Hoxa2 plays a direct role in murine palate development

Authors

  • Tara M. Smith,

    1. Laboratory of Molecular Biology, College of Pharmacy and Nutrition, University of Saskatchewan, Saskatoon, Saskatchewan, Canada
    Search for more papers by this author
    • T.M. Smith and X. Wang contributed equally to this work.

  • Xia Wang,

    1. Laboratory of Molecular Biology, College of Pharmacy and Nutrition, University of Saskatchewan, Saskatoon, Saskatchewan, Canada
    Search for more papers by this author
    • T.M. Smith and X. Wang contributed equally to this work.

  • Wei Zhang,

    1. Laboratory of Molecular Biology, College of Pharmacy and Nutrition, University of Saskatchewan, Saskatoon, Saskatchewan, Canada
    Search for more papers by this author
  • William Kulyk,

    1. Department of Anatomy and Cell Biology, College of Medicine, University of Saskatchewan, Saskatoon, Saskatchewan, Canada
    Search for more papers by this author
  • Adil J. Nazarali

    Corresponding author
    1. Laboratory of Molecular Biology, College of Pharmacy and Nutrition, University of Saskatchewan, Saskatoon, Saskatchewan, Canada
    • Laboratory of Molecular Biology, College of Pharmacy and Nutrition, 116 Thorvaldson Building, 110 Science Place, University of Saskatchewan, Saskatoon SK, S7N 5C9
    Search for more papers by this author

Abstract

The cleft palate exhibited by Hoxa2 null murine embryos has been described as being secondary to abnormalities of tongue musculature, and Hoxa2 was presumed to not play a direct role in palate development. However, we detected Hoxa2 expression in the developing palate at both the mRNA and protein levels between embryonic day (E) 12.5 and E15.5. Organ cultures of Hoxa2−/− palates maintained in the absence of the tongue showed decreased fusion rates than either Hoxa2+/− or Hoxa2+/+ palate cultures. Knocking down Hoxa2 expression with antisense retroviral constructs resulted in decreased fusion rates than corresponding controls. An overall increase in cell proliferation was observed in Hoxa2 null palates providing a potential mechanism by which Hoxa2 may be affecting palate development. Hoxa2 also repressed the expression of its downstream targets Msx1, Bmp4, Barx1, and Ptx1 within the palate. These results demonstrate the cleft palate phenotype of Hoxa2 null embryos is not solely due to abnormal tongue musculature, and indicate a direct role of Hoxa2 in regulating murine palatogenesis. Developmental Dynamics 238:2364–2373, 2009. © 2009 Wiley-Liss, Inc.

INTRODUCTION

The formation and fusion of the secondary palate in vertebrates is a complex and tightly regulated process. In the mouse, the secondary palate initially develops as two bilateral projections of the maxillary prominences at embryonic day (E) 11.5. The palatal shelves initially grow vertically downward on either side of the tongue. Then, at E14, the tongue flattens and allows the palate shelves to elevate and assume a horizontal position above the tongue. The elevated palate shelves then grow medially toward one another and contact to form the midline epithelial seam (MES). The MES subsequently undergoes a rapid degradation to form a mesenchymal confluence across the secondary palate, at which point palatogenesis is considered complete (Ferguson,1988).

Hoxa2 is a member of a highly conserved family of transcription factors that share a 60 amino acid region known as the homeodomain. The homeodomain is comprised of a helix-turn-helix motif that is capable of recognizing and binding to DNA, and influencing transcription of downstream target genes (Akin and Nazarali,2005). The Hoxa2 gene encodes for encodes a 41-kDa protein (Nazarali et al.,1992; Tan et al.,1992) that acts as the selector gene for second branchial arch patterning (Gendron-Maguire et al.,1993; Rijli et al.,1993; Grammatopoulos et al.,2000; Pasqualetti et al.,2000).

Deletion of the Hoxa2 gene in mice leads to an altered specification of neural crest cells, resulting in numerous craniofacial abnormalities (Gendron-Maguire et al.,1993; Grammatopoulos et al.,2000; Santagati et al.,2005). One of these abnormalities is a cleft secondary palate, which occurs in 81% of Hoxa2−/− embryos (Gendron-Maguire et al.,1993; Rijli et al.,1993; Barrow and Capecchi,1999). A mutation in the HOXA2 gene in humans leads to similar abnormalities including a partial cleft secondary palate (Alasti et al.,2008). Because Hoxa2 is expressed in the second and third branchial arches, but not in the first branchial arch (Prince and Lumsden,1994) from which the palatal shelves later emerge, it was originally presumed that Hoxa2 could not play a direct role in murine palate development. Barrow and Capecchi (1999) hypothesized that the high incidence of cleft secondary palate in Hoxa2 null embryos results from abnormal attachment of the hyoglossus muscle to the greater horn of the hyoid bone—a third branchial arch derivative. They reasoned that this defect would prevent the depression of the lateral edges of the tongue and create a physical barrier between the opposing palate shelves, thereby inhibiting their contact and fusion. However, a subsequent study by Ohnemus et al. (2001) reported that the hyoglossus is always properly inserted in the hyoid of Hoxa2−/− mutant mice regardless of the presence or absence of a cleft palate phenotype. This suggests the possibility that palatal clefting may instead be a primary defect in these embryos. Although Hoxa2 is absent from the first branchial arch at early stages of morphogenesis, this does not preclude it from being expressed at later stages within the growing palatal shelves. Indeed, a previous study in our laboratory provided immunocytochemical data demonstrating Hoxa2 protein expression within both epithelial and mesenchymal cells of the developing murine secondary palate (Nazarali et al.,2000).

Our current study provides further evidence for a direct involvement of Hoxa2 in the development of the murine secondary palate. Using quantitative real-time reverse transcriptase-polymerase chain reaction (RT-PCR), Western blot analysis and immunohistochemistry, we have confirmed and further characterized the endogenous expression of Hoxa2 in the secondary palate throughout the period of murine palatogenesis. In addition, we observed that cell proliferation was elevated in the palatal shelves of Hoxa2−/− mice, indicating that genetic ablation of Hoxa2 function can directly influence palatal growth. Moreover, the palatal tissue of Hoxa2−/− mice demonstrated increased expression of four putative Hoxa2 target genes—Ptx1, Barx1, Msx1, and Bmp4. This provides further evidence that Hoxa2 plays an intrinsic role within the palatal tissues by repressing expression of these downstream target genes.

Using an organ culture model in which the palatal shelves are cultured in the absence of the tongue, we further demonstrate that palates from Hoxa2 null mice display significantly lower fusion rates than palates from heterozygous or wild-type embryos. Similarly, a knock down of Hoxa2 expression in organ cultures of wild-type palates using antisense Hoxa2 retroviral constructs also caused a significant decrease in palatal fusion rates. Collectively, these findings point to a direct role for Hoxa2 in regulating murine palatogenesis.

RESULTS

Endogenous Hoxa2 Expression Within the Developing Palate

To determine whether the Hoxa2 gene may be directly involved in regulating palate development, it was important to demonstrate that Hoxa2 is expressed endogenously within the developing palate. Using quantitative real-time PCR and Western blot analysis, Hoxa2 mRNA and protein, respectively, were detected in the developing palate of wild-type mice at E12.5, E13.5, E14.5, and E15.5 (Figs. 1, 3), which spans the period during which the palatal shelves emerge, elevate, and subsequently fuse. Hoxa2 expression was highest at the two earliest time points, reaching an apparent peak at E13.5. By E14.5, Hoxa2 mRNA expression decreased to a significantly lower level than at either E12.5 or E13.5 (P < 0.05), and it remained low at E15.5. As expected, Hoxa2 transcripts were not detectable in palate tissues of Hoxa2−/− mice (Supp. Fig. S1, which is available online).

Figure 1.

Relative quantitative expression of the Hoxa2 gene throughout murine palatogenesis. Relative Hoxa2 expression levels in wild-type mouse palate shelves were determined by quantitative real time reverse transcriptase-polymerase chain reaction using Taqman primers and probes. Bars represent mean ± SEM, n = 3. *P ≤ 0.05 between bars indicated by brackets.

Immunohistochemical analysis of sections from the anterior, medial, and posterior regions of palates between E12.5 and E15.5 verified the presence of Hoxa2 protein expression within the developing palate. Hoxa2 was expressed in both the epithelium and the mesenchyme at E12.5 (Fig. 2a–c), and at increased levels by E13.5 (Fig. 2e–g). Hoxa2 expression appeared to generally decline throughout the palate by E14.5, with the highest expression observed in the midline epithelial seam (Fig. 2i–k). By E15.5, the contacted palate shelves had fused to mesenchymal confluence and low levels of Hoxa2 expression were persistent throughout the palate (Fig. 2m–o). In contrast, no Hoxa2 protein signal was detected in the palates of Hoxa2 null mice confirming the specificity of the labeling (Figs. 2d,h,l,p, 3).

Figure 2.

Immunofluorescence staining of coronal sections from mouse embryos. Anterior, medial, and posterior palate sections are in the same orientation and from the same embryo. ac, eg, ik, mo: Green nuclear staining represents the expression of Hoxa2 protein at embryonic day (E) 12.5 (a–c), E13.5 (e–g), E14.5 (i–k), and E15.5 (m–o). d,h,l,p: Sections are from the respective anterior palatal region of Hoxa2 null mice and show no staining. Arrow is pointing to the epithelium. M, mesenchyme; T, tongue. Scale bar = 5 μm.

Figure 3.

Western blot analysis of protein extracted from palatal shelves of 12.5-, 13.5-, 14.5-, and 15.5-day-old mouse embryos. Proteins (30 μg) were separated on 10% polyacrylamide-sodium dodecyl sulfate gel and, after transfer to a polyvinylidene difluoride membrane, were probed with Hoxa2 antibody. Lane 1, embryonic day (E) 12.5; Lane 2, E13.5; Lane 3, E14.5; Lane 4, E15.5; Lane 5, E13.5 Hoxa2−/−; and Lane 6, E14.5 Hoxa2−/−.

Altered Cell Proliferation in the Palatal Shelves of Hoxa2−/− Mice

To determine whether a loss of Hoxa2 expression in the palate resulted in intrinsic changes in the growth of the palate primordial, we compared levels of cell proliferation and apoptosis within the anterior, medial, and posterior regions of the palate in wild-type and Hoxa2 null palates at stages E12.5 through E15.5. Compared with normal wild-type Hoxa2+/+ palates, a general trend toward increased cell proliferation rates was noted in the Hoxa2 null palate shelves. In the anterior region, proliferation rates were significantly increased at E12.5, E14.5, and E15.5 (P < 0.001; Fig. 4a); in the medial (Fig. 4b) and posterior (Fig. 4c) regions, rates significantly increased at E12.5 and E14.5 (P < 0.001). A two-factor analysis of variance (ANOVA) of the proliferation data revealed a significant interaction between stage of development and genotype in all three palatal regions (anterior, medial, and posterior).

Figure 4.

Altered cell proliferation rates within Hoxa2 null palates. Cell proliferation rates within the developing palate shown as mean percentage of cells that are 5-bromo-2′-deoxyuridine (BrdU) positive per μm2 ±SEM. AC: Sections from the anterior (A), medial (B), and posterior (C) regions of the palate from wild-type and Hoxa2 null embryos were analyzed at E12.5, E13.5, E14.5, and E15.5 (n = 4). **P ≤ 0.001 between bars indicated by brackets.

Palates were also examined for apoptosis at E12.5, 13.5, and 14.5 in both wild-type and Hoxa2 null embryos. No difference in apoptosis was observed (Supp. Fig. S3).

Altered Gene Expression in the Palatal Tissue of Hoxa2−/− Null Mice

The expression of four genes known to be present in the palate—Ptx1 (Lanctôt et al.,1997), Barx1 (Welsh et al.,2007), Msx1 (Zhang et al.,2002), and Bmp4 (Zhang et al.,2002)—was analyzed at E12.5, E13.5, E14.5, and E15.5 using quantitative real time RT-PCR and Western blot analysis. Our results show Ptx1 is expressed throughout palatogenesis in wild-type embryos (Figs. 5a, 6). Hoxa2 null palates show a significant increase in Ptx1 mRNA expression at E13.5 (P < 0.01), with levels equivalent to wild-type at E12.5, E14.5, and E15.5. Ptx1 protein also exhibited an increase in expression at E13.5 and E14.5 in the Hoxa2 null palate shelves (Fig. 6).

Figure 5.

Altered Ptx1, Msx1, Bmp4, and Barx1 mRNA expression in Hoxa2 null palate shelves. AD: Relative quantitative expression of Ptx1 (A), Msx1 (B), Bmp4 (C), and Barx1 (D) throughout murine palatogenesis in Hoxa2+/+ (wild-type) and Hoxa2−/− (knockout) embryos as determined by quantitative real-time reverse transcriptase-polymerase chain reaction using Taqman primers and probes. Bars represents mean ±SEM, n = 4. A single asterisk represents a significance P ≤ 0.05, and a double asterisk represents a significance P ≤ 0.01 between bars indicated by brackets.

Figure 6.

Altered Msx1, Bmp4, Ptx1, and Barx1 protein expression in Hoxa2 null palate shelves. Protein expression of Msx1, Bmp4, Ptx1, and Barx1 was analyzed by Western blot of palatal protein samples from Hoxa2+/+ (wild-type) and Hoxa2−/− (knockout) at embryonic day (E) 12.5, E13.5, and E14.5.

Our results also demonstrate Msx1 mRNA is expressed at all four ages in the wild-type palate (Fig. 5b). Hoxa2 null palates exhibit a significantly higher expression of Msx1 at E12.5 (P < 0.01), but have levels comparable to wild-type for the remainder of palatogenesis. Protein expression of Msx1 was also visibly increased at E12.5 as determined by Western blot analysis (Fig. 6). Bmp4 is known to be downstream of Msx1 in the palate (Zhang et al.,2002), so an increase in Msx1 expression would be expected to be followed by an increase in Bmp4 expression. As expected, Bmp4 expression increased at E12.5 in Hoxa2 null palates (P < 0.05) with levels comparable to wild-type for the remainder of palate development (Fig. 5c). Bmp4 protein expression increased proportionally in the Hoxa2 null palate shelves compared with wild-type at E13.5 and E14.5, representing a slight delay between increased mRNA message and an increase in the protein signal (Fig. 6).

In addition, an increase in the expression of Barx1 mRNA was observed in the E13.5 Hoxa2 null palates (P < 0.05) when compared with wild-type controls (Fig. 5d). Of interest, a similar increase in Barx1 was also observed at the protein level at E13.5 (Fig. 6).

Effects of Hoxa2 Ablation on Palate Development and Fusion

Histological data comparing wild-type palatal development with that of the Hoxa2 null mice (Supp. Fig. S2) indicates palatal shelves did not elevate in the Hoxa2 knockout mice. An earlier study speculated that the occurrence of cleft palate in Hoxa2−/− mice is due to physical interference from the developing tongue resulting from abnormal attachment of the hyoglossus muscle to the greater horn of the hyoid bone (Barrow and Capecchi,1999). However, we observed that the hyoglossus was properly attached to the hyoid in both wild-type and Hoxa2 null embryos that exhibited cleft palate (Supp. Fig. S4), in agreement with the findings of Ohnemus et al. (2001). This suggests that the high incidence of cleft palate in Hoxa2−/− embryos is not merely a secondary defect due to abnormal positioning of the tongue. To determine whether genetic ablation of Hoxa2 function might disrupt processes of palatal growth and fusion in a more direct intrinsic manner, we used an in vitro organ culture model in which whole palate explants were cultured in the absence of the tongue. Palates isolated from Hoxa2 null embryos exhibited a high incidence of cleft defects even when cultured in the absence of the tongue (Table 1). Palate cultures from Hoxa2−/− embryos had significantly lower fusion rates than cultured palates from either Hoxa2+/− or Hoxa2+/+ animals (P < 0.01). In addition, more palates in Hoxa2−/− cultures either did not contact or contacted without fusion than in cultures of Hoxa2+/− or Hoxa2+/+ palates.

Table 1. Effects of Hoxa2 Gene on Cultured Fetal Mouse Palates in Different Genotypes
Genotype of mouse embryosFrequency of contacted but not fused palatesFrequency of fused palatesFrequency of non-contacted palates
  • *

    P ≤ 0.05,

  • **

    P ≤ 0.01 compared to Hoxa2−/− mutant,

  • ***

    P ≤ 0.01 compared to heterozygous.

  • ****

    P ≤ 0.05,

  • *****

    P ≤ 0.01 compared to wild-type.

Wild-type10.0% (2/20)90.0%** (18/20)0%* (0/20)
Heterozygous7.5%** (3/40)78.7%** (37/47)14.9% (7/47)
Hoxa2−/− Mutant38.5%*** (5/13)44.4%******** (8/18)27.8%**** (5/18)

In other experiments, Hoxa2 expression was knocked down in palate cultures of wild-type embryos by infection with antisense Hoxa2 retroviral particles at three different titers (Table 2) at both E12.5 and E13.5. The palates were assessed after 72 hr (E15.5). The fusion rates of Hoxa2+/+ palates exposed to Hoxa2 antisense retroviral particles were significantly lower than in wild-type palates infected with a control retroviral vector. Moreover, the low fusion rates of wild-type palates subjected to Hoxa2 antisense knock down were comparable to those of cultured palates isolated from Hoxa2−/− mice. Quantitative real-time RT-PCR and Western blot analysis confirmed that the retroviral Hoxa2 antisense infections effectively knocked down Hoxa2 expression in the palate cultures (Supp. Fig. S5).

Table 2. Effects of Retrovirus Infection on Wild-Type Cultured Mouse Palates
Virus and titer (cfu/ml)Frequency of contacted but not fused palatesFrequency of fused palatesFrequency of non-contacted palates
  • *

    P ≤ 0.05,

  • **

    P ≤ 0.01 compared to respective control.

Control   
0.016 × 1055.3% (1/19)81.8% (18/22)13.6% (3/22)
0.08 × 1050.0% (0/19)86.4% (19/22)13.6% (3/22)
0.8 × 1050.0% (0/17)77.3% (17/22)22.7% (5/22)
Antisense Hoxa2   
0.016 × 1050.0% (0/11)45.8%* (11/24)54.2%** (13/24)
0.08 × 10510.0% (1/10)40.9%** (9/22)54.5%** (12/22)
0.8 × 10510.0% (1/10)45.0%* (9/20)50.0% (10/20)

Palatal fusion rates were altered in the Hoxa2 null embryos, but to determine whether loss of the Hoxa2 gene led to inhibition of palatal fusion, E14 palate shelves from wild-type or Hoxa2 null embryos were excised and arranged in pairs with their medial edges in contact and cultured for 72 hr (Shiomi et al.,2006; Nakajima et al.,2007). Fusion occurred with equal success in the Hoxa2 null embryos at approximately 86% (Fig. 7; Table 3).

Figure 7.

Histological analysis for fusion of cultured palate shelves. After 72 hr in culture palate shelves were fixed, serial sectioned, and stained with cresyl violet to determine whether mesenchymal confluence was achieved. Hoxa2 null and wild-type palates were equally capable of fusion.

Table 3. Effects of Hoxa2 Gene on Cultured Fetal Mouse Palates of Different Genotypes
Genotype of embryoNo. of palates that fusedTotal no. of palates cultured% of fused palates
Wild-type121485.7
Heterozygous141593.3
Hoxa2−/− mutant6785.7

DISCUSSION

Our data show that Hoxa2 mRNA and protein are normally expressed throughout the palate in both the epithelium and mesenchyme during the critical periods of palatal shelf outgrowth, elevation, and fusion. Thus, although the Hoxa2 gene is not expressed in the first branchial arch at stages before palatal process formation (Prince and Lumsden,1994), its expression is activated within the palatal processes that subsequently emerge from the maxillary prominences of the first branchial arch. The intrinsic expression of Hoxa2 in the palatal primordia suggests a possible role for Hoxa2 in the regulation of early palatal growth. Moreover, we observed that genetic ablation of Hoxa2 gene function in the Hoxa2−/− homozygous mice led to elevations in the rate of both cell proliferation within the palatal tissue as well as increases in the expression of putative Hoxa2 target genes (Msx1, Ptx1, Bmp4, and Barx1). This suggests that the intrinsic action of Hoxa2 within the palate is to regulate palatal tissue growth and patterns of downstream gene activation.

Barrow and Capecchi (1999) postulated that the cleft palate in Hoxa2 null embryos is a secondary defect due to abnormal hyoglossus insertion to the greater horn of the hyoid bone leading to abnormal positioning of the tongue. However, our data indicate that the hyoglossus inserts properly into the hyoid regardless of the presence or absence of Hoxa2 expression (Supp. Fig. S4), supporting similar findings by Ohnemus et al. (2001). Although other intrinsic or extrinsic factors could influence abnormal positioning of the tongue, our data raise the possibility that the malpositioning of the tongue may not be the primary cause of cleft palate in Hoxa2 null embryos and are consistent with a more direct role for Hoxa2 in palate development. Indeed, when palatal explants of Hoxa2 null mice were cultured in the absence of the tongue, lower palatal fusion rates and lower contact rates were observed in comparison to palates from wild-type embryos. Furthermore, antisense retrovirus-mediated knock down of Hoxa2 expression at E12.5 in wild-type palates resulted in decreased palatal fusion rates by E15.5 relative to controls, with Hoxa2 antisense cultures resembling those of the Hoxa2−/− phenotype. The fusion rates for explant cultures of both Hoxa2−/− palates and palates with retroviral knock down of Hoxa2 expression were in the range of 41–46%, which is considerably higher than the 20% reported for Hoxa2 null embryos in vivo (Gendron-Maguire et al.,1993; Rijli et al.,1993; Barrow and Capecchi,1999). Thus, interactions with the developing tongue could still play a role in contributing to the higher frequency of cleft palate seen in Hoxa2 null embryos in vivo. Compared with wild-type embryos, Hoxa2 null embryos may have disrupted synchrony between the rates of tongue and palate development, causing even a normally developed tongue to become a physical barrier between the rising palatal shelves. To demonstrate whether Hoxa2 null palates were capable of fusion, E14 palatal explants were arranged in pairs with their medial edges in contact and in proper anterior–posterior orientation and allowed to culture for 72 hr (Shiomi et al.,2006; Nakajima et al.,2007). Palatal fusion was not impaired in Hoxa2 null embryos (Fig. 7; Table 3), suggesting palatal growth rate may be affected. Interestingly, we observed an overall increase in the rate of cell proliferation in the palatal shelves of Hoxa2 null mice (Fig. 4), whereas apoptosis was not affected (Supp. Fig. S3). An increase in proliferation leading to a cleft palate phenotype was previously reported in mice lacking the Fgf antagonist Spry2 (Welsh et al.,2007). The incidence of cleft palate penetrance in these animals is 83%, which is very similar to the frequency in Hoxa2 null mice (81%).

Thus, although the in vitro knockout and knockdown organ cultures showed a decrease in the rate of fusion, mutant palate shelves appear to be capable of fusion if they are placed in close opposition as observed with the fusion organ cultures. The fusion organ cultures measure the ability of the palates to fuse with one another and eliminate any confounding issues that arise from the ability of the palate shelves to grow together and contact. The inability of the palate shelves to contact with one another properly (likely due to increased mesenchymal proliferation and altered palatal growth pattern) is believed to be the reason the in vitro knockout and knockdown cultures showed lower fusion rates.

Hoxa2 acts as a transcription factor and would presumably affect palatogenesis through regulating the expression of downstream targets. We have demonstrated that Ptx1, Barx1, Msx1, and Bmp4 genes all exhibit altered expression profiles in the developing palates of Hoxa2 null embryos compared with wild-type embryos, providing further support for a direct and intrinsic role for Hoxa2 in palatogenesis. Through the regulation of these genes, Hoxa2 might act to regulate cell proliferation within the developing palate.

Within the branchial arches, Hoxa2 is known to activate Msx1 expression (Santagati et al.,2005). However, our data suggest that Hoxa2 represses Msx1 expression in the developing palate. A significant increase (P < 0.01) in Msx1 mRNA and protein expression was observed in E12.5 null mutants (Figs. 5b, 6). Cell proliferation in the anterior mesenchyme has been shown to be regulated by Msx1 and its downstream targets including Bmp4 (Zhang, et al.,2002). We found that Bmp4 mRNA expression was increased at E12.5 together with Msx1 (Fig. 5c). Bmp4 protein was also visibly increased at E13.5 and E14.5 and may indicate a delay in protein translation (Fig. 6). Msx1 and Bmp4 work within a feedback loop (Zhang et al.,2002), which might explain why mRNA levels return to wild-type levels by E13.5. Together, these data suggest that Hoxa2 may influence cell proliferation and growth within the palate through regulation of Msx1 and Bmp4.

Barx1 is expressed predominantly in the posterior regions of the palate shelves and is believed to play an important role in regulating cell proliferation in this region (Welsh et al.,2007). Increased expression of Barx1 in the Hoxa2 null palate (Figs. 5d, 6) at E13.5 would, therefore, match the increased level of proliferation observed in the posterior regions of the palate.

Ptx1 is ectopically expressed within the second branchial arch of Hoxa2 null embryos, and transgenic expression of Hoxa2 in the first branchial arch is sufficient to block Ptx1 expression. A partial reversion to the wild-type phenotype was observed in double mutants lacking both Hoxa2 and Ptx1 (Bobola et al.,2003). This suggests Ptx1 is an important downstream target of Hoxa2 in the second branchial arch. Our data show Hoxa2 also acts as a repressor of Ptx1 in the developing palate. Ptx1 expression is significantly increased in the Hoxa2 null palates at E13.5 (Figs. 5a, 6) concurrent with peak Hoxa2 expression in wild-type palates (Figs. 1, 3). This coincident expression could explain why Hoxa2 appears to have a significant effect on Ptx1 expression at this early stage. Thus, Hoxa2 may block expression of Ptx1 in the palate, similar to its role in the second branchial arch.

Our histological data indicates palatal shelves did not elevate in the Hoxa2 knockout mice (Supp. Fig. S2). The mechanism of palate shelf elevation remains largely unclear and poorly defined at the present time. Various mechanisms have been put forward and these have not changed significantly from those reviewed by Lazzaro in1940 and later by Ferguson in1988. General consensus among researchers is that palate shelf elevation is a rapid process triggered by both intrinsic forces within the palate shelf and also influenced by other oral and craniofacial structures. Extracellular matrix was implicated by Lazzaro (1940) and studies have supported this and have identified progressive accumulation of hyaluronan to play an important role in palatal shelf elevation (reviewed in Ferguson,1988). Collagen fibers, epithelial covering, and polarized alignment of mesenchymal cells have also been postulated to play a role in palatal shelf elevation (Ferguson,1988). The neurotransmitter γ-aminobutyric acid (GABA) appears to play a role in palatal shelf elevation because it is reported that GABA agonists generate a cleft palate by inhibiting palatal shelf elevation, whereas GABA antagonist stimulate this process (Miller and Becker,1975; Wee and Zimmerman,1983). GABA as well as glutamic acid decarboxylase 67 (Gad 67), an enzyme important in GABA biosynthesis, is present in the palatal shelf (Wee et al.,1986; Asada et al.,1997; Hagiwara et al.,2003). It is possible that loss of Hoxa2 which alters mesenchymal proliferation patterns affects GABA neurotransmitter signaling and changes the turnover of extracellular matrix which could contribute to delayed lifting of the shelves; however, this will have to await further studies.

In summary, we have demonstrated endogenous Hoxa2 expression within the developing murine palate and have shown Hoxa2 expression is necessary for normal palatogenesis. Although Hoxa2 is not initially expressed within the first branchial arch, its expression appears to be switched on later in development within its derivative, the palate, where Hoxa2 plays intrinsic roles in regulating both cell proliferation and gene expression during murine palatogenesis. Absence of Hoxa2 did not inhibit the palatal fusion process; it does, however, induce an increase in mesenchymal cell proliferation, which will alter the palatal growth pattern and likely affect palatal elevation, preventing the palatal shelves from coming together for fusion.

EXPERIMENTAL PROCEDURES

Hoxa2 Transgenic Mice

Hoxa2+/+, Hoxa2+/−, and Hoxa2−/− C57 black mice were obtained by timed heterozygous matings. They were staged according to Kaufman (1992) and were considered E0 days pregnant on the day the vaginal plug was found. Hoxa2+/+, Hoxa2+/−, and Hoxa2−/− mice genotypes were confirmed by PCR analysis (Gendron-Maguire et al.,1993). CD-1 mice were used for wild-type palatal organ cultures treated with Hoxa2 antisense retroviral particles.

Whole Palatal Organ Culture

Whole mouse palatal organ cultures were established according to the protocol outlined by Abbott et al. (1999) with modifications as indicated below. Pregnant females at embryonic stage E12.5 were anesthetized by halothane (MTC Pharmaceuticals). Fetuses were aseptically removed from the uterus in Hanks' balanced solution (Sigma). The maxillary regions were dissected by removing the mandible and tongue as well as the brain and the spinal cord from the posterior region of the palate. Palates were placed, intact, in sterile 60-ml glass bottles containing 10 ml of Richter's Improved Zinc MEM Option and F12 Nutrient Medium (Invitrogen) (1:1), supplemented with 1% fetal bovine serum, 6 mg/ml bovine serum albumin, 10 μg/ml transferrin, 10 ng/ml selenium, 50 μg/ml sodium ascorbate, 2.4 mg/ml glucose, 0.6 mg/ml L-glutamine, 50 μg/ml streptomycin, and 50 units/ml penicillin (Sigma). Bottles were flushed with 50% O2, 45% N2, and 5% CO2 for 2 min and then incubated at 37°C, circulated at 12–15 rpm. Every 24 hr the media was changed, flushed with the same gas mixture, and cultures incubated as indicated above. After 72 hr (E15.5), the palates were assessed and relevant parameters such as frequency of contacted but not fused, frequency of noncontacted palates, and frequency of fused palates were measured with a micrometer under a dissecting microscope.

Cell Proliferation Assay

Cell proliferation was assessed by intraperitoneal injection of timed-pregnant mice with 5-bromo-2′-deoxyuridine (BrdU) at a concentration of 100 mg/kg body weight. BrdU is a thymidine analogue that is taken up and incorporated into the DNA of proliferating cells, including those in the developing embryos. One and a half hours after injection, mice were killed, and the harvested embryos fixed and sectioned as described in the Immunohistochemistry section. Sections were immunohistochemically stained with a monoclonal anti-BrdU antibody (Sigma) and the Mouse on Mouse staining kit (Vector Laboratories) to detect proliferating cells, using a modified procedure as follows: briefly, the sections were pretreated by rehydrating the tissue sections in phosphate buffered serum (PBS) 0.1% Triton X-100 before being exposed to 1 N HCl on ice for 10 min and then 2 N HCl for 10 min at room temperature followed by a 40 min incubation at 37°C. Sections were neutralized with 0.1 M sodium borate, followed by three 5-min washes in PBS-1% Triton-X100. Endogenous biotin and avidin was blocked using the Avidin-Biotin blocking kit (Vector Laboratories) as per the manufacturer's protocol. After incubating the section in the blocking reagent overnight at 4°C, the remainder of the procedure was as per the manufacturer's protocol. Finally, sections were counter stained with Hoescht and mounted in Prolong (Molecular Probes). Sections from the anterior, medial and posterior regions of wild-type and Hoxa2 null embryos at E12.5, E13.5, E14.5, and E15.5 were analyzed. The mitotic index was only calculated for the mesenchyme region of the palate due to an inability to gain accurate total cell counts from the epithelium. The percentage of total cells that were BrdU positive per unit area was used as the mitotic index.

TUNEL Assay

Cell survival and apoptosis was measured using the DeadEnd Fluorometric TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling) System (Promega) as per the manufacturer's protocol. Tissue from the anterior, medial, and posterior regions of wild-type and Hoxa2 null embryos at E12.5, E13.5, and E14.5 were examined for apoptotic cell death.

RNA Isolation and Reverse Transcription

Total RNA was isolated from excised wild-type and Hoxa2 null embryo palate shelves using the RNeasy Protect Mini Kit (Qiagen) as per the manufacturer's protocol. RNA concentration was determined by optical density. First-strand DNA synthesis was performed using the SuperScript first-strand synthesis system for RT-PCR (Invitrogen) and random decamer primers, as per the manufacturer's protocol. The final concentration of RNA for all RT reactions was 20 ng/μl.

Quantitative Real Time RT-PCR

Gene expression was quantified using the Taqman primers and labeled probe system and the ABI 7300 (Applied Biosystems). Wild-type and Hoxa2 mutant embryo palates at E12.5, E13.5, E14.5, and E15.5 were tested for Msx1, Bmp4, Barx1, and Ptx1; only wild-type tissue was tested for Hoxa2. All reactions were performed using the Taqman Universal Master Mix (2X), FAM-labeled Taqman Gene Expression assays for gene of interest, VIC-labeled Taqman Endogenous Control β-actin, and 10 ng of cDNA. All reactions were run in replicates of 4, with n = 4. Thermocycling parameters were as follows: 2 min at 50°C, 10 min at 95°C, 40 cycles of 15 s at 95°C, plus 70 s at 60°C.

Immunohistochemistry

The Hoxa2 rabbit polyclonal antibody was generated using oligopeptide SPLTSNEKNLKHFQHQS (Hao et al.,1999; Nazarali et al.,2000). Time staged embryos were fixed in 4% paraformaldehyde followed by immersion in 20% sucrose in PBS. Frozen embryos in embedding medium were sectioned (8 μm thick), and tissue sections were collected on gelatin-coated coverslips. Coverslips were dried for 2 hr at room temperature before being subjected to immunohistochemical analysis with TSA #22 (Invitrogen) using the following modified procedure. Coverslips with tissue sections were incubated twice in PBS, 5 min each time, followed by a 1-hr incubation in a 3% hydrogen peroxide solution. This was followed by two 5-min washes in PBS and then 30 min in the manufacturer provided block solution. After blocking, sections were incubated overnight at 4°C with the Hoxa2 specific antibody at a dilution of 1:5,000 in the block solution. Sections were rinsed in PBS twice (5 min each) and then incubated in biotinylated anti-rabbit secondary antibody (Vector Laboratories) at 1:100 dilution for 1 hr at room temperature. Sections were then washed twice (5 min each) in PBS, followed by a 30-min incubation in the avidin-horseradish peroxidase antibody at 1:100 dilution. The sections were washed twice in PBS for 5 min each, then incubated in the Alexa Fluor 488-conjugated tyramide diluted to 1:200 in the amplification buffer for 10 min, and then counterstained with Hoechst before being mounted in Prolong (Molecular Probes).

Western Blot Analysis

Protein was isolated from wild-type and Hoxa2 null mutant embryonic palatal shelves by lysis in RIPA buffer (150 mM NaCl, 10 mM Tris, 0.1% sodium dodecyl sulfate [SDS], 1% Triton X-100, 1% deoxycholate, 5 mM ethylenediaminetetraacetic acid). Protein was quantified using the Bio-Rad Detergent Compatible Protein Assay kit (Bio-Rad) to ensure equal loading on gel. Samples were then boiled for 20 min with loading buffer and loaded on a 10% polyacrylamide-SDS gel. After separation the proteins were transferred to a PolyScreen PVDF membrane, which was blocked overnight in 3% skim milk in PBS (SM-PBS) at 4°C. The membrane was then exposed to primary antibody for 1 hr at 28°C (Hoxa2, 1:1,000; Ptx1, goat polyclonal, Santa Cruz, 1:100; Msx1, rabbit polyclonal, Developmental Studies Hybridoma Bank, 1:500; Bmp4, goat polyclonal, Santa Cruz 1:200; Barx1, goat polyclonal, Santa Cruz, 1:200). This was followed by 3 consecutive washes of 10 min each in PBS with 0.08% Tween-20. After the washes were complete, the membrane was incubated with the secondary antibody, goat anti-rabbit IgG horseradish peroxidase conjugate (Bio-Rad) in SM-PBS at a dilution of 1:3,000 for 1 hr at 28°C. After 3 washes of 40 min each in PBS-Tween-20 the membrane was exposed with a chemiluminescent reagent (DuPont NEN) and signal was detected by exposing to X-ray film. The membrane was then washed overnight in PBS at 4°C before being incubated with the anti–β-actin at a dilution of 1:1,000, and incubated for 1 hr at room temperature, followed by 3 washes in PBS with 0.08% Tween-20. After the washes, the membrane was exposed to the anti-mouse IgM horseradish peroxidase conjugate (Bio-Rad) in SM-PBS at a dilution of 1:1,500, washed and membrane exposed to X-ray film as indicated above.

Hoxa2-Antisense Retroviral Particles

Hoxa2 cDNA was restriction digested from the recombinant plasmid pRSVD0C-Hoxa2 with HindIII and XhoI (Tan et al.,1992) and used as a template to PCR amplify the Hoxa2 antisense sequence using the following primers: forward 5-ggctcgagccatgaattacgaatttgagcg-3, reverse 5-ggaagcttttagtaattcagatgctgtaggtcg-3. The Hoxa2 antisense DNA was then cloned into a retroviral vector pLEGFP-C1 (BD Biosciences Clontech) at the same two restriction enzyme sites; the sequence of the pLEGFP-Hoxa2 antisense construct was confirmed with an ABI PRISM system. The EcoPack2-293 packaging cell line (BD Biosciences Clontech) was cultured as per the manufacturer's instructions, and the recombinant retrovirus expressing Hoxa2 antisense was generated by transfecting the recombinant construct pLEGFP-Hoxa2 antisense into the packaging cells using SuperFect reagent (Qiagen). Desired transfection efficiency was verified by EGFP protein expression. Subsequently, a selection of stable retrovirus producer cells were harvested from a medium containing 400 μg/ml of G418 (Sigma) after 14 days. The retrovirus producer cells were then cultured to 90% confluence, and the supernatant containing retroviral particles collected. Viral stocks were centrifuged and filtered through a 0.45-μm filter and stored at −80°C. Viral titer was determined by infecting NIH-3T3 cells.

Transduction of Palatal Organ Culture

Viral stocks were added to a palatal organ culture medium to reach final concentrations of 105, 104, and 103 cfu/ml at both E12.5 and E13.5. The palates were assessed after 72 hr (E15.5) and measured for the same parameters as the whole palatal organ cultures above. The desired transduction effect was verified by observing EGFP expression using confocal microscopy. The ability of the antisense retroviral construct to knock down Hoxa2 expression was confirmed by quantitative real time RT-PCR and Western blot analysis.

Fusion Cultures

Pregnant CD-1 mice (E14) were killed by cervical dislocation after anesthesia with Halothane (MTC Pharmaceuticals) and the embryos were aseptically removed in Hanks' Balanced Salts solution (Sigma-Aldrich). The palate shelves were aseptically dissected and palate shelf cultures set up as previously described (Shiomi et al.,2006; Nakajima et al.,2007) with two modifications: the addition of 1 M HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) to a final concentration of 20 mM and addition of a penicillin–streptomycin antibiotic solution to a final concentration of 0.1% (Sigma-Aldrich). In brief, the E14.0 palate shelves were excised and arranged in pairs with their medial edges in contact and in proper anterior–posterior orientation on Nucleopore Polycarbonate Track-Etch filters (8.0 μm pore; Whatman, Inc.). Two filters were floated in a 35-mm tissue culture dish containing 1.4 ml of BGJb medium (Invitrogen), supplemented as indicated above, and incubated with 5% CO2 at 37°C for 72 hr. After 72 hr incubation, cultures were observed under a light microscope to determine whether the palate shelves fused. Cultured palatal shelf explants were sectioned in a coronal plane and stained with cresyl violet to ensure that the midline epithelial seam was degraded. Fusion rates were compared between wild-type, heterozygous, and Hoxa2 null mutant embryos.

Histological Analysis

The Hoxa2−/− and wild-type newborn palate specimens were fixed in Bouin's solution, embedded in paraffin, then 4-μm-thick sections were stained with hematoxylin and eosin.

Statistical Analysis

Statistical analysis on quantitative real-time RT-PCR data was performed using a Student's t-test between wild-type and Hoxa2 null palate shelves at each stage. A Chi-square paired-sample test was applied to compare the frequencies of contacted and fused palates. Cell proliferation data were analyzed statistically by comparing mean values using two-way ANOVA and Bonferroni post-tests for comparing groups. A significant P value of less than or equal to 0.05 was adopted for all comparisons.

Acknowledgements

We thank Dr. Filippo Rijli for his review of our manuscript. A.J.N. was funded by a grant from the Natural Sciences and Engineering Research Council of Canada.

Ancillary