|Home | About | Journals | Submit | Contact Us | Français|
Craniofacial development is characterized by reciprocal interactions between neural crest cells and neighboring cell populations of ectodermal, endodermal and mesodermal origin. Various genetic pathways play critical roles in coordinating the development of cranial structures by modulating the growth, survival and differentiation of neural crest cells. However, the regulation of these pathways, particularly at the epigenomic level, remains poorly understood. Using murine genetics, we show that neural crest cells exhibit a requirement for the class I histone deacetylase Hdac3 during craniofacial development. Mice in which Hdac3 has been conditionally deleted in neural crest demonstrate fully penetrant craniofacial abnormalities, including microcephaly, cleft secondary palate and dental hypoplasia. Consistent with these abnormalities, we observe dysregulation of cell cycle genes and increased apoptosis in neural crest structures in mutant embryos. Known regulators of cell cycle progression and apoptosis in neural crest, including Msx1, Msx2 and Bmp4, are upregulated in Hdac3-deficient cranial mesenchyme. These results suggest that Hdac3 serves as a critical regulator of craniofacial morphogenesis, in part by repressing core apoptotic pathways in cranial neural crest cells.
Vertebrates are unique among metazoans in their reliance on neural crest cells to form a wide array of head structures. These neural crest-derived craniofacial components are analogous to mesodermally- and ectodermally-derived structures in invertebrates (Gans and Northcutt, 1983). In midgestation in the mouse, neural crest cells populate the pharyngeal arches – a series of paired outpouchings that flank the developing pharynx, as well as the frontonasal prominence, an area that eventually gives rise to the nose and forehead. The first pharyngeal arch, which is the largest and most rostral of the arches, contains the neural crest cells that will form structures in the face and neck (Alappat et al., 2003). Driven by the proliferation and continued influx of migrating neural crest cells, two distinct outgrowths arising from the first arch, known as the maxillary and mandibular prominences, grow ventrally and flank the developing oropharynx, eventually fusing at the ventral midline in what becomes the face. Failure of neural crest cells to migrate or proliferate appropriately can result in hypoplasia of these structures and the absence of fusion, which can manifest as abnormal facies and clefting (Ito et al., 2003; Vallejo-Illarramendi et al., 2009).
In addition to appropriate proliferation and migration, neural crest patterning depends upon carefully controlled apoptosis of premigratory crest cells. In the developing hindbrain, a premigratory neural crest population arises in each of eight rhombomeres. However, the majority of neural crest cells in rhombomeres 3 and 5 undergo apoptosis, and the few surviving cells make small contributions to the discrete streams of migrating neural crest that populate and pattern the developing pharyngeal arches (Graham et al., 1993; Sechrist et al., 1993; Birgbauer et al., 1995; Köntges et al., 1996; Ellies et al., 2002). Loss of this selective apoptosis through ablation of rhombomere 4 causes increased survival and migration of crest cells from rhombomeres 3 and 5, resulting in ectopic muscle attachment sites on the developing mandible (Ellies et al., 2002). A signal for selective apoptosis of neural crest cells specifically in these two rhombomeres is induced by increased expression of Msx2 (Graham et al., 1994). A related member of this muscle segment homeobox family of genes, Msx1, is also a proapoptotic factor in neural crest cells and controls programmed cell death by regulating several caspases in the apoptotic pathway (Tríbulo et al., 2004).
Several human syndromes have implicated MSX genes in craniofacial development. Some patients with autosomal-dominant Boston-type craniosynostosis harbor a missense mutation in MSX2 and present with a variety of malformations including abnormal skull shape and cleft palate (Warman et al., 1993; Jabs et al., 1993). Additionally, deletion of the MSX1 gene in patients with Wolf-Hirschhorn syndrome manifests with a spectrum of ear, tooth, and skull defects (Ivens et al., 1990). These disorders highlight the critical role that Msx factors play in regulating how the neural crest contributes to the derivative structures that make up the calvaria and face.
Murine genetics have also proven invaluable in deciphering the genetic programs that coordinate the proliferation, migration and apoptosis of cranial neural crest cells. For instance, models of conditional deletion and overexpression of Msx1 and Msx2 have further delineated roles for these molecules in various aspects of craniofacial development. Msx1 has been studied extensively in the developing tooth and is highly expressed in neural crest-derived dental mesenchyme, where it is required for proper condensation and development of the molar tooth germ beyond the bud stage (Satokata et al., 1994; Chen et al., 1996). Deletion of the Msx1 gene in mice results in reduced Bmp4 expression in the dental mesenchyme, supporting the well-studied interaction of Msx genes and Bmp4 (Graham et al., 1994; Chen et al., 1996; Bei et al., 2000). In addition to its role in neural crest patterning and apoptosis in the developing hindbrain (Graham et al., 1993, 1994), Msx2 also regulates osteogenesis and functions with Msx1 to control cranial neural crest differentiation into bones of the calvaria (Han et al., 2007; Roybal et al., 2010). Interestingly, both overexpression and inactivation of Msx2 causes defects in calvaria and tooth development (Dodig et al., 1999; Satokata et al., 2000), suggesting that the Msx1/Msx2 apoptotic pathway must be exquisitely regulated during craniofacial development.
While neural crest proliferation and localized apoptosis are critical for normal craniofacial morphogenesis, little is known about how the pathways controlling these processes are regulated at the epigenetic level. Here, we demonstrate that the class I histone deacetylase Hdac3 regulates genetic programs involved in murine craniofacial development. Global deletion of Hdac3 results in lethality prior to E9.5 (Bhaskara et al., 2008). Conditional genetic deletion of Hdac3 in neural crest results in cleft palate, hypoplastic teeth and a variety of defects in calvarial structures of neural crest origin. Loss of Hdac3 leads to upregulation of established mediators of neural crest apoptosis, including Msx1, Msx2 and Bmp4. These results suggest that Hdac3-mediated repression of Msx signaling plays a critical role in craniofacial development.
Wnt1-Cre, Pax3Cre, Hdac3flox, Z/EG and p21 null mice were maintained on mixed CD1/B6/129 genetic backgrounds separated by 4–8 generations of interbreeding from pure parental strains (Engleka et al., 2005; Jiang et al., 2000; Mullican et al., 2011; Novak et al., 2000). Mice were genotyped using previously described Cre-specific PCR primers (5′-TGC CAC GAC CAA GTG ACA GC-3′, 5′-CCA GGT TAC GGA TAT AGT TCA TG-3′) (Heidt and Black, 2005), and primers designed to distinguish between the control and floxed Hdac3 allele (5′-GCA GTG GTG GTG AAT GGC TT-3′, 5′-CCT GTG TAA CGG GAG CAG AAC TC-3′). Genotyping for the Z/EG transgene was performed by X-Gal staining tail samples. Littermate embryos were analyzed in all experiments unless otherwise noted. The University of Pennsylvania Institutional Animal Care and Use Committee approved all animal protocols.
These techniques were performed on paraformaldehyde-fixed, paraffin embedded slides as previously described (High et al., 2008). Embryos were dissected in cold PBS, fixed overnight in 2% paraformaldehyde, dehydrated into 100% ethanol, embedded in paraffin and sectioned. H&E and Goldner’s trichrome staining were performed using standard procedures. Primary antibodies used for immunohistochemistry were anti-GFP goat polyclonal (Abcam ab6673, 1:100), anti-GFP rabbit polyclonal (Invitrogen A-11122, 1:200), anti-Runx2 rabbit polyclonal (Santa Cruz sc-10758, 1:20), anti-phospho-histone H3 rabbit polyclonal (Cell Signaling #9701, 1:50) and anti-Hdac3 rabbit polyclonal (Santa Cruz sc-11417x, 1:10). Radioactive in situ hybridization for Msx1,Msx2 and PlexinA2 was performed as previously described (Engleka et al., 2005). Msx1 probe corresponds to RefSeq #NM_010835 (bp 862-1741), Msx2 to #NM_013601 (bp 159-1143) and PlexinA2 as previously described (Brown et al., 2001). TUNEL staining was performed as previously described (Jain et al., 2011). All control and mutant histological and immunohistochemical images shown for comparison were taken at the same exposure and contrast settings, using NIS Elements software. Pharyngeal arch size was quantified using ImageJ software by counting number of pixels in at least three serial sections from control and mutant frontal sections. All quantification was done in a blinded manner.
Cell number, proliferation and apoptosis were quantified by manually counting nuclei, phospho-histone H3-, or TUNEL-positive cells, respectively, in adjacent sections of the anatomically-defined palatal shelf region in a blinded manner.
Samples were dehydrated into 100% methanol, embedded in 1% low-melt agarose, cleared overnight in 1:2 (v/v) benzyl alcohol and benzyl benzoate, and scanned using the Bioptonics OPT scanner (3001M) (Sharpe et al., 2002). Image stacks were reconstructed using OsiriX software. Image contrast was optimized to show anatomic detail.
E12.5 anterior craniofacial mesenchyme was microdissected and dissociated to single cell suspension with 0.25% Trypsin-EDTA. Cells were incubated at 37 degrees C in 5% CO2 for 16 hours before transfection with 50nM control (Santa Cruz sc-37007), Msx1 (Dharmacon #M-044089-01) or Msx2 siRNA (Dharmacon #M-047845-01) using Lipofectamine RNAiMax (Invitrogen #13778030). 24 hours after transfection, cells were fixed with 4% paraformaldehyde (10 min, room temperature) and permeabilized with 0.2% Triton X-100. Immunocytochemistry was performed with anti-phospho-histone H3 rabbit polyclonal (Cell Signaling #9701, 1:100). Cell number and proliferation were quantified by manually counting nuclei and phospho-histone H3-positive cells in a blinded manner.
Neonatal mice were eviscerated and skinned after being soaked in room temperature water for 3 hours followed by a 1 minute heat shock at 65 degrees C to loosen the skin. The samples were then fixed for 3 days in 100% ethanol at room temperature. Samples were stained for 2 days at room temperature with Alcian blue solution (0.3mg/ml Alcian Blue, 80% ethanol, 20% acetic acid) to visualize cartilage, and then rinsed and postfixed in 100% ethanol at room temperature, overnight. Samples were then incubated for 24 hours with Alizarin Red solution (0.065mg/ml Alizarin Red S in 0.5% KOH) to visualize bone, followed by incubation in 0.5% KOH until soft tissues were mostly digested. The 0.5% KOH solution was replaced with 25% glycerol in water (added very slowly to sample) and incubated at room temperature until tissues cleared.
For E12.5 cranial mesenchyme expression profiling, Wnt1-Cre; Hdac3f/f and littermate Cre-negative control embryos were microdissected with tungsten needles in cold PBS. RNA was obtained using the Qiagen RNeasy spin column, with on-column DNAse I digestion. Complementary DNA (cDNA) was synthesized according to kit instructions with the Superscript III system (Invitrogen). Quantitative RT-PCR was performed in triplicate using SYBR Green (Applied Biosystems). Gapdh was used as a reference control gene. Quantitative RT-PCR primers (Table S1) were designed using IDT software.
The student’s two-tailed t test was used to ascertain differences between groups. A p-value of less than 0.05 was considered significant.
In order to explore the epigenomic regulation of craniofacial development, we deleted the class I histone deacetylase Hdac3 in premigratory neural crest cells using the Wnt1-Cre transgene and a floxed Hdac3 allele (Hdac3f) (Jiang et al., 2000; Mullican et al., 2011). Using the Z/EG reporter allele (Novak et al., 2000), we observed that both Wnt1-Cre; Hdac3f/f (which we have termed Hdac3Wnt1NCKO); Z/EG and Wnt1-Cre; Z/EG control embryos demonstrate intact migration of neural crest cells into the developing face and pharyngeal arch region at E9.5 (Fig. 1A and Supplemental Fig. 1). Immunohistochemistry of staged embryos demonstrates that Hdac3 is widely expressed in the developing head at E9.5 and E10.5, including in neural crest, ectoderm and endoderm (Fig. 1B and data not shown). Expression of PlexinA2, a neural crest marker (Brown et al., 2001), appears unaltered in Hdac3 mutant embryos at E9.5 (Supplemental Fig. 1), further indicating that neural crest migration is normal at this stage. In Hdac3Wnt1NCKO embryos, Hdac3 protein expression is lost in neural crest-derived craniofacial mesenchyme, while expression is retained in ectoderm and endoderm (Fig. 1B). In the absence of neural crest expression of Hdac3, the maxillary and mandibular prominences of the first pharyngeal arch demonstrate mild hypoplasia as early as E10.5 (Figs. 1B and C). This is in contrast to the caudal pharyngeal arches, which demonstrate no hypoplasia at E10.5 in the absence of Hdac3 expression (Singh et al., 2011).
Embryos in which Hdac3 is deleted in neural crest cells are found at expected Mendelian ratios in late gestation and are viable until birth, but uniformly succumb at P0 (Singh et al., 2011). Hdac3Wnt1NCKO mice are born with microcephaly, micrognathia, a shortened snout and eyelid closure defects, with preservation of body morphology (Fig. 2A). Optical projection tomography and histology of Hdac3Wnt1NCKO P0 heads reveals cleft palate, which is also apparent in late gestation embryos (Figs. 2B and C). The palate defect is characterized by a large posterior cleft, without additional facial clefting (Fig. 2 B). Cleft palate is fatal in perinatal mice; afflicted pups are unable to generate suction and nurse, and subsequently die at P0 from dehydration and accumulation of air in the digestive tract (Condie et al., 1997; Qiu et al., 1995). Consistent with these reports, Hdac3Wnt1NCKO pups are unable to feed, as evidenced by the lack of a milk spot (Fig. 2A). The gross craniofacial defects observed in Hdac3Wnt1NCKO mice are fully penetrant and are summarized in Table 1. We observed similar craniofacial abnormalities using a second neural crest driver, Pax3Cre (Engleka et al., 2005); these abnormalities are also fully penetrant both at P0 and in late gestation (Supplemental Fig. 2 and unpublished observations).
In the adult vertebrate, osteoblasts are among the most abundant cell types generated by cranial neural crest. These neural crest-derived osteoblasts contribute to the bones of the face, the skull base and the entirety of the calvaria, with the exception of the parietal bone (Santagati and Rijli, 2003). Hdac3 has previously been implicated in multiple stages of osteoblast differentiation and maturation. In committed, undifferentiated osteoblasts that express the Osx-Cre transgene, deletion of Hdac3 leads to subtle abnormalities in calvarial osteoblast differentiation, and progressive abnormalities in trabecular bones that lead to perinatal runting and death early in adulthood (Razidlo et al., 2010).
Consistent with a pro-osteogenic role for Hdac3 at the earliest stages of bone development in neural crest, Hdac3Wnt1NCKO mice - in which Hdac3 is deleted prior to osteoblast specification (before the onset of Osx-Cre expression) - exhibit severe bone defects in the calvaria and viscerocranium (Fig. 3). Alcian blue/alizarin red staining of the late embryonic and perinatal mice reveals decreased ossification of the calvaria, particularly in the region of the frontal bone, where ossified bone is nearly undetectable by alizarin red staining (Fig. 3A). The open frontal fontanel leads to hemorrhage in some newborn pups following parturition (Fig. 2B). The parietal bone develops normally in Hdac3Wnt1NCKO mutants, consistent with its mesodermal origin (Fig. 3A). In mutant embryos, ossification abnormalities of the viscerocranium were also identified in the mandible, skull base and tympanic ring (Fig. 3A).
Additional bone abnormalities in E17.5 mutant embryos are detectable by Goldner’s trichrome staining. Neural crest-derived calvarial bones, including the frontal bone, are thin and show minimal mineralization (Fig. 3B). Additionally, at variable penetrance, absence of ossification in the sphenoid bone plate results in encephalocoele, in which the brain herniates into the nasal sinuses (Fig. 3B). The bone defects observed in Hdac3Wnt1NCKO mice are summarized in Table 1.
To establish whether loss of Hdac3 in the developing calvaria affects migration and survival of neural crest, or differentiation of osteogenic precursors, we fate mapped cranial crest cells to the E12.5 calvaria using the Z/EG reporter, measured apoptosis with TUNEL staining and evaluated pre-osteoblast differentiation by Runx2 immunohistochemistry (Supplemental Fig. 3). We observed that mutant cranial crest exhibit normal migration to the calvaria and normal expression of the critical ossification regulator Runx2 at the frontal bone primordium (Ishii et al., 2003), but demonstrate increased apoptosis compared to controls. These observations suggest that the bone abnormalities in the Hdac3Wnt1NCKO mice are associated with decreased survival of neural crest cells, and are unlikely to be due to a primary abnormality in differentiation or migration.
Seminal work in amphibians and avian models established that the cranial neural crest makes up a significant portion of the developing tooth, specifically the dental mesenchyme (LeDouarin, 1982; reviewed by Maas and Bei, 1997). Further studies in mouse demonstrate that patterning of the developing molar tooth germ is influenced by the interaction of the epithelium and mesenchyme. Recombination experiments with neural crest and epithelium revealed that early (E9.5-12.0) oral epithelium can signal to non-dental mesenchyme to form teeth; however, later in development, the dental mesenchyme possesses odontogenic potential to induce non-oral epithelium to become enamel (Kollar et al., 1969; Mina et al., 1987; Lumsden 1988).
In murine tooth development, signals from the epithelium initiate condensation of neighboring mesenchyme composed of neural crest cells at E11.5 (Chen et al., 2000). As epithelial cells invaginate, they envelop neural crest cells, which will eventually form the pulp of the tooth (Fig. 4A). In Hdac3Wnt1NCKO embryos, the early stages of odontogenesis proceed normally, with tooth bud morphology at E15.5 appearing similar to littermate controls through the bud and cap stages (Figs. 4 A–D). However, expansion of the neural crest-derived mesenchyme does not occur between E15.5-E17.5, resulting in hypoplastic teeth with the absence of normal pitting (Figs. 2C, 4E and F). The tooth bud hypoplasia in mutant embryos parallels the hypoplasia observed in other neural crest-derived structures in late gestation (Fig. 2).
In order to further delineate the mechanism by which deletion of Hdac3 results in hypoplasia of neural crest structures, we performed a histological analysis of staged control versus Hdac3Wnt1NCKO embryos. In normal palatogenesis, the anterior palate is formed by fusion of the maxillary prominence with the frontonasal prominence. Hdac3Wnt1NCKO embryos exhibit posterior cleft palate, suggesting that secondary palatogenesis is disrupted (Fig. 2B). The posterior aspects of the hard palate (referred to as the secondary palate) are formed from two outgrowths of neural crest-derived mesenchyme that lie on either side of the stomodeum. These outgrowths - the palatal shelves - are detectable at E11.5 in the mouse at both anterior and posterior levels (Figs. 5A and A′). Between E11.5 and E13.5, the palatal shelves, driven by neural crest cell proliferation, expand towards the mandible (Figs. 5A, A′, B and B′). Between E13.5 and E14.5 the palatal shelves elevate relative to the tongue and fuse in the midline with the nasal septum (Figs. 5C and C′). This newly formed structure subsequently ossifies, giving rise to the mature secondary palate (Figs. 5D and D′).
Disruption of any of the stages of secondary palatogenesis - palatal shelf formation, expansion, elevation, fusion or ossification – can lead to cleft palate in mice, and similar mechanisms are believed to contribute to secondary cleft palate in humans (He et al., 2011; Ito et al., 2003; Richarte et al., 2007; Vallejo-Illarramendi et al., 2009; Wu et al., 2008). Histological analysis of Hdac3Wnt1NCKO embryos reveals that the palatal shelves form appropriately but are hypoplastic at E12.5 (Figs. 5 E, E′, F and F′). Despite appropriate elevation of the medial aspect of the palatal shelves by E14.5, the palatal shelves do not meet at the midline (Figs. 5G and G′). These results suggest that cleft palate in Hdac3Wnt1NCKO embryos is due to a failure of palatal shelf expansion.
Expansion of the palatal shelves is dependent upon proliferation and survival of the neural crest cells that make up the palatal shelf mesenchyme (Ito et al., 2003). In E12.5 mutant versus control palatal shelves that were matched for surrounding anatomical landmarks, we detected a significant increase in apoptosis and a trend towards decreased proliferation, as determined by TUNEL and phospho-histone H3 staining, respectively (Fig. 5I). Increased TUNEL staining is also visible throughout additional cranial neural crest-derived structures outside of the palatal shelf region including the tooth bud and calvaria, suggesting that hypoplasia of these structures is also mediated by increased apoptosis (Figs. 5I, ,6E6E and Supplemental Fig. 3).
To discern the nature of the cell cycle dysregulation in Hdac3Wnt1NCKO cranial mesenchyme, we performed expression profiling by quantitative RT-PCR of microdissected anterior cranial tissue of E12.5 mutant embryos and littermate controls (Fig. 6A). While Hdac3 expression in mutant tissue is significantly downregulated, expression of the other class I Hdacs is unchanged (Fig. 6B). Strikingly, the expression of multiple cell cycle regulators is altered in Hdac3Wnt1NCKO cranial mesenchyme. Cdkn1a (p21), Trp53 (p53), Cdkn1c (p57), Ccnd1 (Cyclin D1), Ccnd3 (Cyclin D3), CcnG1 (Cyclin G1) and Cdk2 are significantly upregulated in the absence of Hdac3, while Cdkn2c (p18) expression is downregulated (Fig. 6C). This pattern of gene dysregulation is consistent with G1/S arrest of neural crest cells.
Hdac3 has previously been shown to directly repress the cell cycle inhibitor gene Cdkn1a (p21) in multiple tissues (Trivedi et al., 2008; Wilson et al., 2008). In order to determine whether the substantial upregulation of Cdkn1a observed in Hdac3Wnt1NCKO cranial mesenchyme is sufficient to account for cleft palate and microcephaly in these embryos, we deleted Hdac3 in neural crest on a Cdkn1a null background (Deng et al., 1995). Homozygous Cdkn1a−/− mice develop normally and are viable and fertile (Deng et al., 1995). We found that loss of p21 does not rescue the cleft palate and microcephaly observed in Hdac3Wnt1NCKO embryos, suggesting that dysregulation of additional cell cycle modulators contributes to altered regulation of neural crest expansion in Hdac3-deficient cranial neural crest cells, and ultimately to cleft palate and craniofacial hypoplasia (Supplemental Fig. 4).
In multiple aspects of embryogenesis, including neural crest development and limb development, the homeobox transcription factors Msx1 and Msx2 and the signaling molecule Bmp4 function to initiate apoptosis (Barlow and Francis-West, 1997; Graham et al., 1994; Lallemand et al., 2005; Marazzi et al., 1997). Msx1 is expressed in the palatal shelves and dental mesenchyme, and regulates development of these neural crest-derived structures (Satokata et al., 1994; Chen et al., 1996; Maas and Bei, 1997; Bei et al., 2000; Alappat et al., 2003). Msx2 is expressed in the early dental mesenchyme and is also crucial for normal tooth development (Maas and Bei, 1997; Winograd et al., 1997). Bmp4 is primarily expressed by epithelial structures in the developing face, which signal to neural crest-derived mesenchyme to mediate processes such as palatogenesis and odontogenesis, and its expression can be induced by Msx genes in neural crest derivatives (Zhang et al., 2002, Mitsiadis et al., 2010).
We measured the expression of important regulatory genes involved in craniofacial development using a candidate approach, and detected significant upregulation of Msx1, Msx2 and the target gene Bmp4 in the anterior cranial mesenchyme of E12.5 Hdac3Wnt1NCKO embryos, but found no significant differences in the expression of palate-specific markers Osr2 and Shox2 (Fig. 6D and Supplemental Fig. 5) (Lan et al., 2004; Yu et al., 2005). In situ hybridization reveals substantially more expression of Msx1 transcripts in the dental mesenchyme of mutant embryos than in littermate controls (Fig. 6E). We also observe increased expression of Msx1 in the developing palatal shelves of Hdac3-deficient embryos compared to littermate controls (Fig. 6F). Similar to Msx1, in situ hybridization for Msx2 shows elevated expression specifically in the dental mesenchyme of E12.5 Hdac3Wnt1NCKO embryos (Fig. 6E).
Interestingly, transgenic overexpression of Msx2 leads to cleft secondary palate, skull malformations, micrognathia, tooth hypoplasia and eyelid dysplasia, which phenocopies many aspects of the Hdac3Wnt1NCKO abnormalities (Winograd et al., 1997). Gain of BMP signaling also results in craniofacial abnormalities that partially phenocopy the abnormalities identified in Hdac3Wnt1NCKO embryos (He et al., 2010). These observations suggest that the hypoplasia of neural crest-derived structures observed in Hdac3Wnt1NCKO embryos may be in part mediated by derepression of Msx1 and Msx2 leading to increased apoptosis (Graham et al., 1994; Tríbulo et al., 2004; Park et al., 2005). Consistent with this hypothesis, we observed increased apoptosis in Hdac3 mutants in dental mesenchyme and palatal shelves, coincident with increased Msx expression (Figs. 6E and F).
To determine whether Msx gene upregulation has a functional role in mediating the observed cell cycle dysregulation and failure of neural crest structures to expand during development, we performed siRNA-mediated knockdown of Msx1 or Msx2, individually or in combination, in cultured neural crest mesenchyme from control and Hdac3Wnt1NCKO embryos. We observed a partial restoration of wild-type levels of neural crest proliferation following knockdown of either Msx1 or Msx2 (Supplemental Fig. 6). Taken as a whole, these results suggest that derepression of Msx1 and Msx2 in the absence of Hdac3 causes decreased proliferation and increased apoptosis in cranial neural crest cells, manifesting as abnormal facies, cleft palate, dental hypoplasia and bone deficiencies.
In addition to the Msx gene products, the T-box transcription factors Tbx2 and Tbx3 have been shown to play significant roles in palate expansion and neural crest development (Zirzow et al., 2009; Mesbah et al., 2012). Tbx2 and Tbx3, like Msx1 and Msx2, are normally expressed in neural crest-derived cranial mesenchyme and are important in cell cycle regulation during palatogenesis (Zirzow et al., 2009). Loss of Tbx2 or Tbx3 in craniofacial development leads to cleft palate due to excessive proliferation, and both proteins have been identified as inhibitors of the cell cycle in palatogenesis (Zirzow et al., 2009). We observe significant upregulation of these members of the T-box transcription factor family in Hdac3-deficient craniofacial mesenchyme, consistent with the observed cell cycle dysregulation in neural crest-derived tissue (Fig. 6D). Overall, the pattern of upregulation of these Msx, T-box transcription factors, cell cycle regulators and the signaling molecule Bmp4 – all known inhibitors of neural crest survival - is consistent with a model in which Hdac3 represses core inhibitors of the neural crest cell cycle in order to drive craniofacial development.
At multiple stages of neural crest development, the growth and survival of neural crest cells comes under positive and negative regulation from surrounding cell types, but is also subject to cell autonomous regulation (Sauka-Spengler and Bronner-Fraser, 2008). The homeodomain transcription factors Msx1 and Msx2 are important mediators of neural crest apoptosis in both early and late stages of craniofacial morphogenesis, but the temporal and spatial regulation of their expression is poorly understood. Interestingly, both increased and decreased Msx gene levels in developing neural crest lead to severe craniofacial abnormalities (Bei and Maas, 1998; He et al., 2010). That craniofacial morphogenesis is so exquisitely sensitive to Msx expression speaks to the importance of fine regulation of the timing and patterning of Msx1 and Msx2 transcription during neural crest development.
In this study, we have demonstrated that loss of Hdac3 in neural crest results in severe craniofacial malformations including microcephaly, cleft palate, impaired bone formation in the skull and hypoplasia of the teeth. Inactivation of Hdac3 by Wnt1-Cre results in cell cycle dysregulation in neural crest-derived structures, including increased expression of p21, p53, p57, Cyclin D1, Cyclin D3, Cyclin G1 and Cdk2. Hdac3Wnt1NCKO embryos exhibit elevated expression of Msx1 and Msx2 in the anterior cranial mesenchyme, and knockdown of Msx1 and Msx2 in mutant tissue partially normalizes neural crest proliferation. Incomplete rescue in this assay may have been due to incomplete knockdown of Msx1 and Msx2 using siRNA approaches, but may also suggest that factors other than Msx1 and Msx2 are functionally altered in Hdac3 mutants and contribute to the cell cycle dysregulation that we observe. Indeed we observe upregulation of Tbx2 and Tbx3 in Hdac3-deficient cranial mesenchyme, and abnormal T-box factor expression is also likely to contribute to the abnormal phenotype in Hdac3Wnt1NCKO mice. Dysregulation of Tbx factors and other regulators of neural crest development is also likely to explain the strong – but not complete – overlap between the phenotype of Hdac3Wnt1NCKO mice and Msx2 transgenic overexpressing mice (Winograd et al., 1997).
Previous work investigating the regulation of Msx1 transcription in cell lines identified several regulatory elements in the Msx1 upstream genomic region (Takahashi et al., 1997). LacZ reporters of a “minimal” Msx1 promoter containing these regulatory regions were strongly expressed in the neural crest-derived anterior cranial mesenchyme of E12.5 embryos (Takahashi et al., 1997). Similar work with the Msx2 promoter region has also identified highly conserved enhancers that contribute to Msx2 expression in the craniofacial mesenchyme when fused to lacZ reporters (Liu et al., 1994). In light of the canonical role of Hdac3 as a transcriptional repressor, it is reasonable to hypothesize that Hdac3 may directly repress Msx1 and Msx2 expression in neural crest by deacetylating histones at these promoters/enhancers, either at known or novel regulatory regions. However, our in vivo chromatin immunoprecipitation experiments have not demonstrated Hdac3 occupancy of previously described enhancer regions (Brugger et al., 2004; Hussein et al., 2003). As a global modulator of gene expression, Hdac3 is likely to affect the expression of multiple pathways involved in craniofacial development. Additional work is needed to determine whether Hdac3-mediated repression of Msx1 and Msx2 occurs via histone deacetylation at uncharacterized enhancer regions at these loci, through derepression of additional intermediate genes or via direct deacetylation of non-histone targets. Interestingly, recent results suggest that critical functions of Hdac3 in the embryo are independent of its interactions with NCOR1 and SMRT, co-repressors that are necessary for Hdac3 catalytic activity (You et al., 2013). Thus, Hdac3 may mediate important developmental processes through a non-canonical function unrelated to deacetylation of histone targets.
Our finding of decreased bone formation in Hdac3 conditional mutants is consistent with the pro-ossification role that Hdac3 is known to play in osteoblast precursors; however, by deleting Hdac3 prior to the specification and expansion of osteoblast precursors, we observe drastic deficiencies in bone formation not seen after deletion of Hdac3 in committed Osx-expressing cells (Razidlo et al., 2010). Our results demonstrate increased apoptosis of neural crest-derived mesenchyme in the developing calvaria, despite intact neural crest migration and normal expression of the osteoblast regulatory molecule Runx2. Paradoxically, deletion of Msx1 and Msx2 in neural crest leads to heterotopic bone formation, particularly in the frontal bone region, while Msx2 transgenic overexpression also results in ectopic bone formation, with decreased mineralization in the interparietal bone (Liu et al., 1995; Roybal et al., 2010; Winograd et al., 1997). While some of the bone abnormalities in Hdac3Wnt1NCKO embryos could be attributed to derepression of Msx gene expression, it is likely that additional targets of Hdac3 underlie the skeletal abnormalities in these mice. An interesting future area of research will involve determining additional targets of Hdac3 that regulate the early stages of bone development in neural crest progenitor cells and additional populations of osteoblast precursors.
(A) Neural crest migration evaluated by PlexinA2 expression and fate mapping. (Inset i, ii) E9.5 mutant and littermate controls demonstrate similar expression of PlexinA2, a marker of neural crest. (Inset iii, iv) Neural crest-derived cells fate mapped by Wnt1-Cre and the Z/EG reporter populate the first pharyngeal arch similarly in mutant and control embryos. (B) Quantification of the area occupied by cranial crest-derived GFP+ cells in the first pharyngeal arches of mutant and control embryos reveals no significant difference. Scale bars: (A): 200μm.
(A) Gross images of E15.5 embryos. Mutant embryos exhibit fully penetrant microcephaly, with micrognathia and a shortened snout. (B) H&E stained frontal sections of E14.5 embryos. Neural crest deletion of Hdac3 with Pax3Cre results in cleft palate (*). Scale bars: (A, B): 800μm.
(A, B) GFP, Runx2, and TUNEL staining of E12.5 heads. (Insets i, ii) Neural crest cells properly migrate to the outer layers of the developing calvaria and frontal bone primordium, shown here by fate mapping with the Z/EG reporter. (Inset iii) Mutant calvaria exhibit a normal pattern of Runx2 staining by immunohistochemistry, suggesting that differentiation of osteogenic precursors is maintained at this developmental stage. (Inset iv) TUNEL staining of adjacent sections reveals that neural crest cells exhibit significantly decreased survival in the frontal bone primordium of Hdac3Wnt1NCKO embryos. (C) Enlargement of TUNEL-stained images shown in (B) and quantification of TUNEL-positive cells per high power field. (*) denotes p <0.05. Scale bars: (A–C): 200μm.
H&E stained frontal sections of E16.5 heads. Neural crest deletion of Hdac3 on a p21+/− or p21−/− background results in cleft palate (arrow). Scale bar: 400μm.
Quantitative RT-PCR of microdissected E12.5 cranial mesenchyme. Mutant embryos express similar levels of Osr2 and Shox2 - known regulators of palate development - as littermate controls.
(A) siRNA-mediated knockdown of Msx1 and/or Msx2 in microdissected E12.5 cranial mesenchyme partially restores proliferation of mutant cells to wild-type levels as measured by phospho-histone H3 immunohistochemistry. (B) Validation of siRNA-mediated Msx knockdown. Transfection of cranial mesenchyme with 50nM siRNA produces partial but significant loss of Msx transcripts as measured by quantitative RT-PCR. (*) denotes p <0.05.
This work was supported by R01 HL071546, the WW Smith endowed chair, and the Spain Fund for Cardiovascular Research (J.A.E.), the Medical Scientist Training Program grant (N.S., M.G.), T32 HL007954-13 (M.G.) and K99-R00 HL098366 (C.M.T). We would like to acknowledge Ashley Cohen for animal husbandry and Kurt Engleka for OPT imaging and data analysis. We would also like to thank Mitch Lazar, Theresa Alenghat and Shannon Mullican for generously providing us with the Hdac3 floxed mice.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.