Dlx5 and Dlx6 expression in the anterior neural fold is essential for patterning the dorsal nasal capsule

The formation of craniofacial structures results from interactions between Hox-negative cephalic neural crest cells (CNCCs) and adjacent instructive epithelia (Bee and Thorogood, 1980; Couly et al., 2002). The vertebrate nose builds upon an embryonic cartilaginous frame, the nasal capsule, an evolutionary-conserved structure composed of ventral and dorsal components: the mesethmoid and ectethmoid cartilages, respectively. In avians, the mesethmoid supports upper beak formation, whereas the ectethmoid comprises elements of the olfactory system, including the lamina cribosa, the crista galli apophysis and the conchae.

Ablation and grafting experiments have demonstrated the importance of signalling from the endoderm to CNCCs for the specification and patterning of different craniofacial elements (Couly et al., 2002; Couly et al., 1993). In particular, we have shown that Shh-Gli1 signalling from the anteriormost territory of the endoderm, which we named ‘EZ-I’ (Benouaiche et al., 2008) to rostroventrally migrating CNCCs is required for the edification of the chick mesethmoid cartilage. By contrast, the molecular and cellular mechanisms driving ectethmoid formation have not been elucidated.

In the chick embryo, the ectethmoid is formed by CNCCs emigrating from the prosencephalic ridges of the neural plate. These cells follow preferentially a subectodermal route, contact the presumptive olfactory ectoderm at Hamburger Hamilton stage 10 (HH10) and remain in its close proximity up to HH18 (Couly and Le Douarin, 1990; Couly et al., 1993; Le Douarin et al., 2007). Then, neural crest-derived chondrogenic ectomesenchyme accumulates within the nasal prominences. CNCC depletion at the neurula stage prevents any craniofacial chondrogenesis, including that of the nasal capsule. The observation that CNCCs are not specified before migration and constitute an equivalence group (Couly et al., 1992) implies that instructive signals are also required for ectethmoid formation. As the ectethmoid develops in the absence of the endoderm (Benouaiche et al., 2008; Couly et al., 2002; Couly et al., 1998; Ruhin et al., 2003), the patterning instructions may reside in other territories neighbouring CNCCs, such as the prospective olfactory ectoderm.

Starting at the end of gastrulation in amniotes, the presumptive olfactory ectoderm expresses the homeobox genes Dlx5 and Dlx6, which play an important role in olfactory-fate specification of placodal precursors (Acampora et al., 1999; Bhattacharyya and Bronner-Fraser, 2008; McLarren et al., 2003; Pera and Kessel, 1999; Pera et al., 1999; Yang et al., 1998). In mouse, genetic lineage tracing using lacZ reporter activity shows persistent Dlx5 expression throughout nasal differentiation (Acampora et al., 1999). Maintenance of Dlx5 expression is necessary for the acquisition of olfactory fate in both mouse and chick embryos (Bailey et al., 2006; Bhattacharyya et al., 2004; Bhattacharyya and Bronner-Fraser, 2004; Bhattacharyya and Bronner-Fraser, 2008; Depew et al., 1999; Long et al., 2003; McLarren et al., 2003; Price et al., 1991; Quint et al., 2000; Simeone et al., 1994; Solomon and Fritz, 2002; Szabo-Rogers et al., 2008; Szabo-Rogers et al., 2009). Furthermore, Dlx5-null neonates display nasal capsule hypoplasia, which is often asymmetric (Acampora et al., 1999; Depew et al., 1999; Levi et al., 2003; Long et al., 2003). The double inactivation of Dlx5 and Dlx6 further accentuates nasal capsule defects, but the phenotype has not been yet described in detail (Beverdam et al., 2002; Depew et al., 2002; Levi et al., 2003; Robledo et al., 2002). Noticeably, nasal defects in Dlx5/6^−/− heads are not shared with either single or compound mutations of other Dlx family members (Depew et al., 2005).

A recent study has demonstrated that the olfactory pit plays a late role in controlling the morphogenesis of skeletal elements of the avian nasal capsule (Szabo-Rogers et al., 2009). These data were obtained from embryos in which CNCCs had already migrated into the developing olfactory territory and do not provide information about the initial interactions directing CNCCs towards ectethmoid formation. As the olfactory pit derives from the Dlx5/Dlx6-positive anterior part of the neural fold, we decided to explore the possibility that this territory could be endowed with early inductive capacities. Using mouse mutants and surgical and molecular analyses of chick embryos, we demonstrate that an anterior region of the neural fold ectoderm (NF-ZA) is the territory responsible for ectethmoid induction through a Dlx5/Dlx6-dependent mechanism.

We conclude that the complete nasal capsule results from the juxtaposition of two independent structures formed by CNCCs competent to respond to different endodermal or ectodermal cues.

All animal procedures were approved by national and institutional ethical committees. Fertilised eggs were obtained from Morizeau Farms, France (chicken, Gallus gallus) or Cailles de Chanteloup Farms, France (quail, Coturnix coturnix japonica) and incubated at 38°C in a humidified atmosphere to reach the appropriate Hamburger Hamilton (HH) developmental stage (Hamburger and Hamilton, 1992; Teillet et al., 1998).

The genotypes of Dlx5;Dlx6 mutant mice and embryos were determined by PCR using the Dlx5;Dlx6-mutant allele-specific primers as reported previously (Beverdam et al., 2002). Homozygous mutant embryos were obtained by breeding Dlx5;Dlx6 heterozygous parents. The developmental stage of embryos was determined according to the detection of vaginal plugs, considering noon of the day of plug observation as E0.5.

All products were from Sigma-Aldrich (France) unless otherwise stated. Embryos were collected and dissected at room temperature in phosphate-buffered saline (PBS). Immunoperoxidase detection was performed as previously described (Couly et al., 2002). Quail nuclei were detected with the quail-specific monoclonal antibody QCPN at 1/500 (obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA 52242). Dlx5 was immunodetected using the anti-DLX5 HPA005670 rabbit polyclonal antibody from Sigma diluted 1:100 in PBS gelatin 0.2%. Whole-mount skeletons were visualised according to standard staining protocols using Alcian Blue for cartilage and Alizarin Red for bone.

Identification of different skeletal cartilaginous components of the nasal capsule in both mouse and chick embryos was performed on the basis of morphological features (overall shape and size; presence or absence of nasal cavities and turbinates; presence of the olfactory epithelium associated with the ectethmoid; presence of the olfactory nerve) and positional relations with respect to surrounding structures such as the nasal or the premaxillary bones.

Experiments were carried out in ovo on windowed chick embryos at neurula stages (Couly et al., 1998). Sub-blastodermal injections of India Ink (Pelikan, France) or Fast Green were used to delineate the embryos. Unilateral or bilateral excisions were first performed on each sides of the neurectodermal ridge. The NF-ZA extends for 150 μm anteroposteriorly, starting at 150 μm from the rostralmost terminus of the embryo (anterior neural ridge or presumptive territory of the adenohypophysis). At the five-somite stage, the NF-ZA is 150 μm rostral to the cephalic neural crest-producing domain (see Fig. S1 in the supplementary material). Operated embryos were then incubated up to 13 days (until HH40) for chondroskeletal analyses.

Grafts of NF-ZA were obtained from five-somite quail embryos following the same dissection procedure described for NF-ZA ablation in the chick. NF-ZA grafts were transplanted into chick embryos at the five-somite stage in ovo after performing a unilateral ectodermal incision at the appropriate axial level. The graft was placed either in the anteroventral mesenchyme (below the level of the prosencephalon) or into the presumptive mesenchyme of the first pharyngeal arch (PA1). In some cases, a simultaneous graft of NF-ZA and EZ-I (Benouaiche et al., 2008) was transplanted into the prosencephalic domain of the mesenchyme. Embryos were fixed and sectioned 24 hours after operation (HH14) to detect quail nuclei using the QCPN antibody, or at HH35-40 for chondroskeletal preparations.

siRNA were designed against chicken Dlx5 and Dlx6 transcripts (sequences upon request). For each gene, sets of three siRNA were synthesized (Sigma-Aldrich), including one with a fluorescein (Dlx5) or rhodamin (Dlx6) tag. The siRNA were prepared according to manufacturer instructions, pooled to yield a final 50 μM Dlx5/6 siRNA mix, supplemented with 1 μg/ml Fast Green tracer dye (Sigma-Aldrich) in water. A scrambled siRNA was used as a control. Two microlitres of siRNA solution were slowly deposited under the vitelline membrane in the anterior domain of the developing neurula. Gold electrodes connected to a BTX 830 (Harvard Apparatus) were laid 3 mm apart at various positions above the vitelline membrane along the anterior neural folds. Four pulses, 500 ms apart, of 20 V for 50 ms were delivered. A 10 μl drop of tyrode was applied to cool down the embryo, then the egg was taped and reincubated.

Electroporation efficiency was monitored using a fluorescence binocular microscope, showing fluorescence-filled cells in the anterior neural fold (data not shown). Subsequent immunodetection of Dlx5 at various stages post-electroporation verified the reduction of targeted gene expression (see Fig. S2 in the supplementary material).

Embryos were collected at various stages and processed for either paraffin sectioning or whole-mount skeletal preparation as described earlier.

For paraffin sectioning, embryos and foetuses were fixed in Bouin's solution (75% saturated picric acid, 8% formol, 5% acetic acid) for 5 days (E14.5 and E18.5). After fixation, E18.5 foetuses were decalcified by immersion in Jenkin's solution (48 hours in 4% hydrochloric acid 37%, 3% acetic acid, 10% chloroform, 10% deionised H₂O, 73% ethanol). Fixed embryos were then dehydrated in ethanol, equilibrated in SafeSolv (Labonord, Templemars, France), embedded in paraffin and oriented in blocks. Paraffin blocks were then sectioned at 8-12 μm using a Leica RM2235 microtome. Serial sections of E18.5 foetuses were subjected to Mallory trichromic staining for anatomical analysis (Sato et al., 2008).

The nasal capsule of Dlx5/6^−/− mouse embryos is reduced to an anterior cartilaginous bar (Beverdam et al., 2002; Robledo et al., 2002). To further analyze the early contribution of Dlx5 and Dlx6 to nasal development, we re-examined chondroskeletal and dermatoskeletal preparations of these mutants. As previously reported, the simultaneous inactivation of Dlx5 and Dlx6 results in the transformation of mandibular into maxillary structures (Beverdam et al., 2002; Depew et al., 2002) with embryos displaying symmetric jaws. From E14.5 onwards, the vestigial nasal capsule is reduced to a symmetrical medial cartilage shaft spanning the whole length of the snout (Fig. 1). This long structure results from the midline apposition of two lateral flat cartilages as seen in frontal section of E18.5 embryos (Fig. 1H). The nasal cartilage of Dlx5/6^−/− mouse embryos is devoid of any cavity (Fig. 1F,H), and both sensory and mucosal epithelia are absent. The olfactory nerve is missing while a small olfactory bulb is still present (Fig. 1E,F). The midline position of the cartilaginous rod (Fig. 1E-H) and the absence of cavity renders this structure similar to the nasal septum, which also derives from the midline juxtaposition of two hemisepta (Wealthall and Herring, 2006). Thus, in Dlx5/6^−/− mouse embryos, the dorsal nasal capsule (ectethmoid) is absent, while the ventral, mesethmoid-derived, nasal septum is still present.

To examine the role of Dlx5 and Dlx6 in chick nasal capsule formation, we downregulated their expression by simultaneous electroporation of siRNAs against Dlx5 and Dlx6 (see Fig. S2 in the supplementary material) in HH6/7 embryos. At this stage, Dlx5 and Dlx6 are expressed primarily in the anterior neural ectoderm spanning the presumptive olfactory territory (NF-ZA), as revealed by immunostaining (Fig. 2A) and in situ hybridization [data not shown (Bailey et al., 2006)], highlighting the likelihood that the knockdown of Dlx5/6 activity is stage and tissue specific. Furthermore, CNCCs had yet to undergo delamination and migration, and therefore they were not likely to be targeted by siRNA transfection. Nine days later, at HH35, we analyzed skeletal preparations of the resulting surviving specimens (23/42). Inhibition of Dlx5 and Dlx6 results in a hypomorphic upper beak (Fig. 2B) characterized by a reduced or absent ectethmoid (Fig. 2C,D); other dorsal nasal capsule structures, including the olfactory epithelium, were not observed. By contrast, the mesethmoid and Meckel's cartilages are present in their normal position. To further confirm the importance of Dlx5 and Dlx6 expression in NF-ZA, we performed homotopic (23) and heterotopic (26) grafts of NF-ZA dissected from siRNA-transfected donors. Unfortunately, the failure of these grafted embryos to survive up to stages where it was possible to analyze skeletal morphology has precluded any meaningful interpretation of the outcome.

To gain insight into the formation of the nasal capsule, we then undertook a series of experimental manipulations of the Dlx5/6-rich NF-ZA in the avian embryo. We first determined the contribution of NF-ZA to the development of the olfactory system by performing unilateral homotopic substitution of five-somite chick NF-ZA with stage-matched quail NF-ZA (Fig. 3A). Seven days after operation, in surviving embryos (8/15), a cartilaginous nasal capsule develops normally and quail-derived cells can be identified by QCPN staining in territories previously reported to derive from NF-ZA (Couly and Le Douarin, 1985). Donor cells are present within the olfactory bulb and the olfactory epithelia, but not in the surrounding ectethmoid cartilages (Fig. 3B,C). Bilateral ablation of NF-ZA (Fig. 3D) results in the strong reduction of ectethmoid cartilages (Fig. 3E,F), while the median mesethmoid is still present in surviving embryos (17/87). Heterotopic grafts of territories neighbouring NF-ZA (neural fold zone B; presumptive adenohypophysis; presumptive forebrain; non-neural ectoderm) cannot compensate for its absence (see Fig. S1 in the supplementary material and data not shown).

We then unilaterally grafted supplementary NF-ZAs at the level of the presumptive first pharyngeal arch (PA1) and examined chondroskeletal and dermatoskelatal structures at HH35 and HH38 (Fig. 4). In whole-mount skeletal preparations of all surviving embryos (35/78), we consistently observed large ectopic cartilaginous or bony structures reminiscent of ectethmoidal conchae, apophysa crista Galli and associated dermatocranial elements (Fig. 4B), in contrast to the unoperated contralateral side (Fig. 4B′). Surrounding structures, including Meckel's cartilage, develop normally.

To determine the cellular origin of the supernumerary structures, we first substituted the chick host mesencephalic CNCCs with stage-matched CNCCs from quail donor embryos; then we ectopically grafted chick NF-ZA onto the same host (Fig. 4C). The resulting surviving chimerae (2/8) were analyzed at HH35 for quail cell lineage. QCPN-positive cells are present in all cartilages, including the ectopic ectethmoid (Fig. 4D,E). This observation suggests that the ectopic NF-ZA recruits CNCCs from the mesencephalic region to induce the ectopic ectethmoid. Histological analysis of serial sections of HH28 grafted embryos showed that the ectopic ectethmoid is lined by an internal olfactory epithelium derived from the chick grafted NF-ZA (Fig. 4E). NF-ZA grafts failed to survive and did not induce any ectopic structure when implanted in posterior Hox-positive territories (data not shown).

We then analyzed the effect of simultaneous ablation of NF-ZAs and of EZ-I (Fig. 5A). In surviving embryos (2/9), both components of the nasal capsule are absent (Fig. 5B-C′). Other CNCCs chondroskeletal derivatives of PA1, including Meckel's cartilage, develop normally. Next, we simultaneously grafted NF-ZA and EZ-I in the presumptive PA1 (Fig. 5D); care was taken to maintain the relative orientation of all elements. Analysis of the chondroskeleton of HH35 surviving embryos (5/10) shows a supernumerary nasal capsule consisting of a cartilaginous bar corresponding to the mesethmoid associated to an ectethmoid-like cartilage (Fig. 5E,E′). This supernumerary structure is remarkably similar to the neighbouring nasal capsule, although slightly reduced in size. First and second arch derivatives develop normally.

The development of sensory systems of the craniate head results from dynamic interactions between cephalic neural crest cells and sensory placodes. Here, we show that the pre-placodal olfactory territory of the neural plate (NF-ZA) is necessary and sufficient to induce CNCCs to form the dorsal nasal capsule (ectethmoid). As NF-ZA further gives rise to the olfactory placodes and their derivatives, this anterior ectodermal region is therefore the organizer of the olfactory system.

NF-ZA is characterized, in both the avian and the mammalian embryo, by high levels of Dlx5 and Dlx6 expression (Acampora et al., 1999; Bhattacharyya and Bronner-Fraser, 2008; McLarren et al., 2003; Pera and Kessel, 1999; Pera et al., 1999; Yang et al., 1998). We show that the simultaneous invalidation of Dlx5 and Dlx6 in the mouse and their downregulation in the chick embryo specifically prevents the formation of the ectethmoid.

Our findings imply that both specification and patterning instructions for nasal capsule morphogenesis are already engraved in the pre-placodal presumptive olfactory territory (NF-ZA) at HH7. Our results extend to pre-placodal stages of development, with recent findings obtained at HH15-20 (Szabo-Rogers et al., 2008; Szabo-Rogers et al., 2009) showing that, in chick embryos, signalling from the nasal pits is involved in nasal capsule morphogenesis. This instructive potential of NF-ZA is present well before the onset of CNCCs migration.

In contrast to later manipulations of placodal territory (Szabo-Rogers et al., 2009), which did not induce nasal capsule-specific cartilage outside presumptive frontonasal territories, we show here that mesencephalic CNCCs migrating to PA1 are competent to respond to NF-ZA instructions. Furthermore, after induction, CNCCs give rise to a supernumerary ectethmoid, while the development of neighbouring PA1 CNCCs derivatives is not affected. Collectively, these findings suggest that, at HH15, CNCC-derived facial mesenchyme has become refractory to further ectodermal induction. Our observations thus fill the gap between the early activation of Dlx genes within pre-placodal progenitors and later roles exerted by their Dlx-positive derivatives in organizing the morphogenesis of the olfactory system.

The dorsal induction of the ectethmoid is concomitant with the ventral induction of the mesethmoid by endodermal EZ-I (Benouaiche et al., 2008). We suggest, therefore, that the interactions of CNCCs with the endoderm ventrally, and the pre-placodal ectoderm dorsally, give rise to two independent structures, the mesethmoid and the ectethmoid, respectively, which then assemble to form the nasal capsule (Fig. 6). Moreover, all information needed for final assembly lie within the three components (CNCCs, EZ-I and NF-ZA) because a complete nasal capsule can form ectopically upon simultaneous grafting of EZ-I and NF-ZA. We further show that the assembly paradigm of the nasal capsule stands true across species, as lack of Dlx5/6 in the prospective olfactory domain specifically prevents ectethmoid formation in the nasal capsule of both mouse and avian embryos.

Dlx5 and Dlx6 are expressed in olfactory precursors and derivatives through adulthood (Levi et al., 2003; Robledo et al., 2002). Mouse anterior ectodermal cells fated to express Dlx5 and Dlx6 persist, despite simultaneous invalidation of both genes (Depew et al., 2002). On the contrary, in both birds and mammals, frontonasal CNCCs and their nasal capsule chondrogenic derivatives never express Dlx5 and Dlx6 (Acampora et al., 1999; Bhattacharyya et al., 2004; Levi et al., 2003; Robledo et al., 2002). We find that both in mouse and chick nasal skeletal structures are present after Dlx5 and Dlx6 downregulation, suggesting that frontonasal CNCCs persist even in the absence of these gene products. Collectively, these observations support the hypothesis that the lack of ectethmoid in Dlx5/6^−/− mutants results from disrupted interactions between pre-placodal precursors (NF-ZA) and CNCCs.

The question of which molecular signals mediate the interaction between NF-ZA and CNCCs remains still unanswered. Several signalling activities are present in the developing nasal-olfactory system and might account for these early instructive cues. Among those, Fgf ligands are expressed in mouse (Baker and Bronner-Fraser, 2001; Kawauchi et al., 2005; Minoux and Rijli, 2010) and chick (Abzhanov et al., 2007; Abzhanov and Tabin, 2004; Creuzet et al., 2004) neurulas within the presumptive olfactory domain. Alterations of olfactory structures have been reported in both species upon genetic or molecular invalidation of Fgf signalling (Bailey et al., 2006; De Moerlooze et al., 2000; Sjodal et al., 2007; Trokovic et al., 2003; Wang et al., 1999). In vitro studies have shown that Fgf is involved in determining the early lineage segregation of uncommitted pre-placodal precursors towards an olfactory or an optic fate (Bailey et al., 2006; Sjodal et al., 2007). Along these lines, Bmp activity favours lens placodal fate at the expanse of olfactory fate (Sjodal et al., 2007) and has a later role in promoting facial mesenchymal cells proliferation (Wu et al., 2004).

Collectively, these data and our present work support a two-step model for early morphogenesis of the olfactory system starting from a large Dlx3/5/6-positive preplacodal territory of the anterior neurula. First, as demonstrated in the chick (Bailey et al., 2006), interaction of preplacodal cells with CNCCs, mediated by Fgf signalling (Creuzet et al., 2006), restricts Dlx5 expression to olfactory precursors which includes NF-ZA. Then, as we show here, the NF-ZA induces CNCCs to form an ectethmoid, which subsequently remains intimately associated with the NF-ZA-derived olfactory epithelia. This second step requires the expression of Dlx5 and Dlx6. The nasal capsule is properly specified, but severely deformed in both Dlx5^−/− and Dlx5^−/−; Dlx6^+/− mice (see Fig. S3 in the supplementary material) and does not develop in the absence of both genes (Depew et al., 2005). Collectively, these data suggest that, as for the first pharyngeal arch (Vieux-Rochas et al., 2007; Vieux-Rochas et al., 2010), the allelic dosage of Dlx5 and Dlx6 controls the correct morphogenesis of the olfactory system.

In parallel with the dorsal induction of the ectethmoid, ventral Shh signalling from the anterior-most region of the endoderm instructs postmigratory CNCCs to form the mesethmoid cartilage (Benouaiche et al., 2008; Gitton et al., 2010).

The interaction between these two systems results into the juxtaposition of two independent modules that, together, give rise to a complete nose (Fig. 6). Consistently, we show that a complete and properly-patterned nose is induced by juxtaposed ectopic grafts of NF-ZA and EZ-I, suggesting that the ensemble of these inducing centres constitutes a robust and autonomous organizer.

Recently, it has been suggested that Fgf and Wnt signalling interact (Liu et al., 2010) to define the precise patterning of craniofacial cartilages. This observation provides a paradigm to explain craniofacial variations within and across species (Brugmann et al., 2007; Brugmann et al., 2006). Although Fgf8 expression is retained by cultured explants and grafts of NF-ZA (data not shown), it seems unlikely that Fgf8 is the only signal involved in nose dorsoventral patterning as its frontonasal expression persists in Dlx5/6^−/− mice (Depew et al., 2002) (data not shown).

The fact that the nose is formed through two independent signalling pathways might reflect the fact that it supports smell and respiration, two independent physiological functions that were acquired separately during evolution (O'Malley, 1923). Although variations in the chronology, topography and/or intensity of the interactions may have underlined craniofacial evolution (Minoux and Rijli, 2010), their alteration might as well account for the origin of human nasal defects. More specifically, this modular conception of the nose opens the possibility to adopt a new, much needed, classification method for human frontonasal malformations (Couly, 1981).

This work has been supported in part by grants in aid by Association Française contre les Myopathies (AFM), the Agence Nationale pour la Recherche (ANR), projects Gendactyl and DrOS, and EU community Crescendo LSHM-CT-2005-018652 and the EuroRETT project. We gratefully acknowledge the insightful advice and support of Prof. Nicole Le Douarin. We thank Prof. Barbara Demeneix for critical reading of the manuscript, and Mrs Anastasia Fontaine and Mrs Sabrina Baudry for excellent technical help. E.H. is recipient of a doctoral fellowship from ANR Gendactyl and from the Fondation de la Recherche Medicale (FRM); K.B. is recipient of a doctoral fellowship from the French Government (MRES).