization indicated an impaired osteogenesis of palatine, which could be rescued partially by constitutively active Bmpr1a. Together, our study highlighted the key role of GAG chains in palate morphogenesis.


Introduction
Cleft palate is the most common congenital malformations caused by genetic mutations or environmental stimulus [1,2]. The mammalian palate is composed of cells derived from cranial neural crest cells (CNCCs) and epithelial cells derived from pharyngeal ectoderm [3]. In mice, the palatal process grew vertically on either side of the tongue in the common oralnasal cavity at early stage. Then, the palatal shelves are re-oriented horizontally above the tongue by overcoming the tongue obstruction, and eventually contacting with each other along the midline to fuse into a complete palatal floor covering the top of the oral cavity [3]. Cleft palate is caused by intrinsic factors, such as elevation or fusion abnormalities, or by disharmony in oral structure, such as physical obstruction of the tongue. There has been a significant amount of evidence of impaired intercellular or intracellular signals and coordination of oral structures leading to cleft palate [4][5][6]. Inactivation of Pax9, Gli2, or Osr2 in palatal mesenchyme impairs palatal elevation, while the loss of Msx1 and Lhx8, and the conditional ablation of Tgfbr2 in palatal neural crest suppresses palatal growth [7][8][9][10][11]. Moreover, the mice with homozygous null for Irf6 developed cleft palate via inappropriate oral adhesions and palatal fusion [12,13]. In contrast, little is known about the contribution of the extracellular matrix (ECM) to palate development. Thus, the role of the ECM Int. J. Mol. Sci. 2023, 24, 9634 3 of 16 both anterior and posterior palatal mesenchyme ( Figure 1A,B), while at E13.5, it was concentrated in the center of the anterior palate and the mesenchyme underlying the epithelium in the posterior palate ( Figure 1C,D). At E15.5, the Fam20b expression domain was restricted to the epithelium and ossifying center ( Figure 1E,F).
anterior and posterior palatal mesenchyme ( Figure 1A,B), while at E13.5, it was concentrated in the center of the anterior palate and the mesenchyme underlying the epithelium in the posterior palate ( Figure 1C,D). At E15.5, the Fam20b expression domain was restricted to the epithelium and ossifying center ( Figure 1E,F).
To explore the effect of Fam20b inactivation on GAG chain content, Alcian blue staining was performed. We found that the intensity of Alcian blue staining were apparently reduced in E14.5 Wnt1-Cre; Fam20b f/f palatal shelves, tongue, nasal cartilage, and Meckel's cartilage comparing with wild-type littermates ( Figure 1G-J), suggesting decreased GAG chain accumulation in CNCC-derived organs. More importantly, Wnt1-Cre; Fam20b f/f mice exhibited complete cleft palate after E14.5 ( Figure 1G-J). The lateral palatal processes of Wnt1-Cre; Fam20b f/f were significantly smaller than those of the wild-type littermates, and failed to elevate horizontally at E14.5 ( Figure 1G-J).  To explore the effect of Fam20b inactivation on GAG chain content, Alcian blue staining was performed. We found that the intensity of Alcian blue staining were apparently reduced in E14.5 Wnt1-Cre; Fam20b f/f palatal shelves, tongue, nasal cartilage, and Meckel's cartilage comparing with wild-type littermates ( Figure 1G-J), suggesting decreased GAG chain accumulation in CNCC-derived organs. More importantly, Wnt1-Cre; Fam20b f/f mice exhibited complete cleft palate after E14.5 ( Figure 1G-J). The lateral palatal processes of Wnt1-Cre; Fam20b f/f were significantly smaller than those of the wild-type littermates, and failed to elevate horizontally at E14.5 ( Figure 1G-J).

Wnt1-Cre; Fam20b f/f Palatal Shelves Were Capable of Elevating
To examine if the complete cleft palate in Wnt1-Cre; Fam20b f/f mice resulted from the primary defects in palatal mesenchyme, we performed organ culture ex vivo to examine the elevating capacity of Wnt1-Cre; Fam20b f/f palatal shelves. In the E13.0 Wnt1-Cre; Fam20b f/f mouse heads without tongue and mandible, both the anterior and posterior palatal shelves were able to elevate in the ex vivo organ culture (Figure 2A-D). Moreover, the E14.0 Wnt1-Cre; Fam20b f/f mouse palatal shelves fused with each other, as the WT controls did in the ex vivo organ culture ( Figure 2E,F), suggesting that Wnt1-Cre; Fam20b f/f palatal shelves also retained the ability to fuse. Therefore, the cleft palate or the failed palatal elevation in Wnt1-Cre; Fam20b f/f mice was supposed to be a secondary deformity. We further generated Osr2-Cre; Fam20b f/f mice to examine this conclusion. Worthy of noticing, Osr2-Cre; Fam20b f/f mice showed intact palates, as well as little-affected tongues and mandibles compared with the wild-type counterparts ( Figure 2G-L). Taken together, these results demonstrated that Fam20b inactivation caused few primary defects in palatal elevation, and the failed palatal elevation in Wnt1-Cre; Fam20b f/f mice most likely resulted from the undescending tongue.

Wnt1-Cre; Fam20b f/f Palatal Shelves Were Capable of Elevating
To examine if the complete cleft palate in Wnt1-Cre; Fam20b f/f mice resulted from the primary defects in palatal mesenchyme, we performed organ culture ex vivo to examine the elevating capacity of Wnt1-Cre; Fam20b f/f palatal shelves. In the E13.0 Wnt1-Cre; Fam20b f/f mouse heads without tongue and mandible, both the anterior and posterior palatal shelves were able to elevate in the ex vivo organ culture (Figure 2A-D). Moreover, the E14.0 Wnt1-Cre; Fam20b f/f mouse palatal shelves fused with each other, as the WT controls did in the ex vivo organ culture ( Figure 2E,F), suggesting that Wnt1-Cre; Fam20b f/f palatal shelves also retained the ability to fuse. Therefore, the cleft palate or the failed palatal elevation in Wnt1-Cre; Fam20b f/f mice was supposed to be a secondary deformity. We further generated Osr2-Cre; Fam20b f/f mice to examine this conclusion. Worthy of noticing, Osr2-Cre; Fam20b f/f mice showed intact palates, as well as little-affected tongues and mandibles compared with the wild-type counterparts ( Figure 2G-L). Taken together, these results demonstrated that Fam20b inactivation caused few primary defects in palatal elevation, and the failed palatal elevation in Wnt1-Cre; Fam20b f/f mice most likely resulted from the undescending tongue.   Wnt1-Cre; Fam20b f/f new born die within 24 h after birth with cleft palates and micrognathia. Actually, from E13.5 on, the mandibles of Wnt1-Cre; Fam20b f/f mice started to be shorter and smaller than those of control mice, which became more and more severe in the following stages ( Figure 3A-F). Bone and cartilage staining also confirmed the shortened mandibles of Wnt1-Cre; Fam20b f/f mice ( Figure 3G-L). Moreover, the Wnt1-Cre; Fam20b f/f mandibular bones were excessively curved, which significantly increased the distal angle ( Figure 3I,J). The sagittal view of Wnt1-Cre; Fam20b f/f mice showed that the angle between the highest point of the tongue and the horizontal plane increased significantly (32.67 • ± 0.57 • vs. 38.33 • ± 1.52 • , p < 0.01; Figure 3M,N), which led to the undescending tongue compared with wild-type control ( Figure 3O,P). Since the micrognathia, undescending tongue, and cleft palates in Wnt1-Cre; Fam20b f/f mice highly resembled the human Pierre-Robin sequence [37][38][39], the failed elevation in Wnt1-Cre; Fam20b f/f palatal shelves was a consequence secondary to the shortened mandible. Wnt1-Cre; Fam20b f/f new born die within 24 h after birth with cleft palates and micrognathia. Actually, from E13.5 on, the mandibles of Wnt1-Cre; Fam20b f/f mice started to be shorter and smaller than those of control mice, which became more and more severe in the following stages ( Figure 3A-F). Bone and cartilage staining also confirmed the shortened mandibles of Wnt1-Cre; Fam20b f/f mice ( Figure 3G-L). Moreover, the Wnt1-Cre; Fam20b f/f mandibular bones were excessively curved, which significantly increased the distal angle ( Figure 3I,J). The sagittal view of Wnt1-Cre; Fam20b f/f mice showed that the angle between the highest point of the tongue and the horizontal plane increased significantly (32.67° ± 0.57° vs. 38.33° ± 1.52°, p < 0.01; Figure 3M,N), which led to the undescending tongue compared with wild-type control ( Figure 3O,P). Since the micrognathia, undescending tongue, and cleft palates in Wnt1-Cre; Fam20b f/f mice highly resembled the human Pierre-Robin sequence [37][38][39], the failed elevation in Wnt1-Cre; Fam20b f/f palatal shelves was a consequence secondary to the shortened mandible.

Reduced Cell Density in Wnt1-Cre; Fam20b f/f Palatal Mesenchyme
Although the cleft palate in Wnt1-Cre; Fam20b f/f mice resulted secondarily from the Pierre-Robin-sequence-like phenotype, the primary defects in Wnt1-Cre; Fam20b f/f palates were not excluded. The size of Wnt1-Cre; Fam20b f/f palatal shelves was found to be significantly smaller than that in the control group. Statistical analysis of DAPIstained nuclei showed that the number of total cells in restricted area of anterior palates showed no difference compared with wild type (anterior: 7800 ± 1300 cells/mm 2 vs. 6780 ± 810 cells/mm 2 , p > 0.05), while the cell densities in the restricted medial and lateral regions of the E13.5 Wnt1-Cre; Fam20b f/f posterior palates were significantly reduced (medial: 7400 ± 1370 cells/mm 2 vs. 5366 ± 633 cells/mm 2 , p < 0.05; lateral: 6700 ± 1060 cells/mm 2 vs. 5200 ± 900 cells/mm 2 , p < 0.05; Figure 4A-E) Then, we investigated whether reduced cell density resulted from the altered cell proliferation and apoptosis in Wnt1-Cre; Fam20b f/f palatal shelves. TUNEL assay showed that the percentage of apoptotic cells in the total cells of posterior palate was significantly increased (anterior: 3.2987 ± 0.7942% vs. 3.2112 ± 0.8842%, p > 0.05; posterior: 1.3829 ± 0.8825% vs. 2.1457 ± 0.0842%, p < 0.05) ( Figure 4A-D,F). In contrast, Ki67 staining displayed that the percentage of proliferating cells only in anterior palate was slightly increased (40.7733 ± 0.0118% vs. 44.5367 ± 0.0256%, p < 0.05), but there was no difference in posterior palates (medial: 44.2130 ± 0.3126% vs. 46.3123 ± 0.0176%, p > 0.05; lateral: 52.4143 ± 0.9846% vs. 47.9921 ± 0.6987%, p > 0.05; Figure 4G-K). In summary, the reduced size of Wnt1-Cre; Fam20b f/f palates resulted predominantly from the increased cell apoptosis and the decreased mesenchymal cell density, which most likely also led to cleft palates by disrupting palatal fusion, even in the case of accomplished palatal elevation. Although the cleft palate in Wnt1-Cre; Fam20b f/f mice resulted secondarily from the Pierre-Robin-sequence-like phenotype, the primary defects in Wnt1-Cre; Fam20b f/f palates were not excluded. The size of Wnt1-Cre; Fam20b f/f palatal shelves was found to be significantly smaller than that in the control group. Statistical analysis of DAPI-stained nuclei showed that the number of total cells in restricted area of anterior palates showed no difference compared with wild type (anterior: 7800 ± 1300 cells/mm 2 vs. 6780 ± 810 cells/mm 2 , p > 0.05), while the cell densities in the restricted medial and lateral regions of the E13.5 Wnt1-Cre; Fam20b f/f posterior palates were significantly reduced (medial: 7400 ± 1370 cells/mm 2 vs. 5366 ± 633 cells/mm 2 , p < 0.05; lateral: 6700 ± 1060 cells/mm 2 vs. 5200 ± 900 cells/mm 2 , p < 0.05; Figure 4A-E) Then, we investigated whether reduced cell density resulted from the altered cell proliferation and apoptosis in Wnt1-Cre; Fam20b f/f palatal shelves. TUNEL assay showed that the percentage of apoptotic cells in the total cells of posterior palate was significantly increased (anterior: 3.2987 ± 0.7942% vs. 3.2112 ± 0.8842%, p > 0.05; posterior: 1.3829 ± 0.8825% vs. 2.1457 ± 0.0842%, p < 0.05) ( Figure 4A-D,F). In contrast, Ki67 staining displayed that the percentage of proliferating cells only in anterior palate was slightly increased (40.7733 ± 0.0118% vs. 44.5367 ± 0.0256%, p < 0.05), but there was no difference in posterior palates (medial: 44.2130 ± 0.3126% vs. 46.3123 ± 0.0176%, p > 0.05; lateral: 52.4143 ± 0.9846% vs. 47.9921 ± 0.6987%, p > 0.05; Figure 4G-K). In summary, the reduced size of Wnt1-Cre; Fam20b f/f palates resulted predominantly from the increased cell apoptosis and the decreased mesenchymal cell density, which most likely also led to cleft palates by disrupting palatal fusion, even in the case of accomplished palatal elevation.  To further clarify the primary influence of GAG chains deficiency on palatal development, in situ hybridization was applied to examine gene expression pattern. The expression of Shh converged in the eight palatal rugae in wild-type mice at E15.5 ( Figure 5A), while there was only seven rugaes in Wnt1-Cre; Fam20b f/f palates ( Figure 5B). Moreover, the amount of Shh spots in Wnt1-Cre; Fam20b f/f soft palates was also significantly reduced compared to wild-type soft palates ( Figure 5A,B). The Fgf10 expression in the E13.

The Altered Gene Expression in Wnt1-Cre; Fam20b f/f Palatal Mesenchyme
To further clarify the primary influence of GAG chains deficiency on palatal d opment, in situ hybridization was applied to examine gene expression pattern. Th pression of Shh converged in the eight palatal rugae in wild-type mice at E15.5 (F 5A), while there was only seven rugaes in Wnt1-Cre; Fam20b f/f palates ( Figure 5B). M ver, the amount of Shh spots in Wnt1-Cre; Fam20b f/f soft palates was also significant duced compared to wild-type soft palates ( Figure 5A,B). The Fgf10 expression in the Wnt1-Cre; Fam20b f/f posterior palatal mesenchyme was obviously decreased ( Figure 5 In contrast, Wnt5a transcription was enhanced in the E13.5 Wnt1-Cre; Fam20b f/f ant palates ( Figure 5E,F). Another transcription factor, Osr2 which was specifically expr in lateral mesenchyme of posterior palates, displayed an attenuated intensity in Wnt1-Cre; Fam20b f/f posterior palates ( Figure 5G,H). These altered gene expression terns implicated a primary effect of GAG chain deficiency on palatal development.
Histological staining showed that both Wnt1-Cre; Fam20b f/f and Wnt1-Cre; pMes-caBmpr1a mice had cleft palate, while the pMes-caBmpr1a transgene (Wnt1-Cre; Fam20b f/f ; pMes-caBmpr1a mice) failed to rescue the cleft palates ( Figure 7I-X). However, both E15.5 Wnt1-Cre; Fam20b f/f ; pMes-caBmpr1a palatal shelves were located above the tongue, suggesting the palatal elevation was recovered by the enlarged mandibles. Although the Wnt1-Cre; Fam20b f/f ; pMes-caBmpr1a palatal shelves were still too small to fuse into an intact palate ( Figure 7K,O,S,W), the ectopic cartilages detected in the posterior palates and mandibles of both Wnt1-Cre; Fam20b f/f ; pMes-caBmpr1a and Wnt1-Cre; pMes-caBmpr1a mice ( Figure 7O,P,W,X) implied that the smaller palatal shelves most likely resulted from the enhanced BMP signaling, instead of completely from GAG chain deficiency. Actually, Wnt1-Cre; Fam20b f/f ; pMes-caBmpr1a mice showed a milder chondrogenesis in palates and mandibles than Wnt1-Cre; pMes-caBmpr1a mice, implying the Fam20b-catalyzed GAG chain synthesis played a role opposite to BMP signaling in chondrogenesis ( Figure 7K Fam20b, as a member of secretory pathway kinases, is a newly identified hexokinase whose activity is a key switch in the biosynthesis of a wide array of O-linked PGs. Fam20b-dependent xylose phosphorylation in GAG chain synthesis has been demonstrated in vitro [22,23,44]. Inactivation of Fam20b in oral epithelium led to supernumerary incisors, highlighting the pivotal role of GAG chains in regulating the tooth number by determining the fate of dental epithelial stem/progenitor cells [26,27]. Loss of Fam20b in mouse cartilage led to chondrosarcoma in the knee joint and abnormal bone biomineralization [42], which was consistent with the results in zebrafish wherein inactivation of Fam20b led to aberrant organization of cartilage matrix and skeletal defects [43]; both were linked to abnormal GAG chain biosynthesis. In this study, we investigated the function of Fam20b-dependent GAG chains in craniofacial neural crest cells. Consistent with previous studies [42,43], the GAG chain accumulation in CNC-derived cells was significantly reduced, indicating the deficient synthesis of GAG chains due to Fam20b deletion. Wnt1-Cre; Fam20b f/f mice died of complete cleft palate within 24 h of birth. Except for cleft palate, the Wnt1-Cre; Fam20b f/f mice also exhibited microcephaly, widened fontanels, micrognathia, tongue elevation, temporomandibular joint abnormalities, and reduced mineralization in the skull, maxilla, mandible, and palatine. These results confirmed the essential role of Fam20b in craniofacial morphogenesis and mineralization. PGs are widely distributed in extracellular areas, the pericellular-basement membrane, cell surfaces, and intracellular spaces [17]. They interact with various signals, including BMP, WNT, FGF, and HH, mainly through different types of GAG chains [14,45]. Considering that inactivation of Fam20b affected the synthesis of several types of GAG chains, such as CS, DS, and HS, the impacts of knocking out Fam20b in CNC-derived cells on maxillofacial development could be diverse. In addition, the intracellular signaling pathways and intercellular communication affected by Fam20b loss would also be various. Therefore, further investigations are required to elucidate the specific mechanisms underlying the various craniofacial deformities caused by the inactivation of Fam20b.

The Primary and Secondary Effects of GAG Chains in Palatal Mesenchyme on Palatogenesis
The molecular mechanism controlling the elevation of the palatal shelves are still unclear. It was widely speculated that the anterior palate elevates in a 'flip-up' manner by the elastic force generated by the ECM, while the posterior palatal process remodels itself through regression in distal shelves and outgrowth in the horizontal direction [3]. Due to the capacity of binding a large amounts of water molecules to provide tissue plasticity, GAG chains were considered as the main source of elastic force for elevation [3,46]. In this study, Wnt1-Cre; Fam20b f/f mice exhibited a phenotype highly similar to human Pierre Robin sequence, which was characterized by a primary mandible malformation leading to tongue elevation and secondary cleft palate [37][38][39]. To verify whether elevation failure was secondary to micrognathia, we generated Osr2-Cre; Fam20b f/f mice, in which Fam20b was deleted only in the palatal mesenchyme, but not in mandible [47]. Since the Osr2-Cre; Fam20b f/f palate, tongue, and mandible exhibited no defect, the failed elevation in Wnt1-Cre; Fam20b f/f palates was caused by physical obstruction of tongue malposition rather than the loss of elastic force due to GAG chain deficiency. Furthermore, in vitro organ cultures directly confirmed that the palatal elevation and fusion were little affected by GAG chain deficiency. These results were consistent with those in Wnt1-Cre; Has2 f/f mice [34,35], in which HA was eliminated, overthrowing the putative hypothesis that GAG chains generated the main driving force for palatal elevation.
Our results also showed that GAG chain deficiency leads to an apparent increase in apoptosis in the middle and posterior plate, and a reduced cell density, which caused a smaller palatal process, disabling the contact and fusion of palatal shelves. These results were contrary to the Wnt1-Cre; Has2 f/f and Golgb1 mutant mice, in which reduced ECM was accompanied with an increased cell density in palate mesenchyme [34][35][36]. Overall, our study indicated that the deficiency of Fam20b-catalyzed GAG chain synthesis not only resulted in secondary cleft palate by causing human Pierre-Robin-sequence-like phenotype, but also primarily suppressed cell survival and osteogenesis of palatal mesenchyme.

GAG Chains of Proteoglycans Were Involved in CNCC Osteogenesis by Mediating BMP Signaling
The palatine bone, as well as the mandibular bone, are directly differentiated from CNC-derived cells through intramembrane ossification [48][49][50]. In this study, the deficient synthesis of GAG chains resulted in micrognathia and abnormal palatine osteogenesis. Several signaling pathways were involved in osteoblast induction [48][49][50], among which the BMP pathway played essential roles in mesenchyme condensation and osteogenic differentiation during palatal osteogenesis [51]. We found a significant down-regulation of BMP signaling in Wnt1-Cre; Fam20b f/f palatal mesenchyme. The essential osteogenic transcription factors, including Sox9 and Osterix (Sp7), were also apparently reduced. The consistent results in Osr2-IresCre;Bmpr1a f/f mice [51] indicated that the affected palatal osteogenesis in Wnt1-Cre; Fam20b f/f mice was mainly caused by the suppressed BMP signaling. Subsequently, by enhancing BMP activity in CNC-derived mesenchyme through caBmprIa transgene, we showed a significantly increased osteogenesis in the anterior palate. Although Wnt1-Cre; Fam20b f/f ; pMes-caBmpr1a still had a cleft palate, the elevated palatal process could be attributed to the partially rescued micrognathia by activated BMP signaling, suggesting that the activity of BMP signaling was mediated by GAG chains in maxillofacial osteogenesis. We previously reported that constitutively activated Bmpr1a or suppressed BMP signaling by over-expressing Noggin both resulted in cleft palate and abnormal osteogenesis [52], emphasizing the precise regulation of BMP signaling in palatal development. Currently, we hypothesized that inactivation of Fam20b led to a suppressed BMP signaling by attenuating signal enrichment or interactions between the growth factors and their receptors, but the specific mechanism still needed further explorations.
Currently, up to 27 skeletal disorders are reported to be associated with the mutations of 23 genes involved in GAG biosynthesis, which reflects the key roles of the GAG chains in the development and homeostasis of cartilage and bone [16,17]. Although Fam20b was indispensable in the osteogenesis and chondogenesis of zebra fish and mice, there were few reports on the function of FAM20B in humans. In the latest report, two lethal composite heterozygous variants in FAM20B (NM_014864 c.174_178delTACCT p.T59Afs*19/c.1038delG p.N347Mfs*4) were identified in a newborn who died soon after birth [41] and who suffered from a spectrum of defects highly similar to the autosomal recessive heterogeneous disorder, Desbuquois dysplasia (DBQD), which was characterized by joint laxity and skeletal change [53][54][55][56]. Several cases of DBQD have exhibited dysmorphic facial features, such as midface hypoplasia, flat nasal bridge, micrognathia, and cleft palate [57,58], resembling Pierre-Robin sequence and Wnt1-Cre; Fam20b f/f defects. Up to now, calcium-activated nucleotidase 1 gene (CANT1) and xylosyltransferase 1 (XYLT1) gene have been identified as the causative genes of DBQD [58,59]. Since both CANT1 and XYLT1 are involved in GAG chain synthesis, especially XYLT1, which functions upstream of FAM20B [41], FAM20B was most likely a novel causative gene for DBQD. Overall, our study provides new clues in the exploration of the pathogenesis and effective treatment of bone dysplasia, especially in craniofacial bones, resulting from deficient GAG chain synthesis.

Mouse Lines
All the animal experimental procedures in this study followed the protocol approved by the Animal Care and Use Committee at Dalian Medical University (Protocol No. AEE17038). The Wnt1-Cre mice, Osr2-cre knock-in mice (Osr2-cre KI ), Fam20b f/f mice and pMes-caBmpr1a mice were described in previous studies [27,52,55,56]. To generate Wnt1-Cre; Fam20b f/f mice, we first crossed Wnt1-Cre mice with Fam20b f/f mice to obtain Wnt1-Cre; Fam20b f/+ mice, and then, Wnt1-Cre; Fam20b f/+ mice were mated with Fam20b f/f mice to generate Wnt1-Cre; Fam20b f/f mice. Osr2-Cre; Fam20b f/f mice was obtained in the same way. Genotyping was performed as previously described [27].

Alcian Blue Staining
The mouse embryonic heads were fixed overnight with 4% paraformaldehyde, dehydrated with gradient ethanol, embedded in paraffin, and made into 10 µm paraffin slices. The slices were stained with Alcian blue solution at pH 2.5 for 30 min, and counter-stained with nuclear fast red.

Skeleton and Cartilage Staining
The mouse embryos were fixed with anhydrous ethanol after removal of skin, viscera, and muscles, and then soaked in acetone to remove lipids. The working solution containing 0.1% Alcian blue and 0.3% Alizarin red was used for skeleton staining for 3 days, in which cartilage was stained blue and bone was stained red. Then, the excessively stained tissue was decolorized with 2% KOH. A stereoscopic microscope (Olympus, Shinjuku City, Japan) was used for imaging.

Organ Culture
The mandibles were removed from E13.0 and E14.0 embryonic heads, and collected in a 15 mL conical tube containing 3 mL DMEM, which was supplemented with 20% fetal calf serum (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA). Tubes were fixed in a roller incubator at 37 • C and 5% CO 2 for 12 h culture. Then, the heads were washed in PBS and fixed in 4% paraformaldehyde for paraffin sections and histological staining.

Cell Proliferation and Apoptosis Analysis
Cell proliferation was detected by Ki67 immunostaining. After rehydration by gradient alcohol and antigen retrieval by citric acid, the paraffin sections were incubated with Ki67 (1:400) overnight at 4 • C. The Maxvision TM HRP polymer anti-mouse/rabbit IHC kit (Catalog No. KIT5020, Maixin Ltd., Shenzhen, China) was applied as a secondary antibody for 1 h incubation at room temperature, and then developed by 3,3 -diaminobenzidine (Catalog No. DAB-0031, Maixin Ltd., China) for coloring, and counterstained with hematoxylin. The apoptosis assay was conducted by In Situ Cell Death Detection Kit, POD (Catalog No. 11684817910, Roche, Basel, Switzerland) as instructed by manufacturer.

In Situ Hybridization
The mouse heads were fixed in paraformaldehyde treated with diethyl pyrocarbonate (DEPC), and then, dehydrated with gradient RNase free ethanol. Paraffin embedded heads were made into 10 um sections. Whole mounts and section in situ hybridization used digoxygenin-labelled antisense RNA probes (Catalog No. 11277073910, Roche, Switzerland) for Fam20b, Shh, Fgf10, Wnt5a and Osr2 were followed by incubation with anti-digoxygenin-AP conjugate antibody (Catalog No. 11093274910, Roche, Switzerland). Bm-purple (Catalog No. 114420740010, Roche, Switzerland) was used for color development and nuclear fast red for counter-staining.

Immunohistochemistry
Gradient alcohol rehydration was performed on paraffin sections, and antigen repair was performed by citric acid repair solution with pH 6.0. The endogenous peroxidase and non-specific antigen binding sites were blocked by 3% H 2 O 2 and 0.3% Triton X-100 sheep serum, respectively. Then, slides were incubated overnight at 4 • C with the primary antibodies against phosphorylated-Smad1

Statistical Assay
The experiments were performed with at least three biological replicates for statistical analysis in each group. Ki67 positive cells, TUNEL positive cells, and the total number of cells within the restricted areas were counted using Image J (version 1.46r, National Institutes of Health), followed by statistical analysis using GraphPad Prism 9 (GraphPad Software Inc., San Diego, CA, USA). Each group of data was analyzed using unpaired double-tailed Student's t-test, and p value < 0.05 was considered statistically significant.

Data Availability Statement:
The data of this study are available from the corresponding authors upon reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.