Although remarkable for their stunning array of transitional and end-point phenotypic forms, gnathostome skulls are equally well characterized by their possession of conserved regions of structure (Barghausen and Hopson,1979; de Beer,1985; Goodrich,1958). This applies to both the components of the chondrocranium (the jaw-associated branchial arch (BA) derived splanchnocranium and the neurocranium consisting of the midline trabecular and parachordal basal plates and the otic, optic, and olfactory capsules) and the skeletal elements of the dermatocranium (the dermal skeleton) that forms in association with the chondrocranium and is well conserved along macrotaxanomic lines. Fate-mapping studies in gnathostome vertebrates have demonstrated that the craniofacial complex—including the splanchnocranium, the trabecular basal plate, and the olfactory and optic capsules and their associated dermal skeletons—is derived from a rostral Hox-negative, or “trigeminal,” cranial neural crest (CNC) cell population (Chai et al.,2000; Couly et al.,1993,1998; Kuratani,2005a,b; Le Douarin et al.,2004; Osumi-Yamashita et al.,1994; Santagati and Rijli,2003).
Understanding both the conserved and divergent patterns of structural designs in the gnathostome skull requires explaining how this Hox-negative CNC population is informed to manifest structure within and between taxa. A “Hinge and Caps” (H&C) model has previously been proposed that places CNC competence to respond to localized epithelial signals at the root of much of craniofacial development and evolution (Fig. 1; Depew and Compagnucci,2008; Depew et al.,2002a,b, 2005; Depew and Simpson,2006). Although the Hox-negative CNC carries with it taxa-specific interpretive information, there is ever accumulating evidence that both cephalic endodermal and ectodermal epithelial populations supply patterning information to which the CNC component responds to form the jaws, trabecular basal plate and olfactory and optic capsules (Couly et al.,2002; Creuzet et al.,2005; Depew et al.,2002b; Depew and Simpson,2006; Graham et al.,2008; Hu et al.,2003; Marcucio et al.,2005; Song et al.,2004; Trumpp et al.,1999). Significant among the H&C model's posited localized epithelial patterning centers is that at the lambdoidal junction (or λ-junction), so called for the “λ” shape found at the confluence of the maxillary process of the first BA (mxBA1) and the medial (mFNP) and lateral (lFNP) frontonasal processes (Fig. 1; Depew and Compagnucci,2008; Depew and Simpson,2006; Tamarin and Boyde,1977). The epithelial component of the mature λ-junction, as typically seen in mice around embryonic day 10.5 (E10.5), stems from the surface cephalic ectoderm (SCE) that gives rise to the optic, olfactory, and hypophyseal placodes as well as covers the distal end of mxBA1 (see Fig. 1), and disruption of positioning and strength of signaling factors in this region (as with the application of growth factor-soaked beads) is well known to disrupt skeletal patterning of the region (Barlow and Francis-West,1997; Foppiano and Marcucio,2007; Lee et al.,2001). This position in mammals is particularly developmentally and functionally complex and critical as it outlines the positions of the nostrils (choanae), the formation of the upper lips, the nasal capsules, and the primary and secondary palates (Depew and Compagnucci,2008; Gaare and Langman,1977; Pourtois,1972; Tamarin,1982; Tamarin and Boyde,1976,1977; Warbrick,1960; Waterman and Meller,1973).
Pax6 is expressed early during embryogenesis in the anterior ectoderm, including that associated with the forming neurectoderm, eye, nose, and the λ-junction (Fig. 1; Grindley et al.,1995; Walther and Gruss,1991). Pax6 appears to be involved with optic, olfactory, brain, and pancreatic development but it is also required for proper craniofacial skeletal development (Collinson et al.,2003; Hill et al.,1991; Hogan et al.,1986; Kaufman et al.,1995; Matsuo et al.,1993; Osumi-Yamashita et al.,1997a; Quinn et al.,1996,1997). There are more than 20 known mutations involving the rodent Pax6 locus (collectively referred to as Small eye alleles for the hypoplastic eyes of the heterozygotes) and a range of cephalic phenotypes have been documented for these mutations (Hill et al.,1991; Hogan et al.,1986). While the optic, olfactory, and forebrain nervous system defects in a small number of these Pax6 mutants, in particular the rat rSey/rSey and mouse Pax6Sey/Sey and Pax6Sey-neu/Sey-neu (hereafter Pax6SeyN/SeyN) lines, have been analyzed many times, the craniofacial skeletal deficits encountered and their mechanistic etiologies have been less well characterized (Depew et al.,2002b).
With regard to skeletal development, the most extensively investigated Pax6 mutant has been the rSey/rSey mutant. Analysis of the rSey/rSey mutants demonstrated a number of regional defects in the olfactory and optic capsules and associated dermatocranium. Because Pax6 is not expressed in the mesenchymal populations that directly yield the skeletal structures that are deficient in the rSey/rSey mutants, noncell autonomous explanations for these deficiencies have previously been investigated (Fujiwara et al.,1994; Matsuo et al.,1993; Nomura et al.,2007; Osumi-Yamashita et al.,1997a). As most of the defective skeletal elements are derived from the CNC, this population of cells has been most intensively investigated in the rSey/rSey mutants. For instance, using a DiI-labeling approach, it was seen that a subpopulation of CNC cells migrating from the anterior midbrain stack up behind the malformed eye primordia rather than populate the frontonasal region; all other populations of CNC cells were seen to migrate normally (Matsuo et al.,1993; Osumi-Yamashita et al.,1997a,b). Fate-mapping evidence suggested that this population of anterior midbrain CNC migrated specifically to the lFNP, the derivatives of which were thought to be specifically lost in the mutant. Armed with this apparent correlation, investigators then sought an explanation for this stacking phenomenon, and found a specific upregulation in the HNK-1 epitope (a carbohydrate residue found on various cell adhesion molecules) in the SCE around the eye (Nagase et al.,2001). It was thus concluded that (1) loss of Pax6 resulted in upregulated, focal levels of HNK-1; (2) this traps the anterior midbrain CNC, which then fails to reach the frontonasal region; (3) deficiencies in the elaboration of the lFNP skeletal derivatives ensue.
A number of reasons, however, suggest that this is not a sufficiently complete explanation for the craniofacial skeletal defects found in the Pax6SeyN/SeyN mouse mutants. First, the murine and the rat Small eye mutants exhibit some phenotypic incongruity: for instance, the rSey/rSey mutants exhibit an ectopic cleft within the mxBA1 (Kriangkrai et al.,2006; Matsuo et al.,1993) but mouse mutant embryos do not (see Fig. 2). Second, all but a small subpopulation of Hox-negative CNC has been shown to enter the region of the optic and olfactory primordia of both the rat and murine mutant embryos (Matsuo et al.,1993; Nomura et al.,2007; Osumi-Yamashita et al.,1997a,b). This is significant because, as suggested by various neural fold rotation, regeneration, and ablation studies, the Hox-negative CNC—which includes the freely mixing forebrain crest as well as the posterior and anterior midbrain crest—appears to act as an “equivalence group” with regard to translating patterning information into morphology (Creuzet et al.,2002,2005; Kuratani,2005a,b; Le Douarin et al.,2004). Third, Pax6Sey/+ embryos exhibit a similar stacking of CNC behind the eye to the one that is seen in the rSey/rSey mutants (Kanakubo et al.,2006) and yet do not exhibit severe skeletal defects. And fourth, a long history of experimental evidence had suggested a more direct, primal role for the SCE in regional skeletal development, as exemplified, for instance, by evidence that the control of nasal capsular chondrogenesis is mediated by the regional placodal epithelium (Bell,1907; Burr,1916; Corsin,1971; LaMantia et al.2000; Reiss,1998; Toerien and Roussouw,1977; Zwilling,1940).
We therefore hypothesized that the cranial skeletal phenotype of the Pax6SeyN/SeyN mutant is not simply due to CNC migration defects but to an inability of the SCE to elaborate a fully functioning λ-junction (see Fig. 2). We posited that Pax6 is required for the competence of the SCE to manifest a mature λ-junction, and its absence vitiates the patterning information that the λ-junction usually imparts to the underlying mesenchyme: thus, the CNC receives insufficient information to yield the normal craniofacial skeletal structures associated with the junction.
To explore this hypothesis, we addressed the following issues. First, to clarify whether early embryonic phenotypic incongruity correlates with later skeletal incongruities, we describe herein the Pax6SeyN/SeyN mutant skull and compare it with that reported for the rSey/rSey. We then determine whether Pax6SeyN/SeyN embryos exhibit molecular patterning characteristics of a normal frontonasal region at late pharyngeal stages when the craniofacial primordia are patent. Finally, we directly investigate evidence of a disruption in the ontogenetic maturation of the cellular and molecular characteristics of the λ-junction due to the loss of Pax6. We show that there is a specific disruption in the ontogenetic elaboration (spatial and temporal) of SCE gene expression, before and during the formation of the mature λ-junction, of essential secreted signaling factors or their effectors (such as Fgf8, Bmp4, and Raldh3) and of transcription factors (such as Six1 and Dlx5) variably implicated in upper jaw and nasal capsular development. Thus, we demonstrate that the epithelial environment that the Pax6 mutant Hox-negative CNC encounters in the frontonasal region outside of the eye is molecularly compromised before the CNC begins to execute its morphogenetic programs. This supports a model wherein early Pax6 expression in the SCE acts as an epithelial competence factor essential for the ectoderm to engage in regional cross-talk with the CNC to establish a proper λ-junction.
Embryos were fixed at 4°C overnight in a solution of 4% paraformaldehyde and 0.2% gluteraldehyde. The embryos were then washed in phosphate-buffered saline, dehydrated in a graded ethanol series, and subsequently critical point dried, sputter coated with gold and viewed and photographed in a FEI Quanta FEG scanning electron microscope operating at 10 kV.
Differential staining of bone and cartilage was achieved following established protocols (Depew,2008). For histological studies, neonates were fixed at 4°C overnight in a solution of 4% paraformaldehyde, and then cut in 8–20-μm thick serial sections. Sections were then stained with a Gimori Trichrome stain following Depew et al., (1999).
RNA In Situ Hybridization
Digoxygenin-labelled ribroprobes for Alx4, Bmp4, Dlx3, Dlx5, Eya2, Fgf8, Msx1, Msx2, p63, Pax6, Pax7, Pixt2, Prx1, Prx2, Raldh3, Six1, Six3, Spry1, Sox10, Wnt5a, and Wnt9b were generated following standard protocols as per Depew et al., (1999).
Craniofacial Skeletal Defects of the Murine Pax6SeyN/SeyN Mutant: Similar to, But Distinct From, Those of the rat rSey/rSey Mutant
Pax6 has a known central role in cranial development though the specifics of the cranial skeletal malformations are, to date, only well documented for the rat Small eyemutant (Grindley et al.,1995; Hill et al.,1991; Hogan et al.,1986; Kaufman et al.,1995; Matsuo et al.,1993; Osumi-Yamashita et al.,1997a; Quinn et al.,1997; Roberts,1967). As with the rSey/rSey rat mutants, Pax6SeyN/SeyN mice have catastrophic malformations of their nasoethmoidal structures (Fig. 3 and Supporting Information Fig. 1). The snout is manifestly shortened and lacks naral openings. The rostral midline skeletal foundation of the skull, formed of the trabecular basal plate and its nasal septal extension, extends as a rod beyond the cavum cranii but is severely truncated and dorsoventrally compressed (see Fig. 3). Normally, this midline structure is formed of two trabecular condensations that fuse across the midline to yield a single cartilaginous element. In Pax6SeyN/SeyN mutants, however, the caudal end is abnormally bifurcated, attaching to the presumptive acrochordal cartilage at its lateral extremities: thus, the caudal trabecular ends fail to meet and a caudal gap is initially seen. This gap is typically filled in part by a rostral cartilaginous extension from the acrochordal cartilage, often giving the region a distinct “triton-like” appearance (red arrowheads, Fig. 3). This separation of the caudal trabecular basal plate has not been reported for the rSey/rSey mutants (Matsuo et al.,1993; Osumi-Yamashita et al.,1997a,b). It is likewise evident in the murine Pax6LacZ/LacZSmall eye mutants (St-Onge et al.,1997; Supporting Information Fig. 1).
The other drastic deficiency of the snout is the near agenesis of the nasal capsules, which are represented only by short, thin cartilaginous rods extending in parallel to the midline trabecular plate (black arrowheads, Fig. 3). These rods may fuse caudally to the trabecular cartilage and/or to the dysmorphic preoptic pillars (which variably reach the ala orbitalii; Fig. 3). Capsule-associated paraseptal cartilages fail to form. Olfactory bulbs and mesethmoidal structures do not form, and the rostral brain rests not on cribriform plates but on the ala orbitalii of the optic capsules which are enlarged and extend to meet at the rostral midline. This meeting generates a small dorsal midline spur (yellow arrowhead, Fig. 3c). Despite the lack of an eye or optic nerve, the ala hypochiasmatica and postoptic pillars of the optic capsules develop, helping form a rudimentary optic foramen (orange arrowheads, Fig. 3). The connection of the optic capsule with the lateral skull wall (the taenia marginalis) is usually also dysmorphic.
The regional dermatocranium is likewise compromised in the mutants. Nasal bones do not form: instead, the dorsal “nasal” region is abnormally represented by premaxillae that articulate with each other at the dorsal midline and with the frontals caudally (Fig. 3a). Ventrally, the premaxillae meet without generating a true incisive canal and normal primary palate (i.e., there are no premaxillary palatal shelves per se). Supernumerary premaxillary incisors have been reported for the rat rSey/rSey mutants as well as (in a background dependent manner) in Pax6Sey/Sey and Pax6SeyN/SeyN mice (Kaufman et al.,1995; Osumi-Yamashita et al.,1997a; Quinn et al.,1997); however, they do not characterize our Pax6SeyN/SeyN mice (see Fig. 3). Lacrimals, lacking in the rSey/rSey mutants, are present in our Pax6SeyN/SeyN mice, but they are smaller than usual. Dysmorphic maxillae form just ventral to the short, thin lateral nasal capsular rods; these may be fused to the palatines. Vomers develop, but are compressed and in close association with the maxillae and premaxillae (Fig. 3b). As a true nasopharynx fails to form, the palatal elements essentially abut the neurocranial base. The lower jaw (dentary) is relatively normal and subsequently extends well beyond the upper jaw. Moreover, there are no hypoglossal canals in the exoccipitals, as with the rSey/rSey mutants. Thus, structures associated with the λ-junction, such as external and internal choanae, nasal capsules, and the palate, are malformed or lost in the absence of Pax6.
Mesenchyme Associated With the Mature λ-Junctions of Pax6SeyN/SeyN Mutant Embryos Maintains Some Molecular Characteristics of Normal Frontonasal Regions
Normal patterning and morphogenesis of the complex skeletal structures of the skull requires an exquisite, reciprocal cross-talk between the embryonic cephalic epithelia and mesenchyme. The mesenchyme associated with the mature λ-junction expresses numerous genes known to be essential for the development of the cranial skeleton of the region. As the Pax6SeyN/SeyN mutant embryos clearly possess on either side a solitary, undivided ectomesenchyme-filled frontonasal mass (FNM; Fig. 2), and frontonasal process-associated skeletal elements such as premaxillae are evident in mutant skulls (see Fig. 3), it is clear that the FNM of the Pax6SeyN/SeyN E10.5 embryo would continue to express some of these essential genes.
Previous DiI-labeling experiments suggested that rostral CNC ectomesenchyme in mutant embryos reaches the frontonasal region at embryonic times point before the advent of the morphogenesis of the olfactory placode (Matsuo et al.,1993; Osumi-Yamashita et al.,1997a). To reconfirm this, we examined the expression of Alx3, a marker of this population. Although the optic primordia of the mutant embryos is relatively enlarged and dysmorphic by E9.25, Alx3 positive cells encircle the eye and fill the anterior cephalic region associated with the olfactory placodes and λ-junction in both mutant and wild type embryos (see Fig. 4). Thus, Hox-negative CNC blankets the area subjacent to the SCE, which is in line with previous labeling experiments.
We then analyzed the expression of a number of genes implicated in regional skeletal patterning, including the patterns for the transcription factors Msx1, Msx2, Prx2, and Alx4 and for the secreted signaling molecule Wnt5a (Beverdam et al.,2001; Satokata and Maas,1994; Satokata et al.,2000; ten Berge et al.,1998; Yamaguchi et al.,1999). Pax6SeyN/SeyN E10.5 embryos express Msx1, Msx2, Wnt5a, Prx2, and Alx4 in the mesenchyme associated with the λ-junction, including in their FNMs (see Fig. 4). Normally Msx1, Msx2, and Wnt5a are expressed continuously, without a gap, in and between the mxBA1, lFNP, and mFNP: such continuity is maintained between the mxBA1 and FNM in the mutants (Fig. 4c–h). Between the mxBA1 and the FNP, Prx2 and Alx4 typically exhibit decreased expression at the center of the λ-junction (at the central ramus); such a gap of expression is likewise maintained in mutant embryos (Fig. 4i–l′). Notably, a foci of Prx2 expression is seen in this gap between the FNM and mxBA1 of the Pax6SeyN/SeyN embryos (yellow arrowheads Fig. 4j, j′). The topology of this position suggests the possibility that this foci represents an abortive attempt to generate a lFNP in the mutant. Overall, these expression patterns support the notion that, in the absence of normal Pax6, the region of the mature λ-junction retains some level of normal patterning and epithelial-mesenchymal cross-talk.
The Mesenchyme Associated With Mature λ-Junctions of Pax6SeyN/SeyN Mutants Shows Disruptions of Expression Patterns of Transcription Factors Disparately Expressed in the FNP
Just as the presence of bilateral, solitary FNMs, and their specific skeletal derivatives, suggested that some patterning must be retained in Pax6SeyN/SeyN embryos, the absence of paired frontonasal processes (i.e., both a lFNP and a mFNP) suggested that expression of genes typically disparately expressed between the mFNP and lFNP mesenchyme around E10.5 would also be altered. Indeed, as expected, this is the case (see Fig. 5). We confirm previous reports (La Mantia et al.,2000; Nagase et al.,2001) that Pax7, which is expressed in the lFNP mesenchyme but not the mFNP mesenchyme, is not expressed in the cephalic mesenchyme of Pax6SeyN/SeyN embryos (Fig. 5b). We also found that the low levels of Prx1 transcript that are usually seen in the lFNP are likewise lacking in the topographically comparable region in Pax6SeyN/SeyN embryos (Fig. 5d). Perioccular, lFNP-associated expression of Pitx2 and Eya2 are lost or severely downregulated (Fig. 5e–h). Moreover, the expression of Sox10, which typically delineates the trigeminal ganglia that sends ophthalmic and maxillary axons around the forming eye and toward the frontonasal processes, reveals an apparent lack of axon extension around the region of the eye at E10.5 (Fig. 5j).
Epithelial Expression Patterns of Transcription Factors at the Mature λ-Junctions Are Altered in Pax6SeyN/SeyN Mutants
To further address the notion that the lFNP rather than the mFNP is specifically depleted in mutant embryos, we examined the expression of epithelial markers disparately arrayed in the various rami of the mature λ-junction, namely Dlx3, Dlx5, Pitx2, Six3, and p63 (Lagutin et al.,2003; Lin et al.,1999; Liu et al.,2003; Lu et al.,1999; Mills et al.,1999; Yang et al.,1999). Dlx3 has been implicated in the formation of epithelial structures (Morasso et al.,1996; Radoja et al.,2007), whereas Dlx5 has been shown to be essential for nasal capsular development (Depew et al.,1999). In the Pax6SeyN/SeyN mutant embryos, Dlx3 epithelial expression is abnormally restricted to the central rami (Fig. 5k,l). Although most of the SCE expresses Dlx5 early in embryogenesis, by E10.5 expression is normally restricted to rimming the internal nasal pit with the greatest expression medially associated with the presumptive region of the future vomeronasal organ (Fig. 5m,m′). In Pax6SeyN/SeyN embryos, Dlx5 expression is extensively restricted to just the topology of where one would expect the forming VNO to have been if the FNM of the mutant represented only the mFNP component of a normal embryo (Fig. 5n,n′).
In addition, we find that the perioccular expression of Pitx2 is altered in the mutant embryos, while the maxillary and substalk oral expression is maintained (Fig. 5f′). p63 is usually expressed diffusely throughout the BA and FNP epithelium at E10.5, but has increased intensity in the maxillary, odontogenic, medial pit, and lateral pit rami as well as in the optic end of the nasolacrimal groove (Fig. 5o,o′). In Pax6SeyN/SeyN mutant embryos, diffuse expression in the BA is evident; however, only a strong line along the presumptive odontogenic ramus is retained (Fig. 5p,p′). Moreover, a focal increase in expression is seen rostral to mxBA1 and ventral to the malformed optic primordia (Fig. 5p,p′; yellow arrowheads). Finally, Six3 fails to be transcribed in epithelium of the olfactory pit lines of mutant embryos as has been suggested elsewhere (Fig. 5r; Purcell et al.,2005). Thus, by E10.5, the ectoderm of the λ-junction is characterized by the loss or misexpression of key genes required for normal craniofacial development.
Aberrant Expression of Bmp4, Fgf8, Raldh3 and Wnt9b in the Mature λ-Junction
As the gene expression patterns of a number of the transcription factors described above are known to be induced and/or maintained by coordinated expression of a number of secreted signaling factors likewise known to be essential for craniofacial development (reviewed in Depew et al.,2002b), we examined the patterns of expression of signaling factors, or of effectors of their signaling, for the Bmp, Fgf, Wnt, and retinoic acid pathways. The spatiotemporal manifestation of retinoic acid signaling has been addressed in a number of studies of the development of the optic and olfactory systems, including studies involving the loss of Pax6. Raldh3, involved in synthesis of the signaling molecule retinoic acid, is normally expressed at the heart of the developing λ-junction. It has been shown to be absent in some Small eye mutants (Anchan et al.,1997; Enwright and Grainger,2000; Suzuki et al.,2000). Similarly, we found it is absent in Pax6SeyN/SeyN mutant embryos at E10.5 (Fig. 6b).
Wnt9b, essential to the formation of a normal λ-junction (Juriloff et al.,2006), is typically concentrated around the center of the mature λ-junction, broadly extending along the nasolacrimal ramus to the optic lens, and along the outside margins of the lateral pit, medial pit, and odontogenic lines (Fig. 6c,c′). This pattern is disrupted in Pax6SeyN/SeyN mutant embryos, where the nasolacrimal and odontogenic rami (diffusely) express Wnt9b. Notably, a small foci of expression is seen in the epithelium covering the rostroventral end of the aberrant optic primordia of mutant embryos (Fig. 6d,d′).
Bone morphogenic protein (Bmp) signaling has been highly implicated in the development of the region of the λ-junction (Abzhanov et al.,2004; Barlow and Francis-West,1997; Foppiano et al.,2007; Jiang et al.,2006; Lee et al.,2001; Liu et al.,2005; Wu et al.,2004). In particular, Bmp4 has a dynamic spatiotemporal pattern of expression in both regional epithelium and mesenchyme. At E10.5, Bmp4 is usually expressed in epithelium in the center of the junction, with extensions into the maxillary, mid-outer medial and lateral pits, and odontogenic rami, as well as on either side of the nasolacrimal ramus (Fig. 6e,e′). It is also expressed in the dorsal aspect of the developing eye. Contrary to earlier reports (Furuta and Hogan,1998), we find that Bmp4 expression is upregulated in Pax6SeyN/SeyN mutant embryos, especially in the cells overlying the malformed eye primordia (Fig. 6f,f′). Moreover, expression is higher, but slightly patchy, around the central, median, and maxillary rami of the junction. The maxillary end of the FNM of Pax6SeyN/SeyN mutant embryos also expresses variable levels of Bmp4, with the highest levels extending along the odontogenic line. Notably, there is a focally increased level of Bmp4 expression in the area between the FNM, the mxBA1 and the eye (green arrowhead Fig. 6f,f′) that roughly corresponds to where focal expression of Prx2 (see Fig. 4) and p63 (see Fig. 5) are also found.
Like Bmp4, Fgf8 has also been highly implicated in cranial development, and has a dynamic spatiotemporal pattern of expression in the regional epithelium (Abu-Issa et al.,2002; Creuzet et al.,2004; Frank et al.,2002; Hu et al.,2003; Kawauchi et al.,2005; Riley et al.,2007; Storm et al.,2006; Szabo-Rogers et al.,2008,2009; Trumpp et al.,1999; Tucker et al.,1999). In the mature λ-junction of an E10.5 mouse, Fgf8 is highly expressed in the medial and lateral pit lines (Fig. 6g, g′), as well as in the oral epithelium associated with the junction of mxBA1 and mdBA1. At this time, lower levels of Fgf8 are seen in the odontogenic ramus and expression in the central/median ramus of the junction is just beginning to appear. We find that Fgf8 expression at this stage in Pax6SeyN/SeyN mutants is greatly altered: while a discrete line of expression is detected along the presumptive odontogenic ramus, expression in the epithelium of the FNM is restricted to a few scattered cells lateral to the odontogenic line (Fig. 6h,h′). Moreover, there is an abnormal, small foci of Fgf8 expression in the median body just below the eye (yellow arrowhead, Fig. 6h,h′). Sprouty 1 (Spry1), an Fgf8 responsive gene believed to mediate, in part, Fgf signaling, is also abnormally expressed in the mutant embryos (Fig. 6i–j′). Unlike in the wild type, there is no Spry1 expression in the epithelium of the FNP; conversely, we found increased Srpy1 signal in the tissue surrounding the optic primordia and stalk (yellow arrowheads, Fig. j,j′).
Failure in the Normal Ontogenetic Progression of the SCE Into a Mature λ-Junction Begins Early
We posited that the etiology of the skeletal defects in Pax6SeyN/SeyN mutants stemmed from a disruption of the signaling interactions at the λ-junction. We in turn presented evidence that, while retaining some regional patterning information, the normal formation of a mature λ-junction at E10.5 has been disrupted in Pax6SeyN/SeyN mutant embryos as reflected by a number of genes that are regionally mis-expressed in the mutant embryos at this time.
We were interested in addressing the essential issue of the ontogeny of the disruption of the normal maturation of the λ-junction, specifically in the molecular and cellular environment that the Hox-negative CNC encounters as it engages in a developmental dialogue with a regional epithelia depleted of Pax6. We therefore examined wild type and mutant embryos at a time (E9-9.25) when the CNC has arrived but has not begun to extensively proliferate and generate noticeable primordia at the λ-junction.
As might be anticipated by its absence at E10.5, Raldh3 expression, which at E9.0 normally encompasses the entire region of the presumptive central/median and nasolacrimal rami plus much of the associated maxillary and pit epithelia, is entirely absent in the mutant (Fig. 7b). At E9.25 the nascent mxBA1 normally meets the suboptic tissue and the presumptive frontonasal ectoderm at an angle: the maxillary portion of this angle typically expresses a high level of Bmp4 while a lower level of expression is seen in the suboptic epithelium (Fig. 7c,c′). Moreover, the optic placodes typically express Bmp4, though in a graded fashion where the highest levels are dorsally restricted. Presaging what is seen at E10.5, in mutant embryos at E9.25 the epithelium covering the presumptive optic tissue displays uniformly increased levels of Bmp4 expression, both in intensity and in topographic extent (Fig. 7d,d′). The maxillary and suboptic expression is less obviously altered. However, a change in expression level is also observed in the region that will yield Rathke's pouch (blue arrowhead, Fig. 7d′).
We further find that the early ontogeny of Fgf8 expression in the SCE that eventually forms the λ-junction is severely altered (Fig. 7e–h). At E9.0, Fgf8 mRNA is normally highly detectable in the BA1 ectoderm and the developing commissural plate (CP) that derives from the anterior neural ridge (also Fgf8 positive). Notably, Fgf8 is also typically expressed in a more diffuse manner in the ventrolateral ectoderm (here referred to as the VLE) to the CP—epithelium that mostly lies between the CP and the presumptive, forming olfactory placode. In the Pax6SeyN/SeyN mutant at E9.0, Fgf8 expression in the CP is slightly broader, as is that in the VLE, than in the wild type embryo (red arrowhead Fig. 7f). One day later, olfactory placodogenesis has normally already occurred, the placode has begun to invaginate, and the underlying mesenchyme has begun to differentially proliferate to form nascent mFNP and lFNP. Typically, at this time in wild type embryos, the VLE continues to express Fgf8, as do the medial and lateral pit lines of the forming frontonasal processes (Fig. 7g). The maxillary ramus is highly Fgf8 positive although the central/median ramus of the maturing λ-junction still remains Fgf8-negative. By E10.0 Fgf8 expression in the λ-junction epithelium of Pax6SeyN/SeyN mutant embryos is remarkably different (Fig. 7h). Expression in the VLE is restricted to a small patch just lateral to the Fgf8-positive CP, and rather than intense expression in the epithelium of the pit lines of the forming frontonasal processes, the solitary FNM of the mutant only exhibits a small discrete patch of expression just lateral to the VLE. Strikingly, the epithelium at the presumptive central/median ramus exhibits a crescent-shaped swath of Fgf8-positve cells. These discrete patches of Fgf8 expression clearly foreshadow the topographically maintained—although further restricted—patterns of expression found half a day later at E10.5 (see Fig. 6).
Just as we find that the ontogenetic progression of expression of these secreted signaling molecules is disrupted in Pax6SeyN/SeyN embryos, we find that the expression of a number of transcription factors previously implicated in the regulation of the cephalic ectoderm and of cranial skeletal formation, including Six1 and Dlx5, are also altered before and during placodogenesis stages (Fig. 7i–n). For instance, we find a drastic decrease in the expression of Six1 in the olfactory placodal and suboptic epithelium at both E9.0 and E10.25 (red arrowheads, Fig. 7l,l′,n).
Thus, lack of Pax6 results in a transcriptionally deficient SCE already by E9-9.25 which leads to compounded defects by E10.5 (see Fig. 8). The clear disruption of the maturation process of the region of the λ-junction—as reflected for instance by an inability to properly express (in appropriate spatial and temporal manners) Raldh3, Bmp4, Fgf8, Dlx5, and Six1—indicates that the absence of Pax6 in the SCE vitiates the manifestation of the λ-junction, a region of critical importance for the formation of nasal and upper jaw skeletal structures.
Craniofacial patterning and morphogenesis requires an exquisitely timed and positioned reciprocal cross-talk between the pertinent embryonic epithelia and mesenchyme. This is particularly true of the development associated with the morphologically and molecularly complex λ-junction, whose organization reflects the future positions of the internal and external choanae and the formation of the upper lips, nasal capsules, upper incisors, and primary and secondary palates. Elaboration of this cross-talk is manifested in the induction and maintenance of intricate patterns of gene expression. The spatial and temporal details of these patterns underlie the precise translation of patterning processes and information into appropriate skeletal morphologies.
Outside of dental enamel, cephalic epithelial tissues involved in this dialogue are not known to directly yield cranial skeletal structures: their importance to skeletal form hence largely lies in their influences on the CNC mesenchyme that forms the cranial tissues. Understanding the ontogeny of such influences on the CNC, including the inception and maturation of the λ-junction from the SCE, remains a key endeavor. Evidence for the developmental importance of the SCE for cranial skeletal development has come from a number of directions, including experimental embryological studies involving ectoderm extirpation, tissue recombination, and bead implantation. For example, embryological studies manipulating portions of the nascent SCE in chicks have identified a subregion, identified as the “facial ectodermal zone,” and factors expressed therein such as Fgf8, Shh, and Bmp4, as critical for the development of subcomponents of the cranial skeleton (Hu et al.,2003; Marcucio et al.,2005).
Although generally not emphasized, phenotypic analyses of mutations of genes expressed in the ectoderm (and not the associated CNC) have also clearly demonstrated the necessity of a properly informed and competent SCE (Depew et al.,2002b). The essential issue in such studies has been how to detail and explain the nature of the noncell autonomous actions exerted by these genes on the CNC: this has particularly been true in the case of the Pax6 mutants. Previous studies of Pax6 mutations have suggested that this occurs via an upregulation in expression of HNK-1, a putative adhesion-related molecule, and that this inhibits the migration of a subpopulation of midbrain CNC to the frontonasal region which subsequently impairs the development of specific cranial skeletal elements (Matsuo et al.,1993; Osumi-Yamashita et al.,1997a,b).
An inherent premise of this “inhibited migration” hypothesis has been that the CNC reaching the lFNP is “prepatterned” before its final destination: absence of structure is due to absence of a strictly foreordained population of cells. However, accumulating evidence, including that suggesting the “equivalence” of the Hox-negative CNC and that demonstrating a specific role for the SCE in the formation of regional cranial skeletal structures, is at variance with this premise. Despite a portion of midbrain CNC failing to migrate past the eye, by E9-9.25 there are equipotent midbrain and forebrain CNC that do reach the frontonasal region (see Fig. 4): without recourse to “pre-patterned” crest, there is no reason why fungible CNC that do migrate to the frontonasal region should fail to compensate for the absent fraction. Such reasoning led us to challenge the established view that the etiology of Pax6 mutant cranial skeletal deficiencies lies essentially in a lost subpopulation of CNC. Following from the “Hinge and Caps” model, we hypothesized that the Hox-negative CNC in the Small eye mutants in general, and of the Pax6SeyN/SeyN mutants in particular, fail to generate a full complement of regional structures because the informative epithelial environment of the SCE that they have migrated into contact with has been altered to the point that it cannot support and elaborate the fully functioning λ-junction: this vitiates the complex epithelial-mesenchymal dialogue that normally regulates the formation of the nasal capsules, trabecular basal plate, and upper jaws.
To address this “dysfunctional λ-junction” hypothesis, we first compared the morphologic phenotype reported for the rSey/rSey mutants with that of the Pax6SeyN/SeyN mutants. In most significant respects, the Pax6SeyN/SeyN skull defects mirror those reported for the rSey/rSey: for instance, perinates of both mutants have similar catastrophic malformations of their nasoethmoidal structures, possessing truncated trabecular basal plates and lacking naral openings and elaborated nasal capsules. In the Pax6SeyN/SeyN mutants, however, the caudal end of the trabecular basal plates is abnormally bifurcated, giving the region a distinct triton-like appearance (see Fig. 3). Although this separation of the caudal trabecular basal plate has not been reported for the rSey/rSey mutants (Matsuo et al.,1993; Osumi-Yamashita et al.,1997a), we find that this defect is also evident in perinates of a disparate murine mutant (Pax6LacZ/LacZ; Supporting Information Fig. 1).
Notably, the cranial skeletal defects encountered in both rodent mutants correlate with structures associated with the λ-junction. This strongly suggests that the λ-junction itself is deficient. Indeed, we argue that lack of Pax6 alters the formation and elaboration of a normal λ-junction, thereby altering skeletal morphology, for a number of reasons. First, it is clear that the epithelium of the mature λ-junction in E10.5 Pax6SeyN/SeyN mutant embryos fails to properly express critical markers of the components of the junction. This includes transcription factors, such as Dlx3, Dlx5, p63, and Six3, that have been shown to be necessary for proper cranial skeletal development and morphology.
Second, based largely on the morphological criteria of a lack of a thickened placodal epithelium, it has previously been suggested that the murine Small eye mutants fail to form olfactory placodes (Hogan et al.,1986; Grindley et al.,1997). Although the specifics of the correlation between the absence of a thickened epithelium denoting an olfactory placode and loss of nasal capsular structures has yet to be fully investigated and detailed, extirpation studies have suggested that the epithelium of the olfactory placode is essential to the formation of at least the nasal capsule. Without an olfactory placode, moreover, the morphologic events surrounding the formation of the nasal pits fail to manifest.
We have herein demonstrated that, in the absence of Pax6, gene expression patterns in the SCE of placode-essential transcription factors are disrupted in and around the presumptive olfactory placode epithelium before and during the initiation of placodogenesis. This is exemplified by the loss and/or downregulation of Six1 and Dlx5 in the SCE at E9-9.25 (see Fig. 7). Thus, Six1 and Dlx5, each a contributor to the mature λ-junction, are already mis-expressed or downregulated during placodogenesis and the initiation of λ-junction formation—a period when the crest has arrived and is ready to begin interpreting its environment. Notably, loss of function studies show that both of these genes, in addition to being essential for some aspect of olfactory placodal development, are required for the proper development of the nasal capsules (Depew et al.,1999; Laclef et al.,2003; Zou et al.,2004). This further correlates Pax6 expression in the SCE with the loss of placode development and activity, including placodal influence on the formation of the λ-junction.
Third, the ontogenetic progression of the λ-junction is perturbed by early changes in the patterns of genes encoding numerous secretory signaling molecules, and/or their effectors, known to be critical for craniofacial patterning and development. These include molecules involved in RA, Bmp, and Fgf signaling. Raldh3 mutant mice exhibit defects correlated with a disruption of the λ-junction, including the agenesis of the eye-associated Hardian glands, choanal atresia, and hypoplasia of the lateral nasal capsule (particularly the turbinals) (Dupe et al.,2003). We (and others) have failed to detect cephalic Raldh3 in Small eye mutants (Figs. 6 and 7; Enwright and Grainger,2000; Suzuki et al.,2000), placing Raldh3 at the center of previous observations that Pax6SeyN/SeyN mutant FNP exhibit severely decreased RA signaling (Anchan et al.,1997).
Bmp4 is ontogenetically dynamically expressed in the SCE epithelia, including the λ-junction, and has been shown to be involved in the epithelial–mesenchymal and epithelial–epithelial interactions that regulate gene expression patterns and overall craniofacial development (Abzhanov et al.,2004; Barlow and Francis-West,1997; Foppiano et al.,2007; Jiang et al.,2006; Lee et al.,2001; Liu et al.,2005; Wu et al.,2004). As with RA signaling, regulated levels of Bmp4 appear to be critical to proper craniofacial development: in this regard, we find clear spatial misexpression of Bmp4 in and around the mature λ-junction (E10.5) of Pax6SeyN/SeyN mutant embryos (see Fig. 6). Although increased overall (in particular in the SCE associated with the abortive eyes), Bmp4 expression appears patchier and fails to normally delineate numerous rami of the λ-junction in mutant embryos. Contrary to earlier reports (Furuta and Hogan,1998), we find that E9-9.25 mutant embryos show a clear increased level of Bmp4 expression overlying the presumptive optic lens placode. The altered ontogeny of expression of this secreted signaling factor reflects the ontogenetic alteration of the λ-junction itself, and likely contributes to the craniofacial defects observed.
FGF superfamily signaling has also been implicated in numerous developmental processes, including cranial patterning and growth. Aberrant FGF signaling due to mutation in ligands or receptors, leads variously to craniosynostoses and cleft lip and palate (Dode et al.,2007; Riley et al.,2007). There is moreover strong evidence for the involvement of Fgf8 in the development of the λ-junction and the structures informed by it. In mice, for instance, Fgf8 has specifically been shown to regulate craniofacial pattern and development (reviewed in Depew et al.,2002b). Notably, mice which have either lost Fgf8 in the SCE due to conditional inactivation of a floxed allele (Kawauchi et al.,2005) or which bear hypomorphic alleles of Fgf8 (Abu-Issa et al.,2002; Frank et al.,2002) demonstrate the necessity of Fgf8 for olfactory placode and nasal capsule development. This work has been augmented by experimental embryological studies in birds where over- or under-expression of Fgf8 has been shown to affect the pattern and development of the craniofacial skeleton (Abzhanov and Tabin,2004; Szabo-Rogers et al.,2008). Significantly, the dynamic ontogenetic choreography of Fgf8 expression in the SCE is particularly out of step in Pax6SeyN/SeyN mutant embryos (see Fig. 7). In its ontogeny, as with that of Bmp4 and Raldh3, Fgf8 expression reflects a dynamic change in the signaling environment to be encountered by the Hox-negative CNC responsible for generating the structures malformed or lost in the Pax6SeyN/SeyN mutants.
A preponderance of molecular and morphologic evidence therefore indicates that Pax6 is essential for the competence of the SCE to engage in and execute the exquisitely choreographed spatiotemporal dialogue between the cephalic epithelia and mesenchyme that determines the translation of patterning information into the morphogenesis of the distinct, intricate skeletal morphologies of the skull. Its role in parsing the early SCE into appropriate placodal regions has previously been demonstrated and its interplay with Fgf8 and Dlx5 (among others) during placodogenesis has been appreciated (Bailey et al.,2006; Bhattacharyya et al.,2004). We have demonstrated, however, that Pax6 involvement in craniofacial development in general, and in the maturation of the SCE in particular, is more intricate and pervasive than perhaps its roles just in placodogenesis alone may have previously suggested. Pax6 is normally expressed early (E8-E8.5) in the SCE when the CNC is beginning its migration to the region of the future λ-junction: loss of Pax6, however, sets in motion a cascade of altered, reduced, or absent expression patterns of genes essential for normal craniofacial skeletal development. This includes, among others, early changes in the epithelial expression of genes encoding the placode-essential transcription factors Six1 and Dlx5 and the secreted signaling molecules Bmp4 and Fgf8 (as well as of that for the RA synthesizing molecule Raldh3), at embryonic stages when the CNC has just finished migrating and placodogenesis and the initiation of the λ-junction have just begun (see Fig. 8). Either directly or indirectly, this has a knock-on effect with regard to the elaboration of a normally functioning λ-junction.
We share with previous investigators the notion that loss of Pax6 leads to changes in the SCE, although perhaps not the nature and significance of all such changes. We further believe that in order for the inhibited migration hypothesis to be a valid explanation for the etiology of the craniofacial deficits seen in Small eye mutants, the anterior midbrain CNC has to either be extensively prepatterned with regard to its potential to yield distinct skeletal morphologies or else gather indispensable patterning information around the eye that it then normally passes on to the frontonasal epithelia and/or forebrain CNC—which in turn uses this information in the formation of the λ-junction. Accumulating evidence argues against the former, while the latter has yet to be conclusively demonstrated. In addition, we reconfirm here that significant CNC is present in the Pax6SeyN/SeyN mutant frontonasal region and that this neural crest maintains a degree of normal molecular patterning, including some gene expression characteristic of both the medial and lateral FNP. This suggests that cells competent to form the lFNP are present. We therefore adduce the molecular disruption of λ-junction ontogeny, due to the early loss of SCE Pax6 expression, as underling the failure of the FNP in Pax6SeyN/SeyN mutants to be fully elaborated and generative of the appropriate cranial skeletal elements.
The authors thank current and past members of CFD and the Depew lab for discussions, and K. Yun and JLR Rubenstein for the gift of the Pax6SeyN mutant line. They also thank MA Basson, P Crossley, JLR Rubenstein, and L Selleri for various riboprobe plasmids. Finally, they thank many investigators whose work informed their own but whose citations they have not been able to include.