Wild-type and Hoxa2^−/− embryos were obtained from timed pregnant Hoxa2^+/− (heterozygous) mice. Genotype was confirmed using PCR as previously described (Gendron-Maguire et al., 1993). This research was approved by the University of Saskatchewan's Animal Research Ethics Board and adhered to the Canadian Council on Animal Care guidelines for humane animal use.

Primary MEPM cells were isolated from micro-dissected palatal shelves of wild-type and Hoxa2^−/− mouse embryos at E13.5. The palatal shelves were treated with 0.25% trypsin-EDTA for 15 min, passed through a 70 μm cell strainer and cultured as monolayer cells (Iwata et al., 2012) in DMEM: Ham's F12 (1:1) media containing 10% FBS, 1% antibiotic-antimycotic solution (Sigma). Osteogenic differentiation was carried out as described previously (Kwong et al., 2008) with minor modifications. Briefly, MEPM cells were seeded on 0.1% gelatin or poly-D-lysine coated plates at a cell density of 5 × 10⁴ cells per well in 24 well plates and cultured until they reached confluence. Osteogenic differentiation was induced with differentiation media (DMEM, 10% FBS, 2 mM L-glutamine and 1% antibiotic-antimycotic solution) supplemented with osteogenic inducing agents, including 50 μg/ml L-ascorbic acid 2-phosphate sesquimagnesium salt (Sigma), 10 mM β-glycerophosphate (Sigma), and 100 nM dexamethasone (Sigma). Cells were differentiated for up to 21 days (d) and samples were collected at d8, d15, or d21. To assess the impact of BMP signaling, MEPM cells were treated with dorsomorphin (5 μM) or DMSO and were harvested at d8 for further experiments.

ALPI staining in the palatal shelves in vivo was carried out as previously described (Baek et al., 2011). Embryonic mouse heads were fixed overnight in freshly prepared 4% paraformaldehyde at 4°C and rehydrated in 30% sucrose at 4°C. Frozen coronal sections (10 μm) were prepared on slides coated with 0.5% gelatin. The sections were air dried for at least 2 h and then rehydrated with TBS with 0.08% tween-20 two times for 10 min each. Subsequently, the sections were treated with alkaline phosphatase buffer (100 mM NaCl, 100 mM Tris-HCl pH 9.5, 50 mM MgCl₂, 0.1% Tween-20) for 20 min and stained with alkaline phosphatase buffer containing 4.5 μl/ml of 5-bromo-4-chloro-3-indolyl phosphate (Roche) and 3.5 μl/ml of nitro blue tetrazolium (Roche) for 10 min. The reaction was stopped with PBS containing 20 mM EDTA buffer and counter stained with nuclear fast red. The stained sections were dehydrated in a series of PBS/ethanol, ethanol/xylene and finally mounted in DPX mounting media (Sigma). For osteoblast differentiation in primary MEPM cells, ALPI staining was carried out following the aforementioned protocol after fixing the cells with 4% paraformaldehyde for 15 min.

ARS staining and quantification was carried out as previously described (Gregory et al., 2004) with minor modifications. Briefly, monolayer MEPM cells were fixed with 4% paraformaldehyde for 15 min and stained with 250 μl of 40 mM ARS (Sigma) solution (pH 4.1) at room temperature for 20 min with gentle shaking. Excess dye was aspirated and washed with deionized water before imaging. ARS quantification was carried using an acid extraction method (Gregory et al., 2004). Standard plot of ARS concentration was constructed by serially diluting the 40 mM ARS in the buffer containing 10% (v/v) acetic acid and 10% (v/v) ammonium hydroxide. The absorbance values of the standard concentrations were used to interpolate the concentrations of the test samples.

Total RNA was isolated from the micro-dissected palatal shelves using RNA mini spin column (Bio-Rad) as per the manufacturer's protocol. First strand complementary DNA synthesis (reverse transcription) was performed in 20 μl reactions with 500 ng of total RNA using High-Capacity complementary DNA Reverse Transcription Kit (Invitrogen). qRT-PCR was carried out as described in our previous study (Thangaraj et al., 2017) using SYBR green assay (Applied Biosystems) in 7300-real time PCR system (Applied Biosystems) with primers listed in Table 1.

Western bot analyses were carried out as previously described (Brown and Nazarali, 2010). Briefly, palatal tissues were homogenized in RIPA buffer. Total protein content was quantified using the Bradford assay and proteins were separated in 10% SDS-PAGE. Primary antibodies used were: RUNX2 (1:500; Abcam ab102711), SP7 (1:1500; Abcam ab22552), phosphorylated SMAD 1/5/8 (pSMAD 1/5/8) (1:500; Cell signaling 9511S), SMAD 1/5/8 (1:500; Santa Cruz Biotechnology sc-6031-R), and β-ACTIN (1:2,000; Developmental Studies Hybridoma JLA20). Densitometric analyses were carried out using AlphaView software.

Embryonic mouse heads were fixed overnight with freshly-prepared 4% paraformaldehyde and rehydrated in 30% sucrose at 4°C. Frozen coronal sections (10 μm) were rehydrated with PBS for 45 min, permeabilized with 0.1% Triton X-100 and blocked with 3% skim milk containing 0.1% Triton X-100 in 1X PBS for 1 h at room temperature. Sections were then incubated overnight with the following primary antibodies: RUNX2 (1:200; Abcam ab23981) or SP7 (1:800; Abcam ab22552) in 1X PBS with 0.1% Triton X-100 at 4°C. Double labeling was carried out by co-incubating: Ki67 (1:100; eBioscience 14569882) and RUNX2 (1:200; Abcam ab23981) overnight at 4°C. Sections were then washed three times and treated with secondary antibodies conjugated with Alexa Fluor® 488 (1:200) or Alexa Fluor® 594 (1:400) in 1X PBS with 0.1% Triton X-100 at room temperature for 1.5 h. Cell counting analyses were carried out manually using ImageJ software platform (NIH).

Cell proliferation assay was carried out in MEPM cells using cell-counting kit-8 (Dojindo) as previously described (Iwata et al., 2010). MEPM cells were incubated with CCK-8 reagent for 1 h and the absorbance measured at 450 nm was plotted to calculate the relative cell proliferation rate.

Statistical analyses were carried out using unpaired t-test in the case of two groups. One-way ANOVA or two-way ANOVA with Bonferroni multiple comparison test was used for one or two variate analyses, respectively. A p-value of <0.05 was considered significant.

To investigate the role of Hoxa2 in osteogenesis of the palatal mesenchyme, we first examined changes in osteogenic differentiation in the embryonic palatal shelves of Hoxa2^−/− mice at E16.5, a stage when both the prospective palatal process of the maxilla and the palatal process of the palatine bone evidently ossify (Baek et al., 2011). Staining for ALPI, a marker of osteoblast differentiation showed an expansion in ALPI expression domain in the Hoxa2^−/− palatal mesenchyme compared to wild-type (Figures 1A–H). At the anterior region of the hard palate, ALPI staining was restricted to the nasal half in two condensations of the prospective palatal process of the maxilla on either side of the degraded midline epithelial seam in wild-type embryos (Figures 1A,E). In contrast, the domain of ALPI positive preosteoblast area was increased and expanded toward the oral side covering oral-nasal axis in Hoxa2^−/− palatal shelves (Figures 1B,F). In the posterior region of the hard palate, ALPI staining was present in the ossifying centers of the palatal process of the palatine bone in wild-type embryos (Figures 1C,G), whereas there was an expansion in the expression domain of ALPI positive preosteoblasts toward the oral side in Hoxa2^−/− embryos (Figures 1D,H).

Two well-known regulators of osteogenic differentiation RUNX2 (Komori et al., 1997) and SP7 (previously known as Osterix; Nakashima et al., 2002) have been implicated in the patterning of the palatal bones (Baek et al., 2011). To elucidate the spatial mis-regulation of palatal bone formation in Hoxa2^−/− mice, expression pattern of these two osteoblast-specific transcription factors were assessed at E16.5. Immunohistochemical analyses revealed that RUNX2 (Figure 1I) and SP7 (Figure 1M) expressions were confined to the condensations of the palatal process of the maxilla at the anterior hard palate in wild-type embryos at E16.5, whereas RUNX2 (Figure 1J) and SP7 (Figure 1N) expression domains were increased in Hoxa2^−/− embryos. In the posterior region, along the developing palatal process of the palatine bone, the expression of RUNX2 (Figure 1L) was increased in Hoxa2^−/− embryos compared to wild-type (Figure 1K). In this region, the expression of SP7, a downstream target of RUNX2 and a marker of mature osteoblasts, was not evidently increased in the Hoxa2^−/− palate (Figure 1P) compared to wild-type (Figure 1O). This suggests that cells toward the oral side of the palatal process of the palatine bone are at immature osteoblast stage and may not have developed bone matrix by E16.5. Collectively, these data indicate that Hoxa2 could be a potential inhibitor of osteogenic differentiation in the palatal mesenchyme, which may serve to spatially restrict the expression of osteoblast-specific proteins during palate development in vivo.

Furthermore, gene expression profiles of osteoblast markers were evaluated during the initiation of ossification of the palatal process of the palatal bone and the palatal process of the maxilla at E13.5 and E15.5, respectively. The loss of Hoxa2 in the developing palate resulted in an increase in mRNA expression of osteoblast markers such as Runx2, AlpI and Sp7 at both E13.5 (Figures 2A–D) and E15.5 (Figures 2E–H). At E13.5, mRNA expression of Runx2, AlpI and Sp7 were increased to ~6.36-, ~9.65-, and ~2.62-fold, respectively, in Hoxa2^−/− palate compared to wild-type (Figures 2A–C). At E13.5, mRNA expression of Bglap (previously known as Ocn) was not significantly altered (Figure 2D). At E15.5, mRNA expression of Runx2, AlpI, Sp7, and Bglap were upregulated ~1.86-, ~2.29-, ~1.42-, ~3.27-folds, respectively, in Hoxa2^−/− palate compared to wild-type (Figures 2E–H). Consistent with this, protein expression of RUNX2 was upregulated at E13.5 and E15.5 to ~1.4-fold (Figures 2I,J). SP7 protein expression, both long (Figures 2I,K) and short isoforms (Figures 2I,L) were upregulated at E15.5 to ~1.4-fold. These data reveal that along with regulating the spatial patterning of osteogenic differentiation, Hoxa2 also regulates the expression of osteogenic markers at the molecular level in the developing palate.

To evaluate the potential of Hoxa2 in regulating the temporal pattern of osteogenesis, the primary mesenchyme cells from the wild-type and Hoxa2^−/− palatal shelves were differentiated in vitro for up to 21 days (d). Osteogenesis of mesenchymal cells involves sequential stages of proliferation, osteogenic commitment around day8 (~d8) followed by matrix deposition (~d15) and mineralization (~d21) (Gordon et al., 2010). Differentiating cells were stained for ALPI at d8 and Alizarin Red S (ARS) at d15 and d21. ALPI staining showed an increased osteoblast differentiation at d8 in Hoxa2^−/− MEPM cells compared to the wild-type MEPM cells (Figures 3A,B). In addition, ARS staining followed by quantification of ARS extracted matrix showed that Hoxa2^−/− MEPM cells exhibited increased extracellular matrix deposition ~2-fold at d15 (Figures 3C,D,G) and increased mineralization ~1.5-fold at d21 (Figures 3E,F,H) compared to the wild-type MEPM cells.

Next, the gene expression profiles of osteogenic markers were examined in wild-type and Hoxa2^−/− MEPM cells during osteogenic differentiation in vitro. Runx2 mRNA expression was increased to ~1.9-fold in the Hoxa2^−/− MEPM cells compared to the wild-type during osteoblast commitment stage at d8 (Figure 3I). AlpI and Sp7 mRNA expression were increased ~2.85 and ~3.37-fold, respectively, during matrix deposition stage at d15 in the Hoxa2^−/− MEPM cells (Figures 3J,K). Bglap mRNA expression was increased ~6.37-fold during matrix deposition at d15 and ~8.09-fold during matrix mineralization at d21 in the Hoxa2^−/− MEPM cells (Figure 3L). Thus, loss of Hoxa2 results in upregulation of osteogenic marker expression in a stage-specific manner as early as d8 (osteogenic commitment stage). These data indicate that Hoxa2 may play a role in early osteoblast differentiation by inhibiting the transcription factors regulating osteogenic fate specification.

Hoxa2 peaks in its expression in the developing palate at E13.5 (Smith et al., 2009), a stage when the mesenchymal cells simultaneously proliferate and commit to form preosteoblasts of the prospective palatal process of the palatine bone. This suggests that the cleft palate phenotype in Hoxa2^−/− mice, due to the failure of palatal shelves to elevate and reorient horizontally above the tongue after E13.5 (Barrow and Capecchi, 1999), may be a consequence of abnormal cell proliferation (Smith et al., 2013) and osteogenic differentiation (Wu et al., 2008; Fu et al., 2017; Jia et al., 2017a,b). To gain further insight into the role of Hoxa2 during this early stage of palate development, the rate of mesenchymal cell proliferation and the commitment of mesenchymal cells to osteoprogenitor fate was investigated in vivo at E13.5. Immunohistochemical staining of RUNX2 (Figures 4A,B) was used to evaluate osteoprogenitor commitment in the wild-type and Hoxa2^−/− palatal shelves at E13.5. RUNX2 (Figure 4A) expression in the wild-type was restricted to the bend region in the nasal side of the palatal shelves, whereas the expression domain of RUNX2 (Figure 4B) was increased spatially toward the medial edge of the palate as well as to the oral side of the palate in the Hoxa2^−/− mutants. This is similar to the aberrant expression patterns of RUNX2 observed at E16.5 in the Hoxa2^−/− palatal shelves. In addition, the number of RUNX2-positive osteoprogenitor cells on the nasal side of the palatal shelves were significantly higher in the Hoxa2^−/− mutants (~64%) compared to wild-type (~33%; Figure 4C).

Next, the rate of cell proliferation was assessed at E13.5 using Ki67 immunostaining. The percentage of Ki67-positive cells was significantly increased in the Hoxa2^−/− palatal mesenchyme (~50%) compared to wild-type (~26%; Figures 4D–F). In the nasal side of the palatal shelves, the percentage of Ki67-positive cells was ~53% in Hoxa2^−/− embryos compared to ~26% in the wild-type (Figure 4F). Interestingly, the nasal side mesenchyme displayed a higher proliferation rate of ~53% compared to the oral side of ~23% in Hoxa2^−/− palatal shelves (Figure 4F). In addition, the percentage of proliferating osteoprogenitor cells (RUNX2-positive/Ki67-positive) in the nasal side of the Hoxa2^−/− palatal shelves was higher (~20%) compared to ~11% in wild-type (Figures 4G–I). Furthermore, mRNA expression of cyclin D1 (Ccnd1), a critical G1 phase cell cycle regulator was also upregulated in the Hoxa2^−/− palatal shelves from E12.5 to E14.5 (Figure 4J). These results indicate that Hoxa2 plays a critical role by inhibiting osteoprogenitor commitment and osteoprogenitor proliferation prior to the elevation and fusion of the palatal shelves.

To understand the molecular signaling pathways underlying the aberrant cell proliferation and osteogenic differentiation in the Hoxa2^−/− palatal shelves, BMP signaling was investigated as it is critical for cell proliferation (Zhang et al., 2002) and expression of osteogenic markers in the developing palate (Baek et al., 2011). First, the mRNA expression of BMP ligands critical for osteoblast differentiation such as Bmp2 and Bmp4 in the developing palatal shelves was examined. Bmp2 expression was upregulated to ~3.57-fold at E13.5 and ~1.96-fold at E15.5 in Hoxa2^−/− palatal shelves compared to wild-type (Figure 5A). Similarly, Bmp4 expression was upregulated to ~3.42-fold at E13.5 and to ~1.81-fold at E15.5 (Figure 5B). Immunoblotting analyses revealed that canonical BMP signaling mediated by pSMAD 1/5/8 was also upregulated to ~1.5-fold in the Hoxa2^−/− palate at E15.5 (Figures 5C,D). These results indicate that canonical BMP signaling pathway may be downstream of the Hoxa2 gene network in palate development.

To determine if the upregulated canonical BMP signaling is functionally responsible for the increased mesenchymal cell proliferation and osteogenic differentiation observed in the Hoxa2^−/− palate, dorsomorphin was used to inhibit BMP signaling during osteogenic differentiation of MEPM cells in vitro. Although at higher doses dorsomorphin (10–20 μM) inhibits AMPK signaling (Zhou et al., 2001) and mTOR signaling (Vucicevic et al., 2011), it selectively inhibits BMP signaling at lower doses (Yu et al., 2008). Upon 5 μM dorsomorphin treatment, upregulated mRNA expressions of Bmp2 (Figure 6A), Bmp4 (Figure 6B) and Runx2 (Figure 6C) in the Hoxa2^−/− MEPM cells were restored to the wild-type levels. Moreover, increased protein expression of RUNX2 and pSMAD 1/5/8 (Figure 6D) in the Hoxa2^−/− MEPM cells were reduced after dorsomorphin treatment. The increased cell proliferation (Figure 6E) and osteogenic differentiation (Figure 6F) in the Hoxa2^−/− MEPM cells were also reduced after dorsomorphin treatment. These results indicate that the upregulated canonical BMP signaling is functionally responsible for the increased cell proliferation and osteogenic differentiation during palate development in Hoxa2^−/− embryos. Altogether, the findings reveal that Hoxa2 inhibits osteoprogenitor proliferation and commitment, via BMP signaling, to control the spatial and temporal expression of osteoblast markers for proper palatogenesis.

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.