Dlx family genes encode homeodomain transcription factors with homology to the Drosophila Distal-less gene (Panganiban and Rubenstein, 2002). There are six Dlx genes in mammalian genome, Dlx1 through Dlx6, arranged as three tightly linked pairs (Dlx1/2, Dlx3/4, and Dlx5/6) (Porteus et al., 1991; Price et al., 1991; Robinson and Mahon, 1994; Simeone et al., 1994). Analyses of mice with targeted mutations have revealed that Dlx genes play crucial roles in development of multiple body parts including the forebrain, limb, and jaw (Qiu et al., 1995, 1997; Anderson et al., 1997; Acampora et al., 1999; Depew et al., 1999, 2002; Beverdam et al., 2002; Merlo et al., 2002; Panganiban and Rubenstein, 2002; Robledo et al., 2002; Robledo and Lufkin, 2006; Jeong et al., 2008). All six of the mammalian Dlx genes are expressed in the embryonic primordium of the jaw, called the first pharyngeal arch (PA1), primarily in the neural crest–derived mesenchyme but also in specific areas of the ectoderm (Dollé et al., 1992; Bulfone et al., 1993; Robinson and Mahon, 1994; Simeone et al., 1994; Qiu et al., 1997). PA1 comprises two domains: the maxillary arch (mxPA1), which is the prospective upper jaw, and the mandibular arch (mdPA1), the prospective lower jaw. Dlx1 and Dlx2 are expressed broadly in both mxPA1 and mdPA1, while the expression of Dlx3, Dlx4, Dlx5, and Dlx6 is confined to mdPA1. This nested pattern of Dlx expression is thought to be crucial in differentiating the lower jaw versus upper jaw fate within PA1 (Beverdam et al., 2002; Depew et al., 2002).
Targeted mutations in mice of Dlx1 and/or Dlx2, the focus of this report, resulted in an abnormal upper jaw (Qiu et al., 1995, 1997); the lower jaw appeared unaffected, likely due to functional compensation by other Dlx genes. In Dlx1−/− mutants, the ala temporalis (connection of the skull base to the temporal wall) was hypoplastic (Qiu et al., 1997). Dlx2−/− mutants had more extensive defects including severe reduction of the ala temporalis, fragmentation and hypoplasia of the squamosal bone and zygomatic arch on the temporal wall, dysmorphic incus, appearance of ectopic cartilage, and hypoplasia and partially penetrant clefting of the secondary palate (Qiu et al., 1995). Mice with deletions in both Dlx1 and Dlx2 (Dlx1/2−/−) showed exacerbation of the Dlx2−/− phenotype indicative of functional overlap between Dlx1 and Dlx2. Defects in Dlx1/2−/− mutants included fully penetrant cleft secondary palate and loss of upper molars (Qiu et al., 1997; Thomas et al., 1997).
Cleft palate is a relatively common congenital defect in humans, affecting approximately 1 out of 700 births (Tolarova and Cervenka, 1998; Dixon et al., 2011). Normal development of the palate (palatogenesis) involves precise coordination of growth and morphogenesis of orofacial tissue, which appears to be easily perturbed by genetic and environmental factors (Ferguson, 1988; Gritli-Linde, 2007, 2008; Dixon et al., 2011; Bush and Jiang, 2012). In mice, the secondary palate begins to develop around embryonic day (E) 11.5, when the palatal shelves emerge from the internal side of mxPA1. The palatal shelves grow vertically along the sides of the tongue (E11.5–E13.5), elevate above the tongue into a horizontal position (E14.0), grow towards the midline, where they meet and adhere to each other (E14.5). Finally, the epithelial seam between the shelves disintegrates to complete their fusion (E16.5). Over 200 mouse lines with mutation(s) in one or more genes exhibit cleft palate phenotype (Iwata et al., 2012), and detailed analyses of cellular and gene expression changes in some of these mutants have been instrumental in elucidating signaling and genetic pathways regulating palatogenesis (Gritli-Linde, 2007, 2008; Bush and Jiang, 2012). However, in most of these mutants, palate development was affected only after it had progressed substantially (E13 or later), and thus the analyses focused on mid to late stages (E12.5–E14.5). Therefore, molecular regulation of the initial steps of palatogenesis (around E11.5) remains poorly understood.
Although it was reported almost 15 years ago that mouse Dlx1/2−/− mutants had cleft secondary palate (Qiu et al., 1997), nothing was known about the molecular and cellular etiology behind this phenotype. In fact, given that these mutants suffer from structural defects in a broad region of the face (Qiu et al., 1997), it was not even known whether the cleft palate was due to problems intrinsic to the palatal shelves, or was indirectly caused by a mechanical hindrance due to dysmorphology of surrounding structures. To answer this question, we analyzed the morphological, cellular, and gene expression changes in the developing palate caused by inactivation of Dlx1 and Dlx2. Our data show that the activities of Dlx1 and Dlx2 are important for the initial outgrowth of the palatal shelves in a region-specific manner, and are essential for the normal expression of other important regulators of palatogenesis.
Dlx1 and Dlx2 Are Expressed Prior to and During Early Stages of Palatogenesis But Are Down-Regulated Afterward
To identify the time window when Dlx1 and Dlx2 can potentially influence palatogenesis, we examined their expression from E8.5 to E13.5 at a high spatio-temporal resolution (Fig. 1). Expression of Dlx2, but not Dlx1, was detected at E8.5, in the nascent PA1 and in the area between the cranial neural crest (CNC) and PA1, presumably migrating CNC cells (Fig. 1A–D). Dlx1 expression was first detected in PA1 at E9.0, and by E9.5 both genes were expressed in most of PA1 (Fig. 1E–H). From E10.5, we used a series of coronal sections of the head to examine Dlx expression along the A-P axis of the developing palate. At E10.5, shortly before the formation of the palatal shelves, Dlx1 and Dlx2 were expressed broadly in the mesenchyme of PA1 at the middle and posterior level (Fig. 1K–N), including the prospective palate area (ventro-medial domain of mxPA1, Fig. 1I–N). However, at the anterior level, Dlx expression was confined to the lateral domain away from the prospective palate, mainly in the epithelium (Fig. 1I–J). A similar pattern continued at E11.5 in the nascent palatal shelves, although the expression levels were reduced in the nasal half (Fig. 1O–T). The palatal expression of the Dlx genes continued to diminish through E13.0 (Fig. 1U–Z) and became almost undetectable by E13.5 (data not shown) except in the dental mesenchyme of the developing molars (Fig. 1W,X). These results were consistent with earlier reports on the expression of Dlx1 and Dlx2 (Dollé et al., 1992; Bulfone et al., 1993; Robinson and Mahon, 1994; Qiu et al., 1997). Based on these data, we inferred that Dlx1 and Dlx2 might be involved in the beginning of palate development, but unlikely to play any direct roles at later steps of palatogenesis.
Posterior Palatal Shelves Fail to Grow in Dlx2−/− Mutants
To determine when the defects in palatogenesis first arise in Dlx1/2−/− mutants and how they progress, we examined tissue morphology of the palate region in sections from E10.5 to E14.5 (Fig. 2). At E10.5, Dlx1/2−/− mutants exhibited hypoplasia of mxPA1; the tissue mass lateral to the eyes was grossly reduced (Fig. 2C,D). While striking, this phenotype most likely correlates with the defects in lateral facial structures, such as loss of zygomatic arch and hypoplasia of squamosal bone, rather than the cleft palate. Therefore, herein we did not investigate this phenotype, but rather focused on the palate, which arises from the ventro-medial domain of mxPA1.
At all stages examined, anterior palate was relatively normal in the mutants (Fig. 2). Although delay in anterior palatal shelf elevation was observed in some cases (Fig. 2S,T), it was likely to be an indirect consequence of more severe defects in the middle and posterior palate. At the middle level, the shape and cell density of the prospective palate region of mxPA1 appeared similar between the controls and mutants at E10.5 (Fig. 2C,D). When the palatal shelves formed at E11.5, the size (area measured in sections) of the mutant palatal shelves was 79% of the controls (Fig. 2I,J,Y; control: 0.064±0.003 mm2, mutant: 0.050±0.004 mm2). The size difference increased as development progressed, and at E14.5, the mutant palatal shelves failed to elevate above the tongue (Fig. 2O,P,U,V,Y).
The posterior palate was affected the most profoundly. At E10.5, the size of the mutant's prospective palate appeared normal (Fig. 2E,F) although in some mutants mesenchymal cell density appeared slightly lower. However, cell counting did not reveal a statistically significant difference in the posterior palate (control: 8,588±915 cells/mm2, mutant: 8,348±753 cells/mm2, P = 0.744). On the other hand, by E11.5, the mutant palatal shelves were less than half (47.5%) the size of the controls (Fig. 2K,L,Y; control: 0.040±0.006 mm2, mutant: 0.019±0.005 mm2). Furthermore, while the control palatal shelves underwent rapid vertical growth between E11.5 and E13.5 (Fig. 2K,Q,Y), the mutant palatal shelves grew little (Fig. 2L,R,Y). By E14.5, the palatal shelves were essentially non-existent in the mutants at the posterior level (Fig. 2W,X).
In summary, the posterior palate suffered severe growth deficiency from the onset of the palatogenesis in Dlx1/2−/− mutants, while the anterior palate was largely spared. This difference in phenotype along the A-P axis is consistent with the lack of Dlx1 and Dlx2 expression in the anterior palate region (Fig. 1).
Cell Proliferation and Cyclin D1 Expression Are Reduced in the Posterior Palate of Dlx1/2−/− Mutants
To explain the growth deficiency of the palatal shelves of Dlx1/2−/− mutants, we examined cell proliferation and apoptosis, using phospho-histone H3 and cleaved caspase 3 as respective markers. Since Dlx1 and Dlx2 are expressed in the mesenchyme but not the epithelium of the palatal region, we focused our analyses on the mesenchyme. At E10.5, we counted mitotic cells in the ventro-medial domain of mxPA1, and also in the ventro-lateral domain for comparison (Fig. 3A). We found that only the ventro-medial area at the posterior level showed a statistically significant decrease in cell proliferation in the mutants (Fig. 3B,C). Similarly, at E11.5, only the posterior palatal shelves showed a decrease in cell proliferation in the mutants (Fig. 3D). This result is consistent with the severe growth deficiency in the posterior palate of the mutants. Although there was a moderate reduction in the size of the mutant palatal shelves at the middle level, we did not find a decrease in mitotic rate at this position at E11.5. We went on to examine cell proliferation at E12.5, and, surprisingly, found that the mutant middle palate had a higher cell proliferation rate than the controls (Fig. 3E). This result was unique to the middle palate; cell proliferation rates from anterior and posterior levels did not show a statistically significant difference between Dlx1/2−/− mutants and controls at E12.5 (Fig. 3E). At E13.5, we did not detect a significant difference in the cell proliferation rates between the two genotypes at any axial level (Fig. 3F). Increase in cell proliferation in the mutant middle palate at E12.5, although transient, is in apparent contradiction to the decrease in tissue size at the same position found at E13.5. Therefore, more complex mechanisms may underlie the mutant palatal phenotype at the middle level than at the posterior level (see Discussion section).
We examined apoptosis in the developing palate at E10.5, E11.5, E12.5, and E13.5, but did not find significant difference between the controls and mutants (data not shown).
Previous studies showed that Cyclin D isoforms are expressed in the palatal shelves and regulate cell proliferation at E13.5–E14.5 (Ito et al., 2003; Lan and Jiang, 2009; Iwata et al., 2012). We investigated whether expression of these cyclins correlated with the changes in cell proliferation in Dlx1/2−/− mutants at E10.5, E11.5, and E12.5 (Fig. 4). For the expression of Cyclin D1 (Ccnd1) and Cyclin D2 (Ccnd2), and other genes discussed later, we are only presenting the pictures for the middle and posterior levels due to space limitations, but we also examined their expression at the anterior level and found no phenotype (data not shown). In the control embryos, both Ccnd1 and Ccnd2 were expressed strongly in the mesenchyme of the prospective palate domain at E10.5 (Fig. 4A,C,M,O) and in the newly formed palatal shelves at E11.5 (Fig. 4E,G,Q,S), but at lower levels at E12.5 (I,K,U,W). Ccnd2 also showed strong expression in the epithelium at all three stages (Fig. 4M,O,Q,S,U,W). In Dlx1/2−/− mutants, Ccnd1 expression was greatly reduced only at the posterior level at both E10.5 and E11.5 (Fig. 4B,D,F,H). At E12.5, Dlx1/2−/− middle palate had significantly more Ccnd1 expression than the control middle palate (Fig. 4I,J). However, in the posterior palate, Ccnd1 level appeared similar between the two genotypes at this stage (Fig. 4K,L). Ccnd2 expression in the mutants appeared comparable to that in the controls at all the ages examined (Fig. 4M–X), except for up-regulation in the mesenchyme of the developing upper molar, where ectopic cartilage forms (asterisks in Fig. 4R,V; Thomas et al., 1997). Therefore, the changes in the expression of Ccnd1, but not Ccnd2, in Dlx1/2−/− mutants correlate precisely with the cell proliferation phenotype temporally and spatially.
Decreases in Ccnd1 Expression and Cell Proliferation in the Dlx1/2−/− Mutant Palate Are Not Caused by a Change in Shh Signaling
Several studies established that Shh signaling regulates cell proliferation in the palatal mesenchyme around E13.5, either via another signaling molecule Bmp2, or by regulating Ccnd1 and Ccnd2 expression (Zhang et al., 2002; Rice et al., 2004; Han et al., 2009; Lan and Jiang, 2009). Therefore, we investigated whether Shh signaling also regulates cell proliferation in the prospective and nascent palate at E10.5 and E11.5. In control embryos, Shh was expressed in the epithelium of the palate area at all A-P levels (Fig. 5A,C,E,G; data not shown), and its downstream target gene, Ptch1, was expressed in the epithelium and underlying mesenchyme indicating the activation of Hedgehog signaling in these tissues (Fig. 5I,K,M,O; data not shown). To our surprise, we found that the expression of both Shh and Ptch1 was normal in Dlx1/2−/− mutants at E10.5, even at the posterior level where Ccnd1 expression and mitosis were affected (Fig. 5A–D, I–L).
At E11.5, Shh and Ptch1 expression was reduced in the mutants, but the change was mainly at the middle level. In other words, at the posterior level of Dlx1/2−/− mutants, cell proliferation was reduced even though Shh signaling appeared quite normal (Figs. 3C,D, 5K,L,O,P), while at the middle level cell proliferation was not reduced despite severe down-regulation of Shh signaling at E11.5 (Figs. 3D, 5M,N).
Since Dlx1 and Dlx2 are expressed in the mesenchyme but not in the epithelium at the middle level (Fig. 1K,L,Q,R), the change in Shh expression in Dlx1/2−/− mutants would require intermediary factor(s) relaying the effect from mesenchyme to epithelium. Around E13.5, Bmp4 and Fgf10 are the mesenchymal factors that signal to the epithelium and positively regulate Shh expression there (Zhang et al., 2002; Rice et al., 2004). Therefore, we examined the expression of Fgf10 and Bmp4 in controls and Dlx1/2−/− mutants at E10.5 and E11.5 (Fig. 5Q–f).
Fgf10 is normally expressed in the anterior and middle palate but not the posterior palate at E12.5–E13.5 (Alappat et al., 2005; Hilliard et al., 2005). This was also true at E10.5 and E11.5 (Fig. 5Q,S,U,W; Alappat et al., 2005). In Dlx1/2−/− mutants, Fgf10 was down-regulated at the middle level at both E10.5 and E11.5 (Fig. 5Q,R,U,V).
Bmp4 is normally expressed only in the anterior palate at E13.5 (Zhang et al., 2002). However, we found its expression in the palatal mesenchyme throughout the A-P axis at E11.5 (Fig. 5c,e); it was not expressed in the future palate area at E10.5 (Fig. 5Y,a). In Dlx1/2−/− mutants, Bmp4 was severely down-regulated in the middle and posterior palate at E11.5 (Fig. 5c–f). Combined, these data suggested that at least at the middle palate, the same mechanism could regulate the epithelial Shh expression at E11.5 as at E13.5, namely, through Fgf10 and/or Bmp4 signaling from mesenchyme to epithelium.
Expression of Several Transcriptional Regulators of Palatogenesis Is Affected in Dlx1/2−/− Palatal Shelves
In addition to the signaling molecules, many transcription factors regulate palate development (Gritli-Linde, 2007, 2008; Bush and Jiang, 2012). We found that the expression of five transcription factors with known connection to palatogenesis was lost or greatly reduced in the mxPA1 and palatal shelves of Dlx1/2−/− mutants (Fig. 6).
During normal development, a LIM domain-homeodomain transcription factor Lhx6 (Grigoriou et al., 1998), a homeodomain transcription factor Barx1 (Tissier-Seta et al., 1995), a basic helix-loop-helix (bHLH) domain and PERIOD-ARNT-SIM (PAS) domain transcription factor Sim2 (Dahmane et al., 1995; Fan et al., 1996), and a zinc finger transcription factor Osr1 (So and Danielian, 1999) are all strongly expressed in PA1 at E10.5 (Fig. 6A,C,E,G) and the nascent palatal shelves at E11.5 (Fig. 6I,K,M,O,Q,S,U,W). Another zinc finger transcription factor, Osr2, is expressed in small domains of PA1 at E10.5 but is rapidly up-regulated in the palatal shelves one day later (Fig. 6Y,a; Lan et al., 2004). In Dlx1/2−/− mutants, the expression of all five genes was severely down-regulated in mxPA1 and the palatal shelves (Fig. 6A–b).
Mouse mutants for Barx1 and Sim2 have cleft palate defect (Shamblott et al., 2002; Kim et al., 2007; Miletich et al., 2011); while the details on the palate phenotype of Barx1 mutants have yet to be reported, in Sim2 mutants the defect was apparent first at E14.5 as reduced cell density in the mesenchyme and later as smaller palatal shelves (Shamblott et al., 2002). Lhx6 has been implicated in palatogenesis because mice missing the activities of its homolog Lhx8 (also known as Lhx7) (Lhx8−/−) or both Lhx6 and Lhx8 (Lhx6−/−;Lhx8−/−) have cleft palate (Zhao et al., 1999; Denaxa et al., 2009). We found that Lhx6 single mutants (Lhx6−/−) had a fused palate, but skeleton preparations revealed that they had no or very small palatal processes of maxilla (Fig. 6c,d; N=5). Direct evidence for the importance of Osr1 in palatogenesis has not been reported. On the other hand, Osr2 was shown to be essential for normal growth of the palatal shelves from E13.5 (Lan et al., 2004).
Analysis of Dlx2−/− Mutant Palate Phenotype
Inactivating Dlx2 alone is sufficient to cause cleft palate, although with incomplete penetrance (80%; Qiu et al., 1995). Morphological examination at E13.5 revealed growth deficiency in the palatal shelves of Dlx2−/− (Fig. 7A–D), but it was much milder than what was observed in Dlx1/2−/− mutants. Neither Ccnd1 expression (at E10.5 and E11.5) nor cell proliferation rate (at E11.5) appeared significantly different between Dlx2−/− mutant and control palate areas (Fig. 7E–I), although we cannot rule out the possibility that subtle changes below the sensitivity limit of our methods may underlie the palate hypoplasia of Dlx2−/− mutants.
We examined whether the expression of the genes that were down-regulated in Dlx1/2−/− mutants was also affected in Dlx2−/− mutants. Shh and Ptch1 were down-regulated at the medial edge of the middle palate but appeared normal in the posterior palate (Fig. 7J–Q). Fgf10 expression did not appear altered in Dlx2−/− mutant (Fig. 7R,S), but Bmp4 expression was clearly down-regulated (Fig. 7T–W). As for the transcription factor genes, expression of Lhx6, Barx1, and Osr2 was moderately reduced in Dlx2−/− mutants (Fig. 7X–a, f–i), whereas Sim2 and Osr1 expression was greatly reduced in Dlx2−/− mxPA1 (Fig. 7b–e). Therefore, the changes in the expression of Shh, Ptch1, Bmp4, Lhx6, Barx1, Sim2, Osr1, and Osr2 may contribute to the cleft palate phenotype of Dlx2−/− mutants.
Role of Dlx1/2 in Early Palate Development
In this study, we investigated the etiology of the cleft palate of mouse Dlx1/2−/− mutants. We provide evidence that Dlx1 and Dlx2 regulate the initiation of palatogenesis through promoting mesenchymal proliferation in the posterior palate independently of Shh signaling. The early growth defect in Dlx1/2−/− mutants was highly localized to the posterior palate, indicating the existence of heterogeniety along the A-P axis as found at later stages of development (Hillard et al., 2005; Yu and Ornitz, 2011). Decreases in Ccnd1 expression and cell proliferation rate provided molecular and cellular explanations for the growth phenotype in the posterior palate. In addition, our evidence suggests that Dlx1 and Dlx2 non-autonomously promote Shh epithelial expression through Bmp4 and Fgf10 in the middle palate, and further regulate palatal development through promoting mesenchymal expression of the Lhx6, Barx1, Sim2, Osr1, and Osr2 transcription factors.
One confounding result from our analysis of palate development was that, at least at one point (E12.5), Dlx2−/− mutants had a higher cell proliferation rate than the controls in the middle palate (Fig. 3), even though the mutant middle palate was consistently smaller than the controls (Fig. 2). While increase in cell death could counter the effect of increase in cell proliferation, we did not find a significant difference in apoptosis rates between the control and mutant palate at E10.5–E13.5 (data not shown). Another possibility is that cell distribution is somehow altered in the mutant middle palate; for example, cells generated at the middle palate might migrate or be displaced posteriorly, where they would partially fill the void that would otherwise be left in the posterior half of the oral cavity. Cells from the mutant middle palate might also have been shifted toward the dorso-lateral region of the upper jaw, another region that has a growth deficiency (Fig. 2). If more cells “leave” the middle palate through such redistribution than those produced by enhanced cell proliferation in that region, then the net effect would be the decrease in the size of the middle palate.
Timing of Dlx1/2 Action
Two studies from other species suggested that Dlx2 plays important roles in the pre-migratory and/or migrating cranial neural crest cells (CNCCs) before their arrival at the Pas: knockdown of dlx2a expression in zebra fish caused increased apoptosis in migrating CNCCs (Sperber et al., 2008). In addition, ectopic expression of Dlx2 in the premigratory CNCCs of chick embryos inhibited migration of these cells (McKeown et al., 2005). This raises the question of whether the palate defect of mouse Dlx1/2−/− mutants is an indirect consequence of the abnormalities in the pre-migratory/migrating CNCC population. We do not believe this is the case due to the following reasons. First, neither Dlx1 nor Dlx2 is expressed in the pre-migratory CNCCs in mice (Fig. 1; Dollé et al., 1992; Bulfone et al., 1993; Robinson and Mahon, 1994). While Dlx2 is expressed in the migrating CNCCs (Fig. 1; Bulfone et al., 1993; Robinson and Mahon, 1994), we did not find any difference between Dlx1/2−/− mutants and controls in the expression patterns of a gene that marks migrating CNCCs, namely, Tcfap2a (Mitchell et al., 1991) (Fig. 8A,B), apoptosis of the migrating cells (Fig. 8C,D), or the contribution of CNCCs to the PA1 mesenchyme labeled by the Wnt1-Cre reporter system (Fig. 8E,F; Danielian et al., 1998; Soriano, 1999; Chai et al., 2000). Therefore, there is no indication that the migration of CNCCs to PA1 was affected in Dlx1/2−/− mutants. The second line of evidence comes from the fact that the cleft palate phenotype of Dlx1/2−/− mutants is much more severe than that of Dlx2−/− mutants (compare Figs. 2 and 7); this indicates that in Dlx2−/− single mutants, Dlx1 performs some of the normal functions of Dlx2 that are important for palate development. Since Dlx1 is not expressed in the migrating CNCCs (Fig. 8G,H) but is co-expressed with Dlx2 in the post-migratory CNCCs in PA1 (Fig. 1; Dollé et al., 1992; Bulfone et al., 1993; Robinson and Mahon, 1994), the parsimonious explanation is that the actions of Dlx1 and Dlx2 that are important to palatogenesis occur at PA1, and not in the migrating CNCCs. The same logic applies to most other craniofacial defects of Dlx2−/− because they are more severe than those of Dlx2−/− (Qiu et al., 1995, 1997). Therefore, we conclude that in mice, the timing for the major functions of Dlx1 and Dlx2 in the craniofacial development is after CNCCs reach PA1.
Genetic Regulation of the Initiation of Palatogenesis
Despite the recent progress in our understanding of the palate development, what regulates the initiation of palatogenesis is largely unknown. This can be attributed to the fact that in almost all the mouse mutants reported to have cleft palate, the initial formation of the palatal shelves and the early part of the vertical growth (up to E12.5) were unaffected, and defects appeared only in the mid- to late-stages of palatogenesis (E13 and later) (Gritli-Linde, 2007, 2008; Bush and Jiang, 2012). Prior to the current study, there have been only a couple of examples where the inactivation of a gene led to hypoplasia of developing upper jaw before E12.5 that was not caused by problems in CNCC production or migration. Embryos with neural crest–specific inactivation of Alk5 (encoding TGFβ type I receptor) or Smad4 (encoding a transcriptional regulator downstream of TGFβ) had hypoplastic mxPA1 from E10.5–E11 (Dudas et al., 2006; Ko et al., 2007). However, in both cases, it was the increase in cell death, not a change in proliferation, which caused the hypoplasia. By contrast, region-specific reduction in cell proliferation was behind the growth deficiency of posterior palatal shelves in Dlx1/2−/− mutants, which was evident by E11.5. Therefore, Dlx1/2−/− mutants present a unique example where the initial formation of the palatal shelves is disrupted and the phenotype cannot simply be explained by general compromise in tissue integrity.
Signaling Pathways Regulating Cell Proliferation in the Palate
How cell proliferation is regulated in the developing palate is best understood around E13.5, shortly before the palatal shelves change orientation from vertical to horizontal. Shh expression in the epithelium has a central role at this stage; Bmp4, Fgf10 and Fgf7 produced in the palatal mesenchyme signal to the epithelium, where Bmp4 and Fgf10 positively regulate Shh expression, whereas Fgf7 represses Shh expression (Zhang et al., 2002; Rice et al., 2004; Han et al., 2009). Shh then signals back to the mesenchyme to promote proliferation, either via Bmp2 or by directly regulating Ccnd1 and Ccnd2, depending on the position along the A-P axis (Zhang et al., 2002; Lan and Jiang, 2009).
A recent study identified another genetic pathway crucial for cell proliferation in the palate, operating after palatal shelf elevation (E14.5) (Iwata et al., 2012); at this time, Fgf9 is the mitogen positively regulating Ccnd1 and Ccnd3 expression. Fgf9 expression is regulated by TGFβ signaling.
At E10.5–E11.5, when we found reduced cell proliferation in Dlx1/2−/− mutants, Shh is expressed in the epithelium of the ventro-medial quadrant of mxPA1, the location of palatal shelf outgrowth (Fig. 5; Jeong et al., 2004). On the other hand, Fgf9 is expressed more laterally at the dental lamina (Kettunen and Thesleff, 1998; Colvin et al., 1999). However, despite Shh being expressed at the right place and time, it did not mediate the decrease in cell proliferation in the posterior palate of Dlx1/2−/− mutants. Conversely, removing Shh signaling in mxPA1 mesenchyme did not affect the expression of Dlx1 and Dlx2 here, indicating that the Dlx genes are not downstream of Shh signaling (Jeong et al., 2004). Therefore, the regulation of cell proliferation by Dlx1 and Dlx2 at the beginning of palatogenesis (E10.5–E11.5) is through a mechanism independent of Shh, the details of which remain to be elucidated.
Effect of Dlx1/2 on Later Stages of Palate Development
In addition to the early growth defect, the developing palate of Dlx1/2−/− mutants exhibited changes in the expression of several genes implicated in palate development; expression of Shh, Fgf10, and Bmp4, which form a signaling loop important for cell proliferation around E13.5 (Zhang et al., 2002; Rice et al., 2004; Lan and Jiang, 2009), was down-regulated at the middle palate of Dlx1/2−/− mutants by E11.5 (Fig. 5). Also, the expression of Lhx6, Barx1, Sim2, Osr1, and Osr2 was lost or greatly reduced at E10.5 and E11.5 (Fig. 6). Therefore, although the severe early defect in Dlx1/2−/− mutants masks any perturbation of the later steps of palatogenesis, Dlx1 and Dlx2 likely play indirect but extensive roles as upstream regulators of other important factors for this process.
All the experiments using mice were performed following protocols approved by UCSF, NICHD, or NYU institutional committee on the use of laboratory animals. Mice carrying targeted mutant alleles of Dlx2, Dlx1/2, and Lhx6 have been previously described (Qiu et al., 1995, 1997; Choi et al., 2005). For all the mutant lines used in this study, the animals were of mixed genetic background of 129, C57Bl/6, and CD1. As controls for the homozygote mutants (−/−), littermate heterozygote (+/−), or wild type (+/ +) embryos were used without distinction. For each experiment, mutant and control embryos were carefully stage-matched; at E8.5–E11.5, we used the number of somites as a criterion. At E12.5 and beyond, we used a combination of morphological criteria including overall size of the body, size and shape of the limb buds, and degree of digitation.
Preparation of Sections
All the tissue sections used in this study are frozen sections. Embryos were dissected in phosphate-buffered saline (PBS), fixed in 4% paraformaldehyde at 4°C for 1–2 days, incubated in 10% and then 20% sucrose in PBS for cryoprotection, and embedded in OCT compound (Tissue-Tek) for coronal sections. Sections were cut at 12 μm (E10.5)–18 μm (E14.5), and collected on slides such that each slide contains the sections covering the entire antero-posterior (A-P) length of the oral cavity at a 48 μm (E10.5)–180 μm (E14.5) interval. Therefore, for each section in situ hybridization or immunofluorescence experiment presented in this report, we examined the marker expression in the whole sets of sections and drew conclusions, even though no more than three sections (representing anterior, middle, and posterior levels) per genotype are shown in the figures. To select anterior, middle, and posterior sections of the palatal region consistently across different genotypes and ages, we used the following criteria; “anterior” is just anterior to the upper molars. “Middle” is from the level of upper molars to the level of lower molars, and included eyes and optic stalks/nerves in the section. “Posterior” is just posterior to the molars and eyes.
In Situ Hybridization, Cresyl Violet Staining, Skeleton Preparation, and Cell Death Assays
Whole mount and section in situ hybridizations were performed using digoxigenin-labeled RNA probes as previously described (Jeong et al., 2004, 2008). DNA templates for in situ hybridization probes were obtained from other researchers, purchased from companies, or PCR-cloned from mouse E10.5 PA1 cDNA library or tail genomic DNA. Further information on the probes is available upon request. To visualize tissue morphology, the sections were stained with 0.1% cresyl violet solution for 10 min, washed and dehydrated with ethanol and xylene, and coversliped with Permount (Fisher, Pittsburgh, PA). Skeleton staining of newborn animals was performed as previously described (Jeong et al., 2004). Cell death was detected on tissue sections using anti-cleaved caspase3 antibody (Cell Signaling Technology, Danvers, MA) or in whole embryos using LysoTracker (Invitrogen, Carlsbad, CA) as previously described (Jeong et al., 2004; Grieshammer et al., 2005).
Quantitative Analyses of the Size of the Palatal Shelves, Cell Proliferation Rates, and Cell Density
The areas of the (prospective) palatal shelves were measured from sections stained with DAPI (E11.5) or cresyl violet (E13.5) using ImageJ program (Abramoff et al., 2004). The boundary of a palatal shelf was defined as marked in Figure 2. Three control and three mutant embryos were analyzed at each stage, and from each embryo two palatal shelves were measured. Student's t-test was used to determine if the differences between the genotypes were statistically significant.
Immunofluorescence for phospho-histone H3 was performed as previously described (Jeong et al., 2004) using an antibody from Upstate Biochem (East Syracuse, NY) and DAPI for nuclear counter staining. To calculate the percentage of mitotic cells, we first counted total number of cells in a defined area from DAPI staining, by automated counting using ImageJ plug-in followed by manual confirmation. Then the number of phospho-histone H3-positive cells from the same area was counted manually, and this number was divided by the total number of cells. At E10.5, there are yet no palatal shelves that one can demarcate, and thus we defined the prospective palate area as the medial half of the ventral mxPA1, whose dorsal boundary was a straight line at a fixed distance from the mid-point of the ventral epithelium of mxPA1 (see Fig. 3A). The medial boundary of mxPA1 was matched among different samples using stereotypical bendings of the ventral epithelium as landmarks. At E11.5 and E12.5, the countings were performed in the palatal shelves as demarcated in Figure 2. For each stage, three embryos per genotype, two (prospective) palatal shelves per embryo were analyzed, and Student's t-test was used to determine if the differences between the genotypes were statistically significant.
To determine the mesenchymal cell density in the posterior palate at E10.5, we measured the area of the posterior palate (boundaries defined as described above) from DAPI-stained sections using ImageJ, and then used this value to divide the total number of cells in that area (which was counted as described above). Three embryos per genotype, two prospective palatal shelves per embryo were analyzed, and Student's t-test was used to determine if the difference between the genotypes was statistically significant.
We thank Regeneron for Lhx6 mutant mice (MAID #406); Purvi Patel for genotyping mice and embryos; Hideyo Ohuchi, Joy Yang, and Jun Aruga for in situ hybridization probes; Jason Long, Jeremy Cholfin, Inma Cobos, Greg Potter, and Andrea Faedo for sharing reagents and ideas; and Gail Martin, Ophir Klein, Ross Metzger, Maria Barna, and the members of Rubenstein lab and Martin lab for helpful discussions. This work was funded by grants from NIH (NIDCD R01 DC05667), March of Dimes, Weston Havens Foundation, Hillblom Foundation and Nina Ireland to J.L.R.R., a grant from NIH (NIDCR K99/R00 DE019486) to J.J., and by funds from the NICHD intramural research program to H.W.