Trps1^−/− and Trps1^+/− mice were described previously (Malik et al., 2002), and maintained on 129svev and C57BL/6J backgrounds. For timed matings, the day the plug was observed was designated E0.5. All animal work was carried out under approved IACUC protocols.

Skeletal staining of E18.5 C57BL/6J wild type (WT) (n = 5) and Trps1^−/− (n = 4) mouse embryos was performed with 0.03% Alcian blue (Sigma-Aldrich, A3157) and 0.01% Alizarin red (Sigma- Aldrich, A5533) solution. Imaging was performed on Leica M165FC microscope and Leica Application Suite software.

Palatal shelves from E13.5 129svev WT (n = 11), Trps1^+/− (n = 5), and Trps1^−/− (n = 11) mice were dissected. Pairs of palatal shelves were placed on trans-well inserts (8 μm pore size) and incubated in α-MEM (without serum) at 37°C for 48 h as described in Wang et al. (2013).

Embryos were fixed with 4% paraformaldehyde, dehydrated, and embedded in paraffin following standard protocols. In situ hybridization was performed as described previously (Morello et al., 2001; Napierala et al., 2008).

Seven-micrometer-thick sections of paraffin-embedded 129svev mice and palatal shelves were stained with H&E according to standard protocols. The stained tissues were mounted in mounting medium (Richard-Allan Scientific).

C57BL/6J WT and Trps1^−/− mice were fixed in either 4% paraformaldehyde (for detection of Trps1 and Tgfβ3) or Carnoy’s solution (for detection of chondroitin sulfate proteoglycan (CSPG), β-catenin, and Twist1) and embedded in paraffin. Heat-induced epitope retrieval was performed in sodium citrate buffer (10 mM sodium citrate, 0.05% Tween20, pH = 6.5). The primary antibodies used were 1:50 rabbit anti-TRPS1 (Abnova, PAB17465), 1:200 anti-chondroitin sulfate proteoglycan (Sigma, C8035), 1:100 rabbit anti-TGFβ3 (Abcam, ab15537), 1:250 rabbit anti-β-catenin (Abcam, ab50581), 1:100 rabbit anti-Twist1 (Abcam, ab32572), and 1:20 anti-Ki67 (Santa Cruz, sc-7846). Anti-rabbit AlexaFlour 488-conjugated secondary antibody (Thermo Fisher Scientific) and ProLong Gold with DAPI Antifade Mountant (Molecular Probes, Thermo Fisher Scientific) were used. Images were taken with Zeiss AxioCam on a Zeiss Axioskop A1 microscope and ZEN software.

Apoptosis was evaluated using TUNEL assay (Thermo Fisher Scientific) according to the manufacturer’s protocol.

Total head length was measured from the tip of nasal cartilage to supraoccipital bone. The nasal length was measured from the tip of nasal cartilage to the frontonasal suture. The nasal angle was taken between the plane connecting the tip of the nasal cartilage to the superior border of supraoccipital bone and the plane tangent to the nasal bone from the images using a protractor. Mandible length was measured from the most posterior point at the condylar process to the most anterior point at the symphysis. Lengths of the head, nose, and mandible were measured with calipers. Nonparametric Wilcoxon rank sum test was used to determine a difference between WT and Trps1^−/− mice head, nose, and mandible measurements with α = 0.05. The statistical analysis was performed using StataSE software.

TRPS1 haploinsufficiency in humans results in a characteristic facial appearance, which suggests a disturbed development of the skeletal elements. This, in addition to the known role of Trps1 in appendicular skeleton development, led us to studies addressing the role of Trps1 in the formation of the craniofacial skeleton. Trps1^+/− mice have a mild craniofacial phenotype, while Trps1^−/− mice die at birth (Malik et al., 2002). Due to the neonatal lethality of Trps1^−/− mice, we focused our analyses on E18.5 embryos. Comparisons of skeletal preparations of E18.5 WT and Trps1^−/− mice showed that the mean head length of Trps1^−/− head (9.4 ± 0.3 mm) was not significantly different from WT (9.9 ± 0.4 mm) (Figure 1). However, the mean nasal length was significantly shorter in Trps1^−/− (2.1 ± 0.1 mm) than the WT littermate controls (2.5 ± 0.1 mm) (p < 0.05) (Figure 1). The mean nasal angle of the Trps1^−/− mice (31.1 ± 1^°) was also significantly steeper than the WT (26.4 ± 2.8^°) (p < 0.05) (Figure 1). No apparent differences in the nasal cartilage were detected.

Multiple differences between WT and Trps1^−/− mandibles were also detected. The Trps1^−/− mutant phenotype includes unerupted incisors, reduced mineralized regions as well as less cartilage in the coronoid, condylar, and angular processes (Figure 1B). Trps1^−/− mandibles (4.6 ± 0.1 mm) were significantly shorter than the WT counterparts (5.6 ± 0.1 mm) (p < 0.05) (Figure 1). Meckel’s cartilage and clearly demarcated molar crypts were present in both genotypes (Figure 1). Zygomatic arches were less developed in the Trps1^−/− mice in comparison with WT littermates (Figures 2A,B), which is consistent with the craniofacial skeletal phenotype of adult Trps1^+/− mice (Malik et al., 2002). This reduction in mandibular and midfacial structures resembles the jaw hypoplasia observed in patients with TRPS. Additionally, analysis of the skeletal preparations revealed underdeveloped vomer bones in Trps1^−/− mice (Figures 2C,D). Finally, we observed cranial base abnormalities in Trps1^−/− mice, where the basisphenoid was smaller and the presphenoid bone was absent (Figures 2E–H). However, there was no difference in the basioccipital and cranial vault bones between the two genotypes (Figures 2I,J). These results suggest that Trps1 is important specifically for the development of the nose, jaw, and cranial base.

TRPS patients occasionally present with cleft palate (Morioka et al., 1999; Solc et al., 2017) and the same abnormality was noted in our previous studies of Trps1^−/− mice (Kantaputra et al., 2008). Here, the skeletal staining revealed that the palatal shelves were still widely separated at E18.5 in Trps1^−/− mice (Figure 2F). This was further confirmed by histological analysis (Figures 3A,B). Importantly, skeletal preparations and histological analyses detected cleft palate in all Trps1^−/− mice analyzed. Thus, this abnormality shows complete penetrance and is independent of the genetic strain, as it was present in all Trps1^−/− mice on C57BL/6J and 129svev background.

Following the cleft palate phenotype in Trps1^−/− mice, we focused on understanding the role of Trps1 during palatogenesis. Previous studies have shown that during early embryogenesis (E11.5), Trps1 is highly expressed in the first branchial arch mesenchyme (Kunath et al., 2002; Kantaputra et al., 2008). To delineate Trps1 expression in the oral cavity, we used both RNA in situ hybridization and protein detection with immunostaining. Analysis of the Trps1 expression pattern during the growth, elevation, and fusion of palatal shelves revealed Trps1 expression in the mesenchyme of the presumptive vomer condensation and in the tips of the palatal shelves in the maxillary prominence. In addition, the tongue mesenchyme as well as mesenchyme of the mandibular process strongly expressed Trps1 (Figures 3C,D). Specifically, at E12.5, Trps1 protein was detected in the mesenchyme and epithelium of palatal shelves, including the epithelium surrounding the palatal shelves at the future fusion sites (Figure 3E). Trps1 was also found in the nasal region of the palatal shelves (Figures 3E,F). At E13.5, the presence of Trps1 was more widespread in the palatal shelves, mandible, and tongue, especially in the posterior region of the oral cavity (Figures 3G,H). When the palatal shelves were in the adhesion/fusion stage at E14.5, Trps1 was detected in the maxillary mesenchyme and palatal shelf epithelium (Figure 3I). Trps1 expression was evident in the seam of medial edge epithelium (MEE) (Figure 3J), suggesting that Trps1 is involved in palatal shelf fusion.

Given the expression of Trps1 in the palatal shelves and its previously reported effect on cell proliferation (Suemoto et al., 2007; Napierala et al., 2008; Wuelling et al., 2009), we compared proliferative and apoptotic activity by immunofluorescent detection of Ki-67 and TUNEL assay, respectively. These analyses did not detect any apparent differences in proliferation or apoptosis between WT and Trps1^−/− E14.5 mice palatal shelves (Figure 4). Since Trps1 was detected specifically in the MEE, we focused our further analyses on the palatal shelves fusion process. To determine whether Trps1 deficiency disrupts palatal shelves fusion, the ex vivo organ culture approach was used. Palatal shelves isolated from WT, Trps^+/−, and Trps1^−/− mice were placed in close contact with each other and remained in contact during the ex vivo culture (Figures 5A,B). After 48 h of ex vivo culture, fusion of 100% of WT and Trps^+/− palatal shelves occurred (Figure 5C). However, none of the Trps1^−/− palatal shelves pairs initiated the fusion process (Figure 5D). Given that palatal shelf adhesion is a prerequisite for the fusion, we examined the presence of the cell adhesion mediator chondroitin sulfate proteoglycan (CSPG) in palatal shelves (Gato et al., 2002). Immunohistochemistry confirmed the presence of CSPG on the fusion surface of the WT palatal shelves, but CSPG was undetectable on the surface of Trps1^−/− palatal shelves and nasal septum (Figures 5E–H). Importantly, CSPG was detected in the palatal shelf mesenchyme of both WT and Trps1^−/− mice, indicating that Trps1 deficiency results in a cell type-specific loss of CSPG in the epithelium.

It has been shown that the accumulation of CSPG at the fusion surfaces depends on the expression of transforming growth factor-beta 3 (Tgfβ3) in palatal shelf epithelium (Gato et al., 2002; Krauss, 2011). The association between TGFβ3 mutation and cleft palate in both human and animal studies highlights the importance of this signaling molecule in palatogenesis (Proetzel et al., 1995; Carinci et al., 2007; Rienhoff et al., 2013). Thus, we investigated whether the loss of CSPG in Trps1^−/− mice may be due to Tgfβ3 deficiency. As reported before, Tgfβ3 was highly and specifically expressed in the MEE of WT mice palate (Figures 6A,B). In contrast, Tgfβ3 protein was not detected in the Trps1^−/− palatal shelves (Figures 6A,B). This suggests that CSPG deficiency on the fusion surface is a consequence of the loss of Tgfβ3 expression in Trps1^−/− palatal shelf epithelium.

The absence of Tgfβ3 in the Trps1^−/− epithelium led us to investigate other proteins that are essential for palatal fusion and may be involved in Trps1 regulatory networks. We analyzed Twist1, which is the key EMT regulator required for palatal fusion and functions downstream of the Tgfβ3, and β-catenin, which has been implicated in both Trps1 and Tgfβ3 signaling (Yu et al., 2008; He et al., 2011; Fantauzzo and Christiano, 2012). Both Twist1 and β-catenin proteins were primarily present in the epithelium of the WT E14.5 palatal shelves, but their expression was undetectable in the Trps1^−/− palatal shelves (Figures 6A,B). Therefore, these results suggest that Trps1 is required for the epithelial expression of several proteins critical for the palatal shelves fusion.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.