Probing the origin of matching functional jaws: roles of Dlx5/6 in cranial neural crest cells

Gnathostome jaws derive from the first pharyngeal arch (PA1), a complex structure constituted by Neural Crest Cells (NCCs), mesodermal, ectodermal and endodermal cells. Here, to determine the regionalized morphogenetic impact of Dlx5/6 expression, we specifically target their inactivation or overexpression to NCCs. NCC-specific Dlx5/6 inactivation (NCC∆Dlx5/6) generates severely hypomorphic lower jaws that present typical maxillary traits. Therefore, differently from Dlx5/6 null-embryos, the upper and the lower jaws of NCC∆Dlx5/6 mice present a different size. Reciprocally, forced Dlx5 expression in maxillary NCCs provokes the appearance of distinct mandibular characters in the upper jaw. We conclude that: (1) Dlx5/6 activation in NCCs invariably determines lower jaw identity; (2) the morphogenetic processes that generate functional matching jaws depend on the harmonization of Dlx5/6 expression in NCCs and in distinct ectodermal territories. The co-evolution of synergistic opposing jaws requires the coordination of distinct regulatory pathways involving the same transcription factors in distant embryonic territories.


INTRODUCTION
The gnathostome skull is characterized by the presence of articulated, muscularized asymmetric jaws, capable to support predatory behaviours and feeding functions such as mastication and swallowing (Compagnucci, et al., 2013;Depew, et al., 2005). During development, upper and lower jaws derive from the maxillary and mandibular processes of the first pharyngeal arch (PA1) that give rise to the palatoquadrate dorsally and the Meckelian cartilage (MC) ventrally. Hox-free Neural Crest Cells ('NCCs') emigrating from the mesencephalic and anterior rhombencephalic neural folds (Ealba, et al., 2015;Fish, et al., 2014;Minoux and Rijli, 2010;Depew and Olsson, 2008;Noden and Trainor, 2005;Creuzet, et al., 2002;Kontges and Lumsden, 1996;Trainor and Tam, 1995;Couly, et al., 1993) colonize the PA1 and give rise to most facial cartilages, bones and tendons. Before migration, NCCs lack the topographic information needed to unfold jaw morphogenetic processes . In the course of migration and after reaching their final destination in the craniofacial buds, NCCs contact mesodermal, ectodermal and endodermal cells.
Morphogenetic cues exchanged between these different cellular populations result in the formation of functional and morphologically different upper and lower jaws capable to operate in synergy to assure feeding (Barske, et al., 2018;Sato, et al., 2008b;Tucker and Lumsden, 2004).
Surgical deletion and grafting of different regions of the embryonic anterior foregut endoderm and ectoderm have demonstrated that these epithelia convey to NCCs the topographic cues needed for PAs patterning and facial morphogenesis (Noden and Trainor, 2005;Ruhin, et al., 2003;Couly, et al., 2002). The molecular nature of these instructive signals is not completely elucidated, but experimental evidence suggests the involvement of bone morphogenetic proteins (BMPs), FGFs, endothelin-1 (Edn1), Notch/Delta and Sonic hedgehog (Depew and Simpson, 2006) signalling molecules.

4
In the mouse, loss of Edn1 and/or endothelin receptor type-A (Ednra) (Clouthier, et al., 1998;Kurihara, et al., 1994) result in the transformation of the lower jaw into a structure presenting major morphological hallmarks of an upper jaw such as absence of MC, zygomatic arch-like structures and vibrissae. These observations indicate that Edn1 signalling plays a major role in the specification of mandibular identity (Ozeki, et al., 2004;Ruest, et al., 2004).
Remarkably, however, the upper jaw-like structures deriving from the mouse mandibular arch after Edn1-pathway inactivation are hypomorphic and clearly different from the normal upper jaw (Ruest, et al., 2004;Clouthier, et al., 1998). A further demonstration of the importance of Edn1-signalling for the establishment of jaw patterning comes from the constitutive activation of Ednra in upper jaw NCCs where its ligand, Edn1, is not normally present (Ozeki, et al., 2004). We reported that ectopic Ednra activation induces the transformation of maxillary into mandibular structures with duplicated MCs and dermatocranial jaws constituted by four, opposing dentary-like bones. In the same study we obtained a similar transformation forcing the expression of Hand2, a downstream target of the Edn1 pathway, in the Ednra-positive domain (Sato, et al., 2008a). It appears that Edn1 drives lower jaw development through a calmodulin-CamKII-histone deacetylase cascade that de-represses Mef2c, which then transactivates Dlx5/6 expression (Hu, et al., 2015;Verzi, et al., 2007).
Dlx homeobox genes are vertebrate homologues of Drosophila distal-less which play a fundamental role in specifying dorsoventral PA1 patterning (Depew, et al., 2005;Merlo, et al., 2000). The six gnathostome Dlx genes are organized in bigenic tandems and are all expressed in NCCs of craniofacial primordia in spatially and temporally restricted patterns.
Dlx1 and Dlx2 are present in both maxillary and mandibular NCCs while Dlx5 and Dlx6 are expressed only in mandibular NCCs. The targeted simultaneous inactivation of Dlx5 and Dlx6 (Beverdam, et al., 2002;Depew, et al., 2002) in all cells of the developing embryo, including NCCs and epithelial cells, leads to the transformation of the lower jaw into an upper jaw-like structure which, at variance with that obtained after Edn1 or Ednra inactivation, presents a similar size and a remarkable symmetry compared to the existing upper jaw.
The transformations in jaw identity induced by the inactivation of Edn1 signalling are accompanied by the down regulation of Dlx5 and Dlx6 in NCCs (Fukuhara, et al., 2004;Ozeki, et al., 2004;Ruest, et al., 2004;Clouthier, et al., 1998); however, a territory of Edn-1 independent Dlx5/6 expression is consistently maintained in the proximal PA1 Ozeki, et al., 2004). Edn1 is expressed in the epithelium and mesodermal core of the mandibular part of PA1, whereas Ednra is almost exclusively expressed by NCCs suggesting that altered Edn1 signalling affects Dlx5 and Dlx6 expression only in NCCs and not in other cell types. Given the difference between the lower jaw morphology obtained in Edn1/Ednra mutants and Dlx5/6 mutants, these findings suggest that the Edn1-dependent activation of Dlx5/6 in NCCs is necessary to specify mandibular identity, but is insufficient to generate a normal lower jaw morphology (Sato, et al., 2008b;Fukuhara, et al., 2004;Ozeki, et al., 2004;Ruest, et al., 2004;Clouthier, et al., 1998). Interestingly it has been observed that, after inactivation of Dlx5 and Dlx6, maxillary components are also affected despite the fact that these genes are not expressed by maxillary NCCs (Beverdam, et al., 2002;Depew, et al., 2002). This observation could be accounted for by the presence of signalling centres at the extremities of both the mandibular and maxillary arches; the so-called "Hinge and Caps" model of jaw organization (Compagnucci, et al., 2013;Fish, et al., 2011;Depew and Compagnucci, 2008). This model predicts the presence of two opposing morphogen gradients, one emanating from the region of the upper/lower jaw articulation (hinge) and one from the distal extremities of PA1 (caps); the origin and nature of these signals remain still elusive. By lineage analysis, we have shown that the maxillary arch epithelium harbours a cellular contingent derived from frontonasal Dlx5-expressing progenitors suggesting that transient 6 Dlx5/6 expression could program these epithelial cells to provide the cues needed for maxillary arch morphogenesis (Gitton, et al., 2014).
Here, we address the morphogenetic role of Dlx5/6 expression in NCCs. To this end we first invalidate Dlx5 and Dlx6 specifically in NCCs and then we force ectopic expression of Dlx5 in maxillary NCCs that do not normally express this gene. Our findings confirm that Dlx5/6 expression in NCCs is necessary to specify mandibular identity and has an organizer role for craniofacial myogenesis. However, NCC-limited Dlx5/6 expression is insufficient to determine normal jaw morphogenesis; coordination of Dlx5/6 expression in NCCs and in other cellular components is essential for the development and evolution of a functional mouth.

Generation of mouse models with deregulation of Dlx5/6 expression in NCCs
To induce ectopic expression of Dlx5 in the NCC lineage, we crossed ROSA CAG-flox-Dlx5/+ mice, which express Dlx5 in a Cre-dependent manner, with Wnt1 Cre mice (Chai, et al., 2000) to obtain the NCC Dlx5 mouse line in which Dlx5 should be expressed in premigratory neural crest cells (Fig. 1A). The expression of Dlx5 in the pharyngeal arches was analyzed by quantitative RT-PCR (Fig. 1B) and whole-mount in situ hybridization (Fig. 1C, C'). RT-PCR was performed on dissected maxillary, mandibular and second pharyngeal (PA2) arches from E10.5 control and NCC Dlx5 embryos. In all NCC Dlx5 samples, Dlx5 expression was higher than that found in the corresponding controls (Fig. 1B). By in situ hybridization, ectopic Dlx5 expression was detected in the presumptive maxillary and nasal regions, well beyond the normal mandibular territory of expression ( Fig. 1C, C'). This pattern is consistent with the distribution of lacZ expression in Wnt-1::Cre/R26R mice  suggesting that indeed Dlx5 expression has been activated and maintained ectopically in NCCs.
To inactivate Dlx5/6 in NCCs, Dlx5/6 flox/flox mice, in which the homeodomainencoding regions of both Dlx5 and Dlx6 are flanked by lox sequences (Bellessort, et al., 2016), were crossed with Wnt1 Cre-ERT2/+ mice in which tamoxifen exposure induces Crerecombinase activity in cells of the developing neural tube, in migrating NCCs and in dopaminergic neurons, but not in other cell types. To cover most of the period of neural crest delamination and migration (Danielian, et al., 1998) Dlx5/6 flox/flox ::Wnt1 Cre-ERT2/+ timed pregnant dams received two intraperitoneal injections of tamoxifen at embryonic developmental days E6 and E7, generating NCC ∆Dlx5/6 embryos. The need to inactivate both Dlx5 and Dlx6 in NCCs derives from the fact that these two closely related genes are 8 redundant in defining lower jaw identity: both need to be inactivated to transform the identity of the lower jaw into an upper jaw (Beverdam, et al., 2002) , (Depew, et al., 2002). presented a marked mandibular retrognatia, with vibrissae present on both upper and lower jaws. In the upper jaw of NCC ∆Dlx5/6 foetuses, vibrissae were arranged in 5 rows as in control littermates, while ectopic vibrissae in the lower jaw were very close to each other and did not present a recognizable pattern; in these animals some vibrissae developed close to the midline. The palate and the eyelids of E18.5 NCC ∆Dlx5/6 foetuses did not present any obvious malformation.

Craniofacial defects observed after deregulation of
The craniofacial phenotypes resulting from either up-or down-regulation of Dlx5/6 in NCCs were analysed by 3D reconstruction after serial sectioning and imaging of the heads at E18.5 and compared to normal littermates or to heads of foetuses of the same age in which Dlx5 and Dlx6 had been constitutively inactivated (Dlx5/6 -/foetuses) (Gitton, et al., 2014;Beverdam, et al., 2002) (Figs. 3, 4 and Sup. Figs. 1-4 for 3D pdf files). As already described (Gitton, et al., 2014;Beverdam, et al., 2002;Depew, et al., 2002), after constitutive Dlx5/6 inactivation, both lower and upper jaws are transformed resulting in a symmetric mouth. In Dlx5/6 -/mice, the lower jaw acquires morphological and molecular hallmarks of an upper jaw (Beverdam, et al., 2002) , (Depew, et al., 2002) , (Jeong, et al., 2008) while the premaxillary bone is not recognizable (Fig. 3 B In NCC Dlx5 embryos the incus seemed to have partially acquired a malleus-like identity as: 1) it short process was elongated, 2) a small ectopic bone, which could be interpreted as a duplicated gonial bone, appeared adjacent to the incus, and 3) the malleus-incus joint, which is normally of ball-and-socket type, was symmetric ( Fig. 6D, D'; Sup. Fig. 5). The infraorbital foramen was not present in NCC Dlx5 mice also suggesting an upper jaw transformation (Sup. Fig. 4).

Incisor phenotypes
Lower incisor germs of normal E18.5 foetuses are elongated structures that develop within the dentary bone and converge medially towards the distal tip ( Fig. 4A, A' and Sup.  Figure 1). Remarkably, in E18.5 Dlx5/6 -/foetuses, the four incisor germs were not supported by any bone structure (dentary or premaxillary) and appeared as short rods of comparable length oriented along the proximo-distal axis not converging toward the midline (Figs. 3B-B"; 4B, B' and Sup. Fig. 2). The four incisors germs of Dlx5/6 -/foetuses were, therefore, very similar to each other. The lack of supporting bones for Dlx5/6 -/incisors suggests that bones and incisors teeth can develop independently. In NCC ∆Dlx5/6 E18.5 foetuses (Figs. 3C', 4C, C' and Sup. Fig. 3), lower incisors were constituted by short, straight rods oriented towards the midline and protruding for about two thirds of their length from the transformed dentary bone. Upper incisors germs of NCC ∆Dlx5/6 mutants were short and pointed structures entirely supported by premaxillary bones; they were oriented ventrally along an oblique axis towards the midline.

Dlx5/6 expression in neural crest cells is required for proper jaw muscularization
We have previously shown that Dlx5/6 -/mice display agenesis of masticatory muscles due to the lack of instructive cues from NCCs onto cephalic myogenic mesoderm precursors . In E18.5 NCC ∆Dlx5/6 foetuses, craniofacial muscularization was affected with a phenotype strongly resembling that observed in Dlx5/6 -/mice .
The masticatory masseter muscles failed to differentiate normally and were replaced by loose mesenchymal tissue (Fig.5B'). The other masticatory muscles, temporal and pterygoids, were present dorsally but showed defective attachments on the transformed lower jaws (Fig. 5C').
The tongue and associated musculature were severely affected: the suprahyoid muscles (mylohyoid, digastric and geniohyoid) and the extrinsic genioglossus tongue muscles did not differentiate and were replaced by loose tissue containing few disorganized myofibers; the intrinsic tongue musculature was reduced, disorganized and remained as a vestigial mediallylocated structure disconnected from lower jaw bones ( Fig. 5B', C').
In NCC Dlx5 foetuses all craniofacial muscles could be identified, but their attachment points and their shape were changed to adapt to the transformed skeletal elements (Fig. 5 B'', C''). These observations reinforce the notion that cephalic myogenic precursors require Dlx5/6-dependent morphogenetic instructions from NCCs for proper jaw muscle patterning and differentiation (Sugii, et al., 2017).

Dlx5 overexpression in NCCs switches maxillary to mandibular transcriptomic signature
To identify regulatory pathways involved in determining jaw identity, we performed transcriptome analysis on PA tissues of E10.5 control and NCC Dlx5 embryos. When the NCC Dlx5 maxillary arch samples were compared to those from control maxillary arches, 12 and 21 genes were identified as increased or decreased by more than 2-fold, respectively (Table 1). Scatter plot of the signal intensity showed that the upregulated genes corresponded to mandibular marker previously reported to be downstream of Dlx5/6 ( Fig. 8) (Jeong, et al., 2008). By contrast, maxillary marker genes reported to be upregulated in the transformed 13 mandibular arch of Dlx5/6 -/embryos showed only small deviation from the diagonal (Fig. 8).
To confirm this difference and further characterize the Dlx5-downstream genes, we then performed microarray analysis on PA tissues of E10.5 Dlx5/6 -/embryos and compared the results with genes deregulated in the NCC Dlx5 maxillary arch. In the Dlx5/6 -/mandibular arch we identified 18 genes whose expression was increased and 45 genes whose expression was decreased by more than 2-folds (Table 2). Remarkably, 10 of 12 genes up-regulated in the NCC Dlx5 maxillary arch were also down-regulated in the Dlx5/6 -/mandibular arch suggesting a common set of downstream targets involved in lower jaw specification (Table 2). By contrast, only 1 gene, Has2, a putative hyaluronan synthase, was shared between the upregulated genes in Dlx5/6 -/mandibular arch and the down-regulated genes in NCC Dlx5 maxillary arch (Table 2).

DISCUSSION
In gnathostomes, feeding depends on muscularized, articulated jaws capable to support prehension, mastication and swallowing. In modern tetrapods, upper and lower jaws are, in general, distinct anatomical structures. Since many decades, however, it has been remarked that in basal reptiles and amphibians the upper and the lower jaws are essentially mirror images of each other (Romer, 1940). The lower jaw of contemporary species is constituted by two mobile dentary bones that form around an early rod-like Meckelian cartilage and converge towards the centre where they fuse at the mandibular symphysis. The stationary upper jaw is mostly constituted by dermal bones that fuse along the midline and extend medio-laterally to form the palate and the maxillary part of the face. During embryonic development both upper and lower jaws derive from the first pharyngeal arch (PA1) that is colonized by Hox-negative NCCs (Kontges and Lumsden, 1996;Trainor and Tam, 1995;Couly, et al., 1993) which give rise to most cartilages, bones and tendons of the jaws, while the associated musculature originates from the myogenic mesoderm. Endothelin-1 (Edn1) is a key signal at the origin of asymmetric jaw development. Binding of Edn1 to its receptor A (Ednra) in NCCs activates the expression of lower-jaw specific genes including Dlx5/6 and Hand2 (Sato, et al., 2008b). Remarkably targeted inactivation of either Edn1, Ednra, Dlx5/6 or Hand2 result in the transformation of the lower jaw into new structures that present key morphological hallmarks of an upper jaw such as, for example, the presence of vibrissae and the absence of Meckelian cartilage (Ruest, et al., 2004;Ruest, et al., 2003;Clouthier, et al., 1998). It must be pointed out, however, that each of these mutations generates different lower jaw morphologies. Targeted constitutive inactivation of Dlx5/6 results in the simultaneous transformation of the upper and of the lower jaws giving rise to a symmetric mouth in which each of the four jaws (upper/lower, right/left) presents a morphology very similar to each of the others transposed along reflection planes (Beverdam, et al., 2002;Depew, et al., 2002).
We have shown that the Dlx5/6-dependent patterning of the upper jaw derives from their transitory expression in an ectodermal, distal, signalling centre located in the lamboidjunction area and not from their expression in mandibular NCCs (Gitton, et al., 2014).
Constitutive disruption of either 1) Edn1, 2) its receptor Ednra in NCCs or 3) the Edn1 downstream target Hand2 results in jaw phenotypes very different from those observed in Dlx5/6 -/embryos: the upper jaw does not present any obvious malformation while the lower jaw becomes small and hypomorphic and acquires, at the same time, upper-jaw characters such as the presence of vibrissae (Ozeki, et al., 2004;Clouthier, et al., 1998) , (Ruest, et al., 2004;Yanagisawa, et al., 2003;Charite, et al., 2001). In Edn1, Ednra or Hand2 mutants the lower jaw is transformed, but a dorso-ventral symmetry of the mouth is not acquired.
Conversely, activation of the Edn-signalling pathway in the PA1 maxillary NCCs contingent, where it is normally silent, results in the activation of Dlx5/6 and Hand2 leading to the transformation of upper jaws into dentary-like structures articulated with unaffected lower jaws presenting their own Meckelian cartilages (Sato, et al., 2008b). Importantly, it has been shown that Dlx5/6-positive NCCs do not only contribute to the morphogenesis of bones and cartilages, but exert also an instructive role for the patterning and differentiation of the myogenic mesoderm; the constitutive inactivation of Dlx5/6 results in absence of masticatory muscles while in Ednra mutants, the Edn1-independent reactivation of Dlx5/6 in PA1 is sufficient to rescue muscular differentiation . The sum of these results suggests that Dlx5/6 expression in different jaw precursors including, ad minima, NCCs and distal ectodermal cells, has a critical role in sculpturing vertebrate facial structures. Here, by genetically down-or up-modulating the expression of Dlx5/6 specifically in NCCs, we addressed the question of which morphological messages are conveyed by NCC-restricted Dlx5/6 expression and which other cues depend from their expression in other cell types.
We show that, if Dlx5/6 are expressed in either mandibular or maxillary NCCs the morphological and molecular signature of a lower jaw can be recognized while, in their absence, the jaws assume invariably a maxillary identity. In particular the main morphological elements which we have considered as typical traits of mandibular identity are: 1) the presence of a Meckelian cartilaginous rod, 2) the absence of midline fusion between the right and left jaws that are joined only at the distal symphysis, 3) the absence of an infraorbital foramen, 4) the presence of an elongated incisor, 5) the absence of a premaxillary bone, 6) the absence of vibrissae, 7) the presence of tongue and associated musculature and 8) the insertion of masticatory muscles. Maxillary identity on the contrary has been defined by the Remarkably, all mandibular traits are invariably present when Dlx5/6 are expressed in NCCs either in the upper or in the lower jaw. Conversely, maxillary traits appear in the absence of Dlx5/6 expression in NCCs associated to a defective muscular system.
Our first conclusion is that NCC-limited expression of Dlx5/6 is necessary and sufficient to specify a mandibular identity and to maintain the myogenic program for the formation of muscularized jaws.
The second, flagrant, result of this study is that Dlx5/6 must be expressed in cellular contingents different from NCCs to permit the development of matching, functional jaws.
Indeed, the inactivation of Dlx5/6 only in NCCs transforms the lower jaw into an hypomorphic upper jaw that is severely out of register from the existing maxilla. On the contrary, inactivation of Dlx5/6 in all cell types, including NCCs results in the transformation of both upper and lower jaws into similar matching structures that, if muscularized, could prefigure a functional mouth. It must be pointed out that transient expression of the transcription factors Dlx5/6 might be sufficient to change the developmental trajectory a cellular contingent and might confer the instructive capacity needed to modulate morphogenesis. In line with this concept, we have previously shown that the transient expression of Dlx5/6 in epithelial precursors deriving from the lamboid junction (the intersection between presumptive maxillary and frontonasal bud-derived structures) is needed for correct morphogenesis of the upper jaw (Gitton, et al., 2014). The main findings of this paper are summarized in Fig. 9.
It has been shown that FGF8 signals deriving from the mandibular PA1 epithelium are essential for jaw morphogenesis. Indeed, inactivation of mouse Fgf8 in PA1 epithelia results in the loss of most PA1 skeletal derivatives (Trumpp, et al., 1999). Interestingly, mandibular, but not maxillary ectomesenchyme is competent to respond to Fgf8 maintaining Dlx5 expression (Ferguson, et al., 2000) , (Park, et al., 2004). A possible interpretation of our data would be that Dlx5 expression in the lower jaw epithelium could be essential to maintain molecular signals such as Fgf8 and, in so doing, could permit correct lower jaw morphogenesis. Interestingly, at the end of gastrulation Dlx5 expression first appears in lateral bands of the embryonic/extraembryonic junction at the late-streak stage which precedes neural plate formation (Yang, et al., 1998). Later, at E7.25 expression of Dlx5 is detected in the rostral and in the posterior ectoderm. At even later stages, Dlx5 is found in NCC progenitors and in epithelial precursors of the ventral cephalic ectoderm (VCE). During migration, NCCs down-regulate Dlx5 expression and reactivate it only after arrival in the mandibular PA1 in response to an Edn1 signal. At E8.5 the NCCs-derived mesenchyme of the early PA1 is clearly Dlx5-positive (Acampora, et al., 1999), however the epithelial component of PA1 does not express yet express Dlx5 (Yang, et al., 1998). The lower jaw epithelium seems to activate Dlx5 expression between E8.5, when the epithelium is still Dlx5-negative (Yang, et al., 1998), and E10.5, when the epithelial Dlx5 expression is comparable to that of the ectomesenchyme (Vincentz, et al., 2016), (see also Fig. 9). It seems plausible therefore that Dlx5 expression both in NCCs and in the PA1 mandibular epithelium is essential to permit correct lower jaw morphogenesis. VCE cells also down regulate Dlx5 expression, but their derivatives migrate then toward anterior parts of PA1 (Gitton, et al., 2014) and are essential for correct maxillary development. It seems therefore that the integration of signals deriving from three Dlx5-positive cellular progenitors (NCCs, PA1 epithelial cells and VCE cells) is needed for the generation of functional, matching jaws. Dlx5 expression in these different territories might well depend on different regulatory mechanisms. An example of this differential regulation is the observation that the I56i enhancer is active in PA1 NCCs, but not in epithelial structures such as the VCE, the PA1 epithelium or the otic vesicle (Park, et al., 2004).
Vital biological functions such as feeding or reproduction often depend on the interaction between matching structures located either in the same individual (e.g. jaws) or in separate members of the same species (e.g. sexual organs). The coevolution of synergistic complementary structures in separate individuals or in different territories of the same organism has, therefore, been instrumental to ensure species survival. We show that, for the specific case of mouth development, the same cluster of genes is: 1) determining lower jaw identity and 2) tuning the morphogenetic process of both the upper and the lower jaw generating a mouth endowed with opposing and matching mandibles capable to support feeding and/or predation. These two levels of morphogenetic regulation must be coordinated to generate a functional oral organ capable of mastication. These morphogenetic processes 19 occur in different cell types located in spatiotemporally separate embryonic territories suggesting that diverse Dlx5/6-depending pathways are involved. At least in the case of the jaws, the co-evolution of synergistic anatomical structures ostensibly involves the coordination of different regulations involving the same genes in distant embryonic territories.
These observations open very profound questions on how these different regulations might have evolved and be selected. It is remarkable however that both NCCs and the VCE derive from the same Dlx5-positive early progenitors at the very end of gastrulation suggesting interesting evolutionary scenarios that still need to be understood. Probing the origin of matching, muscularized, jaws could open a window to understand more generally how different harmonious, but separate, parts of the body coordinate their morphogenesis.

Mice
Procedures involving animals were conducted in accordance with the directives of the European Community (council directive 86/609), the French Agriculture Ministry (council directive 87-848, 19 October 1987, permissions 00782 to GL) and approved by the University of Tokyo Animal Care and Use Committee and by the "Cuvier" ethical committee (approval n° 68-028r1). Mice were housed in light, temperature (21°C) and humidity controlled conditions; food and water were available ad libitum. WT animals were from Charles River France. Double Dlx5 and Dlx6 (Dlx5/6) null mice (Beverdam, et al., 2002;Merlo, et al., 2000) and the double conditional mutant Dlx5/6 flox/flox (Bellessort, et al., 2016)  Homologous recombination was performed on the ROSA26 locus of B6129F1-derived ES cells. Targeted ES clones were injected into ICR blastocysts to generate chimeras. Chimeras were crossbred with ICR females. R26R CAG-flox-DLX5/+ mice were crossed with Wnt1-cre mice (Chai, et al., 2000) to induce NCC-specific expression of Dlx5 (NCC Dlx5 mice).
Embryos were obtained from pregnant dams, washed and dissected in ice-cold phosphate buffered saline (pH7.4). Embryos selected for fixation were immersed in 4% paraformaldehyde from 15mn to 2h, photographed and further processed (Bellessort, et al., 2016;Sato, et al., 2008b). Specimens selected for biochemical analyses were dissected in PBS and processed for RT-qPCR, Affymetrix, Western blot and Southern blot as described in Supplementary text 1 and (Bellessort, et al., 2016).

Histology and 3D reconstruction
Heads from E18.5 (mutant and wild type foetuses were fixed in Bouin's solution (Sigma, France), embedded in paraffin and complete sets of frontal or parasagittal serial sections (12µm) were prepared. All sections were stained by Mallory's trichrome  and photographed (Nikon Digital Site DS-FI1). Pictures were aligned, piled and registered using the Fiji plug-in of NIH ImageJ "Register Virtual Stack Slices" (http://fiji. sc/wiki/index.php/Register_Virtual_Stack_Slices). 3D segmentation was performed with Mimics (Materialise, Belgium: http://biomedical. materialise.com/mimics) and visualized using Adobe Acrobat 9 pro.

Skeletal preparation and staining
Alizarin red/ alcian blue staining was performed, as previously described (McLeod, 1980). Samples were fixed in 95% ethanol for a week, permuted to acetone for three days and 22 incubated with 0.015% alcian blue 8GS, 0.005% alizarin red S and 5% acetic acid in 70% ethanol for three days. After washing with distilled water, the samples were cleared in 1% KOH for several days and in 1% KOH glycerol series until the surrounding tissues turned transparent. The preparations were stored in glycerol.

In situ hybridization
Whole-mount in situ hybridization was performed, as described previously . Embryos were fixed one overnight in 4% paraformaldehyde in PBT. After dehydration and rehydration with methanol, the embryos were bleached for 1 hour in 7.5% H2O2 in PBT and then washed in PBT 3 times. The samples were treated with 5mg/ml proteinase K for 40 seconds at room temperature, treated with 2mg/ml glycine in PBT to stop the enzyme reaction, and post-fixed in 4% paraformaldehyde and 0.2% glutaraldehyde for 20 minutes on ice. After the pretreatment, the samples were pre-hybridized for more than 1 hour at 70℃ in hybridization mix (50% formamide, 5 x SSC (1 x SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 1% SDS), 50 mg/ml heparin and 50 mg/ml yeast tRNA. With digoxygenin-labeled RNA probe in hybridization mix, the samples were hybridized overnight at 70℃. The samples were then washed 3 times in hybridization mix at 70℃, then in 0.2 M NaCl, 10 mM Tris-HCl (pH7.5), 0.1% Tween-20 for 5 minutes ant treated with 100 mg/ml RNase for 30 minutes at 37℃. After a final wash in 50% formamide, 2 x SSC for 1 hour at 65°C, the samples were pre-blocked with sheep serum, incubated with alkaline phosphataseconjugated anti-digoxygenin antibody, and stained with nitro blue tetrazolium and 5-bromo-4chloro-3-indoyl phosphate. Probes were prepared by RT-PCR and used in published (Kitazawa, et al., 2015;Sato, et al., 2008b).

Quantitative real-time RT-PCR
The maxillary and mandibular processes were dissected from E10.5 control and NCC Dlx5 mice. Total RNA was extracted from five sets of PAs by ISOGEN-II (Nippon Gene).

Gene expression profiling
The maxillary process, the mandibular process and the PA2 were collected from  Genes up-or down-regulated in the NCCDlx5 maxillary arch.        to, tongue; vb, vibrissae; vb*, ectopic vibrissae. Abbreviations in light red represent muscles that did not differentiate and are replaced by loose mesenchymal tissue. Abbreviations: cp, coronoid process; dt, dentary bone; i, incus; jb, jugal bone; m, malleus;
Abbreviations: als, alisphenoidal bone; bo, basioccipital bone; bs, basisphenoidal bone; ect, ectopic structure; pcp, parachordal plate; px, premaxilla; pnc, paranasal cartilage; ptg, pterygoid. Points corresponding to known maxillary and mandibular arch markers are colored green and orange, respectively.   The elongation of the short process of the incus, the presence of a small ectopic bone, which could be interpreted as a duplicated gonial bone, adjacent to the incus, and the fact that the malleus-incus joint, which is normally of ball-and-socket type, is symmetric suggest a partial transformation of the incus in a malleus-like structure.