The vertebrate face develops from the coordinated growth of the facial prominences, which initially consist of undifferentiated neural crest–derived mesenchyme covered by an epithelial layer. At first, the facial prominences (the frontonasal mass, maxillary, lateral nasal, and mandibular) are buds that surround the primitive oral cavity. In the avian embryo between stage 20 and 28 (Hamburger Hamilton,1951), there is considerable proliferation of the mesenchyme driving the outgrowth of the facial prominences (McGonnell et al.,1998; Wu et al.,2004,2006; Szabo-Rogers et al.,2008). This outgrowth culminates with contact of the frontonasal mass and maxillary prominence and fusion of the primary palate. If fusion is disrupted, cleft lip will ensue. During this period, the development of species-specific form also occurs. The formation of wide versus narrow beaks is dependent on the neural crest–derived mesenchyme (Schneider and Helms,2003). The spatial arrangement of proliferating mesenchymal cells and BMP signalling in the frontonasal mass (Abzhanov et al.,2004; Wu et al.,2004,2006) fine tune the morphogenesis of the upper beak. BMPs are also required in lip fusion (Ashique et al.,2002a,b). FGFs promote proliferation and outgrowth of facial mesenchyme (Richman et al.,1997; Mina et al.,2002; Szabo-Rogers et al.,2008) and Sonic Hedgehog is sufficient to induce secondary outgrowth axes in the frontonasal mass (Hu and Helms,1999) and mandibular prominences (Brito et al.,2008). When compared to other signalling pathways, much less is known about the role of WNTs during facial development.
The Wnt family (wingless-type MMTV integration site) of secreted glycosylated factors consists of 19 members in mouse. However, in chicken there are only 18 confirmed WNT ligands. The chicken genome does not appear to have WNT2, 4B, 7C, or 10B at the time of writing. The mouse genome also lacks Wnt4b and 7c. WNT proteins have a range of functions during various developmental processes, such as proliferation, asymmetric division, patterning, and cell fate determination (Veeman et al.,2003; Logan and Nusse,2004; Gordon and Nusse,2006; Karner et al.,2006a; Geetha-Loganathan et al.,2008a,b). WNTs signal through the canonical, β-catenin dependent pathway (Logan and Nusse,2004), or at least two different noncanonical pathways (Veeman et al.,2003; Karner et al.,2006a). Central to the canonical pathway is the regulation of β-catenin. Upon binding to frizzled (Fzd) receptors, Wnt ligands trigger an accumulation of β-catenin in the cytoplasm that is then translocated to the nucleus. There β-catenin binds to transcription factors of the LEF/TCF family, activating downstream gene transcription. One of the “non-canonical” pathways, the Wnt/Ca2+ pathway, leads to release of intracellular calcium, possibly via G-proteins (Sheldahl et al.,1999; Kuhl et al.,2000). This pathway also involves activation of phospholipase C and protein kinase C (PKC). The second “non-canonical” pathway is called the planar cell polarity pathway. Fzd activates JNK and directs asymmetric cytoskeletal organization and coordinated polarization of cells within the plane of epithelial sheets (Karner et al.,2006a,b). This pathway also controls convergent extension movements during gastrulation and has recently also been implicated in the directional migration of cells within the developing palatal shelves (He et al.,2008). Complex crosstalk between these “canonical” and “non-canonical” pathways may regulate the cellular readout of Wnt signalling (Weidinger and Moon,2003).
Mouse models in which Wnt pathway genes were targeted demonstrate the need for canonical and non-canonical WNT signalling in facial morphogenesis. Knockouts for Wnt1, 3a, 4, 5a, 7a, 7b, 9a, 9b, and 11 have been reported. Of these, only three have been shown to have roles in facial development. Targeted deletion of Wnt3a caused death upon birth due to mandibular defects. However, the cause of failure to feed (presumably a cleft palate) was not characterized (Yoshida et al.,2006). Targeted deletion of Wnt9b (Carroll et al.,2005) causes cleft lip in some of the embryos (Juriloff et al.,2006). Full deletion of Wnt5a in mice causes a striking truncation of the upper and lower jaws. Other molecules in the Wnt pathway that have resulted in craniofacial phenotypes when deleted are Ctnnb1 (β-catenin), Dkk1 (Dikkopf), Tcf4/Lef1 (T-cell factor/lymphoid enhancer factor) compound mutants, and Sfrp1/Sfrp2 (Secreted frizzled-related protein) compound mutants. Conditional deletion of Ctnnb1 in neural crest cells eliminates all intramembranous bones in the skull (Brault et al.,2001). The compound null Tcf4−/−/Lef1−/− mice, which have very little response to canonical Wnts, also have a facial phenotype consisting of poorly developed maxillas and shorter nasal septums (Brugmann et al.,2007). The deletion of Dkk1, which is predicted to cause an increase in canonical Wnt signaling, causes a dramatic loss of the entire facial complex due to a very early role in head induction (Mukhopadhyay et al.,2001). Sfrp1 and 2 compound mutants, which should lead to increased canonical and non-canonical signalling, also have defects in the face (Satoh et al.,2006) but far less severe than the Dkk1 knockouts. Thus, regulating the level of WNT signalling is important for several aspects of facial morphogenesis.
Surprisingly, descriptions of WNTs, FZD receptors, or signalling mediators in the developing face are difficult to find in the literature. Furthermore, only partial data on the avian face is available from a handful of studies including expression of SFRP2 and FRZB1 (Ladher et al.,2000), WNT7A and WNT5A (Dealy et al.,1993). During in situ analyses, the head is often presented in sagittal views, if at all. Even the recent study using optical projection tomography on E11.5 mouse embryos is not at high enough resolution to determine which genes are expressed in the different facial prominences and whether expression is epithelial or mesenchymal or both (Summerhurst et al.,2008). Furthermore, expression is very dynamic during facial morphogenesis so several consecutive stages need to be studied. We chose to study the expression of the WNT pathway genes in avian embryos because ultimately it will be convenient to manipulate WNT signalling directly in the developing face. Expression was mapped in stage-15 to -28 embryos thereby covering the critical stages of facial development. These stages include: establishment of jaw identity (Lee et al.,2001), growth of facial prominences (Richman and Tickle,1989), and induction of intramembranous bone (Tyler and Hall,1977; Hall,1978,1980; Tyler,1978) and fusion of the facial prominences (Ashique et al.,2002a; Szabo-Rogers et al.,2008).
To determine potential roles of Wnt signalling during facial development, we have analyzed the expression patterns of WNTs- 1, 2B, 3A, 4, 5A, 5B, 6, 7A, 7B, 8B, 8C, 9A, 9B, 11, 11B, and 16; transducers of WNT signalling pathway CTNNB1 and LEF1; WNT antagonists FRZB1, DKK1, DKK2; and WNT receptors FZD1-8, FZD10 by in situ hybridization and compared their expression patterns (Tables 1, 2). Radioactive in situ hybridization on sections was also performed for some genes to clarify the expression patterns. The genes studied with both methods include WNT3A, 4, 5A, 6, 7A, 11, 16, DKK1, CTNNB1, LEF1.
Table 1. Summary of Expression Pattern of WNT Genes, Antagonists, and Intracellular Mediators of Canonical Signaling in the Facial Prominences
Epi and/or Mes
Frontonasal mass (fnm)
Lateral nasal prominence
These Wnts have never been shown to act via non-canonical pathways.
These Wnts can act in the canonical or non-canonical pathway.
There is also expression in the mandibular and maxillary branches of the trigeminal nerve.
There is also expression in the trigeminal ganglion.
Slightly higher in midline than other regions between stages 24–28.
Table 2. Summary of Expression Pattern of Frizzled Receptor Expression in the Stage-24 Facial Prominences
Lateral nasal prominence
Deeper, ventral to telencephalon and in periocular mesenchyme.
Continuous across fnm.
Higher in globular process.
Cranial edge nearest to fusion zone. Proximally expression is ubiquitous.
Low expression in mesenchyme higher in nasal slit epithelium.
Higher in the caudal edge mesenchyme.
Low expression in mesenchyme higher in nasal slit epithelium.
Slightly higher in globular processes, and in stomodeal epithelium.
Mesenchyme has low expression, stomodeal epithelium has light signal.
Ubiquitous, low in mesenchyme epithelium has light signal.
No data due to deep sections.
Caudal edge has higher expression.
Caudal medial edge distally and throughout mesenchyme proximally.
Ubiquitous, high, except in mesodermal core.
Signal in stomodeal epithelium.
No data due to deep sections.
Only deep sections through telencephalon studied, no expression.
Lateral epithelium and mesenchyme proximally.
No data due to deep sections.
Medial, oral edge in epithelium and mesenchyme, also in dorsal tongue.
Of the 30 genes surveyed, there were 8 genes with no detectable expression in the face as determined with whole-mount in situ hybridization (Supp. Fig. S1A–J, which is available online). These negative genes included: WNT1, 3A, 7B, 8B, 8C, 9A, 11B, DKK2. The apparent lack of expression could be due to levels of transcript being below the level for detection for at least some of the genes. WNT3A was further examined in detail in radioactive in situ. However, no expression in the face was observed (Supp. Fig. S1B, B'). WNT7A was also studied carefully in radioactive in situ since a previous study showed expression in the pharyngeal arches (Dealy et al.,1993). We also found radioactive probes bound to the first and second pharyngeal arch clefts but not to the facial prominence epithelium (data not shown but see Supp. Fig. 2SD). Further radioactive in situ studies, and RT-PCR analyses, would be necessary to determine conclusively whether facial expression exists for the remaining genes.
WNT2B has been reported to be expressed in the ectoderm overlying the midbrain and in ectoderm of the developing eye in the chick (Jasoni et al.,1999; Fokina and Frolova,2006). Between stages 15 and 18, we also found exclusively ectodermal expression for this gene. WNT2B was initially restricted to the ectoderm dorsal to the eye but later extends ventrally to cover the maxillary prominence (Fig. 1A, Supp. Fig. S2A). There is also expression in the pharyngeal arches (Fig. 1A, Supp. Fig. S2A) At stage 24, expression is mainly restricted to the extra-oral surface of the maxillary prominence, lateral edges of the mandibular prominence, and second pharyngeal arch (Fig. 1C). Sections confirm that signal is localized to the ectoderm (Fig. 2A, A'). The expression of WNT2B is not detected in the frontonasal mass at any stage except for faint expression at the tip of the globular processes visible at stage 24 (Fig. 1C). At stage 28, expression is lost in the maxillary prominences and is only retained in the lateral mandibular prominences (Fig. 1D).
WNT5A is exclusively expressed in the mesenchyme. At stage 15, there is strong expression in post-optic region (presumptive maxillary prominence) and first pharyngeal arch (Supp. Fig. S2B) and this pattern is continued to stage 18 (Fig. 1E). At stage 21, the WNT5A signal becomes stronger throughout the mesenchyme of the maxillary and mandibular prominences. Expression is also observed in the lateral nasal prominences and lateral edges of the frontonasal mass (Fig. 1F). By stage 24, WNT5A is predominantly confined to a band of cells midway along the caudal-rostral axis of the maxillary prominence, midline of the mandibular prominence, and is also observed across the frontonasal mass (Fig. 1G, 2B,B'). In the maxillary prominence, the expression of WNT5A shifts to the caudal half by stage 28. By this stage, there is less expression in the midline of the frontonasal mass where the midline cartilages and egg tooth will ultimately form (Fig. 1H).
WNT5B is closely related to WNT5A and expression patterns coincide to a large extent in the maxillary prominence and frontonasal mass. Weak expression in the pharyngeal arches is observed at stage 15 (Supp. Fig. S2C) and 18 (Fig. 1I). There is low expression of WNT5B throughout the facial prominences (Fig. 1J) until stage 24. Then weak WNT5B signal is observed in the lateral edge of the frontonasal mass but relatively stronger expression is present in the maxillary prominences (Figs. 1K, 2C). At stage 28, there is expression in the frontonasal mass and maxillary prominences overlapping with WNT5A and thus some redundancy in function is possible. In contrast to WNT5A, there is no WNT5B expression in the mandible between stages 24–28 (Fig. 1K,L).
WNT9B is expressed exclusively in the ectoderm, similar to WNT2B. In general, however, WNT9B signal is more widespread. At stage 15, there is strong expression in the ectoderm dorsal to the eye and first pharyngeal arch but the signal does not extend to cover the presumptive maxillary prominence (Supp. Fig. S2D). During the later stages (18–21), expression in the extraoral ectoderm of the maxillary and mandibular prominences is present with no detectable signal in the frontonasal mass (Fig. 1M–N). At stage 24, there is increased expression of WNT9B in the frontonasal mass ectoderm and thus all facial prominences become positive for WNT9B (Figs. 1O, 2D–D”'). Interestingly, there is a sharp boundary to WNT9B expression that leaves a band of non-expressing epithelium at the periphery of all the facial prominences. WNT9B overlaps exactly with WNT2B in the maxillary prominences at stage 24. At stage 28, expression of WNT9B is present but at lower levels in the maxillary prominences (Fig. 1P) whereas WNT2B is completely lost (Fig. 1D). WNT9B becomes concentrated in the centre of the frontonasal mass where the egg tooth will later form (Fig. 1P). Our data extend the recent results of another study (Hu and Marcucio,2009) in which stage-24 data were presented.
Like the WNT5 genes, WNT11 is exclusively expressed in the mesenchyme. The expression of WNT11 is undetectable in the face prior to stage 18 (Supp. Fig. S2E, Fig. 1Q). We found that WNT11 is first expressed as a small spot in the maxillary region of the face at stage 18 (Fig. 1Q). At stage 21, WNT11 transcripts are clearly detectable in maxillary prominences and also two faint bands of expression are seen in each of the cranial mandibular prominences (Fig. 1R). Later these domains broaden, covering the anterior maxillary and craniolateral mandibular prominences (Figs. 1S, 2E,E'). WNT11 is also expressed in the lateral mesenchyme of the second pharyngeal arch (Fig. 2E”). At stage 28, the expression is shifted laterally in the maxillary prominence and towards the craniolateral edge of the mandibular prominence and is found in the periocular mesenchyme (Fig. 1T). It is also interesting that WNT11 is not detected in the frontonasal mass at any of the developmental stages. Radioactive in situs showed an identical distribution of transcripts to the whole-mounts, confirming the lack of expression in the frontonasal mass at stages 20 and 24 (data not shown).
Like WNT11, the expression of WNT16 is not detected in the face until stage 18 (Supp. Fig. S2F, Fig. 3A). The first signal appears as a small spot in the maxillary, caudal mandibular, and cranial second pharyngeal arch clefts (Fig. 3A). At stage 20–21, increased expression of WNT16 is seen in the epithelium of lateral nasal prominence and there is an interesting restriction of expression to the caudal nasal pit epithelium, which includes the lateral edge of the frontonasal mass (Figs. 3B, 4A–B'). Expression of WNT16 is also seen in the mandibular branches of the trigeminal nerve (Fig. 3B). There is also faint expression in the anterior epithelium of maxillary prominences, also confirmed by radioactive in situ (Figs. 3B, 4B,B'). At stage 24, paired expression domains in the mandibular prominence corresponding to the mandibular nerve continue to be present on either side of the midline (Fig. 4C'). In saggital sections of stage-24 and -28 embyos, both the maxillary and mandibular branches of the trigeminal nerve are positive for WNT16 (Fig. 4D,D',F,F'). At stage 28, WNT16 is expressed in a subregion of the epithelium lining the nasal passages (Fig. 4E–F') that appears to exclude the specialized olfactory epithelium. It is likely that WNT16 marks the respiratory epithelium (Croucher and Tickle,1989).
WNT4 and WNT6 have recently been shown to act by both canonical and noncanonical pathways. When bound to Fzd6, Wnt4 activates canonical signalling in kidney cells (Lyons et al.,2004). However Wnt4 has also been shown to activate the novel non-canonical pathway, p38 MAPK, in the context of osteogenesis (Chang et al.,2007). WNT6 activates the canonical (Linker et al.,2005) and non-canonical pathways (Schmidt et al.,2007), depending on the context. WNT4 is expressed in the external and stomodeal/oral epithelium of the entire head between stages 20 and 25 (Fig. 5A–D). There is no expression in the nasal pit at stage 20 (Fig. 5A) or in the nasal slit at stage 25 (Fig. 5C). In contrast, WNT6 is barely detectable at stage 20 (Fig. 5E) and is increased at stage 24 in the external head ectoderm. There is additional expression near the lateral edges of the frontonasal mass but not extending up into the nasal slit (Fig. 5F). The intraoral or medial surfaces of the maxillary prominences also express WNT6 distally (Fig. 5G) but not proximally (Fig. 5H). Very low levels of WNT4 and WNT6 were recently detected in the nasal slit epithelium with radioactive in situ hybridization by others (Hu and Marcucio,2009). It is possible that this is because the expression was very restricted in their study and that we may not have included this small region in our sections. Interestingly, unlike the other epithelially expressed WNT genes, it was not possible to observe signals in whole-mount in situ hybridization, perhaps indicating lower abundance of these transcripts.
Expression of WNT Antagonists
FRZB antagonizes both canonical and non-canonical WNT signalling by direct binding of its cysteine-rich domain to the WNT ligands (Bhanot et al.,1996; Bafico et al.,1999). FRZB1 is first expressed in the cranial neural crest (Ladher et al.,2000). Our data at later stages agree with that originally published (Ladher et al.,2000). However, we were able to resolve more clearly the expression around the nasal placode at stage 18 (Fig. 3E). There is also medially restricted expression in the maxillary prominences and relatively higher expression in the midline of the mandibular prominence at stage 25 (Fig. 3G). We also confirmed mesenchymal expression in sections through the facial prominences (Fig. 2G). Until stage 23, FRZB1 transcripts are abundant throughout all facial prominences (Supp. Fig. S3A). From stage 25, mesenchymal expression of maxillary and frontonasal prominences is down-regulated compared to mandibular prominences where it is strongly expressed in the medial portion (Figs. 2G, 3G,H, Supp. Fig. S3B). This striking down-regulation is similar to previous published results (Ladher et al.,2000). At stage 28, there is a small area of expression at the cranial edge of the mandibular prominence and caudal edge of the frontonasal mass (Fig. 3H).
DKK1 specifically antagonizes canonical WNT function by preventing WNT ligands from interacting with LRPs (low-density lipoprotein receptor-like proteins; Niehrs,2006). In the mouse, DKK1 is expressed in the first pharyngeal arch (Monaghan et al.,1999) similar to what we have observed in the chicken at stage 15 (Fig. 6A,A', Supp. Fig. S2H). There is no expression in the frontonasal region until stage 17–18 when expression is seen around the nasal pit (Fig. 3I). The first maxillary expression is also observed at stage 17–18. At stage 21, DKK1 is expressed both in the epithelium and mesenchyme of the maxillary and mandibular prominences (Figs. 3J, 6B,B'). In contrast, in the frontonasal mass expression is restricted to the epithelium surrounding the nasal slit (Figs. 3J, 6B,B'). Stage-24 embryo sections confirmed that DKK1 is restricted to the ectoderm of the nasal slit and caudal edge of the frontonasal mass (Fig. 2H, H”). This trend of exclusively epithelial expression for DKK1 in the frontonasal mass is continued at stage 28 where the caudal edge or frontonasal epithelial zone is positive (Fig. 3L, 6D,D' (Hu et al.,2003). However, the nasal slit epithelium is no longer expressing DKK1. At stage 24, expression in the maxillary prominences is restricted to a cranial and caudal domain on the medial side of the prominence (Figs. 2H”',3K, 6C,C') and at stage 28 a narrow band of mesenchyme subadjacent to the medial epithelium continues to express DKK1 (Fig. 3L, 6D,D'). In the mandibular prominence, there is a reduction of expression from stage 24–28 compared to stage 20–21, although in saggital sections some mesenchymal expression can still be detected (Fig. 6C,C'). This suggests DKK1 can modulate canonical WNT signalling in a temporally and spatially specific manner within each of the facial prominences.
Expression of Canonical WNT Pathway Genes
Between stages 15–18, CTNNB1 expression is highest around the nasal pits (Supp. Fig. S2I). From stage 21, there is enriched expression around the nasal pits and faint bands appeared at the caudal edge of the maxillary prominences (Fig. 3N, Supp. Fig. S4B,B'). At stage 24 and 28, expression of CTNNBI appeared to be more ubiquitous in the face with the strongest signal being localized to the nasal slits (Fig. 3O,P). Expression is also observed in the mesenchyme; however, it was hard to detect the signal above the background. These results are clarified by radioactive in situs, which showed that in fact CTNNB1 expression is very abundant in both the epithelium and mesenchyme at all stages (Supp. Fig. S4A–F). Thus, this important mediator of canonical WNT signaling is expressed in all regions of the developing face.
We found the nuclear mediator of canonical WNT signalling, LEF1, is expressed in the mesenchyme of the maxillary and pharyngeal arches at stage 15 (Fig. 7A,A', Supp. Fig. S2J). Similar patterns are observed at stages 17 to 28 except that maxillary expression becomes more concentrated in the medial edge (Figs. 3Q–T, 7C',D',F'). In the mandibular prominence, expression is throughout at stage 20 but then becomes more abundant near the midline by stage 24 (Figs. 3R–T, 7C',E'). In the frontonasal mass, expression is detected in the midline at stage 24 and this expands across the medial-lateral axis by stage 28. LEF1 transcripts were also found in the lateral edges of the lateral nasal processes between stages 24 and 28 (Fig. 3S,T). Sections confirmed that LEF1 is also expressed in the pharyngeal endoderm of the first pharyngeal pouch (Fig. 7B′,C′,E′).
Expression of FZD Receptors
Expression data were published for several FZD genes in HH stages 3–20 (Stark et al.,2000). Therefore, we focused on older stages. Diffuse mesenchymal expression in ectomesenchymal cells of the maxillary and mandibular prominences was observed for FZD1,2,3,4,5,6 and 7. FZD1, 2, 4, and 7 are also expressed in the frontonasal mass. FZD1,4,7 were also highly expressed in the periocular mesenchyme whilst FZD4 and 7 were highly expressed in the medial maxillary mesenchyme. FZD2 and 7 were the only receptors noticeably lower in the mesodermal core of the mandibular prominence. To confirm this, we hybridized adjacent sections to MYOD, a muscle-specific marker expressed in the mesodermal core. Expression patterns of MYOD, FZD7, and FZD2 were exactly complementary (Fig. 8G”, G”', and data not shown). The most striking expression was noted for FZD10 where medial mandibular prominence had intense expression in the mesenchyme just beneath the epithelium (Fig. 8I, I'). In addition to mesenchymal expression, several of the genes had signal in the epithelium including FZD3, 4, 5, 6, 8, and 10 (Fig. 8E,E',F,F',H,H',I,I').
Our study provides a comprehensive overview of the expression patterns of WNT ligands, the major effector molecules, antagonists, and receptors in the developing face. The data suggest where, when, and which endogenous ligands are controlling different aspects of facial morphogenesis. These basic descriptive studies are necessary before we can go on to test the functions of WNT signalling in craniofacial development.
Multiple FZD Receptors Are Present in the Developing Face
All of the facial prominences express several of the FZD genes. Therefore, there would be multiple receptors capable of binding to the WNT ligands. The expression of FZD10 is intriguing and it may be that only a restricted set of WNTs is capable of activating this receptor. At the present, the binding specificities of the FZD receptors are not well understood. However, it does appear that choice of receptor, in part, determines whether some WNTs activate canonical or non-canonical pathways (van Amerongen et al.,2008). It is, therefore, not possible to predict on the basis of gene expression whether ligands act via canonical or non-canonical WNT pathways. For example, both WNT5A and WNT11 have been shown to activate the canonical pathway in some experimental systems depending on which FZD receptor is present (Tao et al.,2005; Mikels and Nusse,2006). Therefore, one needs to perform functional experiments to determine which of the ligands act via the canonical or non-canonical pathways.
Canonical WNTs Are Localized to the Epithelium
An important overarching observation is that all of the WNTs for which only canonical signalling has been experimentally proven are ectodermally restricted. Since both CTNNB1 and LEF1 and numerous FZD receptors are present throughout the facial mesenchyme, canonical WNT signalling could take place in areas where the ligands WNT- 2B, 9B, 16, and 7A are expressed. Others have determined with reporter assays in mouse and chicken that canonical WNT activity is present in the face. In the chicken, the delivery of the reporter has been conducted with a lentivirus (Brugmann et al.,2007) or with electroporation (Hu and Marcucio,2009). The former study found a higher level of activity in the midline of the frontonasal mass while the latter found more ubiquitous expression in all areas that took up the plasmid including those lateral to the midline. The authors (Hu and Marcucio,2009) attributed the differences in reporter activity to the DNA delivery method and we agree with this interpretation. Thus, it is not clear yet whether all stages or areas of the chicken face have active WNT signalling.
There is generally good correspondence between our chicken WNT ligand expression and the mouse Wnt reporter data (TOPGAL, BATgal; Lan et al.,2006; Brugmann et al.,2007). For example, there is initially very low to absent activity in the frontonasal mass at E9.5, a stage equivalent to stage 15 in the chicken embryo, which is consistent with the fact that none of the WNT ligands could be detected in the frontonasal mass at stage 15. The mouse β-gal staining is strongest in the ectoderm and in the subadjacent mesenchyme in the medial nasal, maxillary, and mandibular prominences. Although there are two WNT ligands present in the ectoderm that could activate non-canonical signalling (WNT4 and WNT6), the expression of the mouse reporter argues in favour of canonical activity taking place at sites of epithelial-mesenchymal interactions in the face. Furthermore, the injection of DKK1 adenovirus, which only blocks the canonical pathway, leads to severe reduction in growth of the frontonasal mass prominence (Brugmann et al.,2007).
Interestingly, several of the WNTs with non-canonical activity, WNT5A, 5B, and 11 are expressed only in the mesenchyme. One of these, Wnt5a, is essential for outgrowth (Yamaguchi et al.,1999; He et al.,2008). These Wnts may also be essential for skeletogenesis and myogenesis as in other regions of the body (Anakwe et al.,2003; Church et al.,2003; Gros et al.,2009). Further experiments designed to selectively block the non-canonical pathways are needed to see whether facial growth and cell differentiation are affected.
Finally, there are overlapping patterns of expression between the canonical WNTs and non-canonical WNTs. Thus, there may be cross-talk between the pathways. Wnt5a has, for example, been shown to antagonize canonical signalling via the Ror2 receptor (Mikels and Nusse,2006). Such mutual antagonism may limit the extent of canonical signalling taking place in the mesenchyme.
WNT Activity During Fusion of the Lip and Palate
The contact of the frontonasal mass (medial nasal prominences in mice) and maxillary prominences to form the lip is also likely to involve WNT signalling. There is already evidence that Wnt9b is involved in lip fusion as there is cleft lip in some of the mutant mice (Carroll et al.,2005; Juriloff et al.,2006). There is high expression of LacZ precisely in the fusing tips of the medial nasal and maxillary prominences at the time of fusion (Lan et al.,2006; Brugmann et al.,2007). Further evidence that canonical Wnt signalling is essential for lip fusion is that Dkk1 overexpression in mice causes cleft lip (Brugmann et al.,2007). Fusion of the facial prominences requires growth such that the prominences can make contact and permit the fusion of competent epithelia. Our data shows that WNT9B is not expressed in the bilayered epithelial seam or mesenchymal bridge, arguing that if the expression patterns are conserved in mice, WNT9B has a crucial role in outgrowth of the facial prominence mesenchyme, thereby facilitating contact.
Additional Wnts may also be involved and from our data we now suggest that WNT16, which also signals via the canonical pathway, could be important in forming the epithelial seam between the frontonasal mass and maxillary prominences. WNT16 is the only WNT ligand expressed in these fusing epithelia. In addition, it is expressed prior to (stage 20–24) and during fusion (stage 28). There are very few other signals (BMP2, 4, 7; Ashique et al.,2002a) expressed on both sides of the fusion zone in the frontonasal mass and maxillary epithelia. Thus far, no detailed descriptions of Wnt16 in the mouse have been reported so it will be interesting to see whether the patterns are conserved and whether this gene is required for lip fusion. WNT11 is also potentially involved in fusion as it is present in the mesenchyme of the cranial maxillary prominence at stage 24. However, expression is shifted laterally at stage 28 just prior to fusion. Therefore, it is possible that WNT11 may initially control outgrowth but not participate actively in fusion of the lip.
The expression of the two WNT antagonists, FRZB1 and DKK1, in our study is very dynamic in the fusion zone. Both are present at high levels at stage 24 but are dramatically decreased at stage 28 when fusion is beginning. A similar pattern has been reported for SFRP2 (Ladher et al.,2000). These data, together with the above, suggest that canonical WNT activity is carefully controlled and is only allowed to increase at the moment when fusion is taking place. A similar observation was made by us previously with the BMP antagonist, NOGGIN (Ashique et al.,2002a). Decreased expression of NOG occurred in the globular processes just prior to fusion, allowing an increase in BMP activity.
Another potentially important role for WNT signalling is in formation of the secondary palate. Here we can point to the regionally restricted expression of non-canonical WNTs in the maxillary mesenchyme as potential players in palatogenesis. Recent analysis of the Wnt5a−/− mice (He et al.,2008) shows there are clefts of the secondary palate. Wnt5a is needed for outgrowth and directional cell migration within the palatal shelves (He et al.,2008). These data and the lack of WNT reporter activity in the palatal shelves (He et al.,2008) point to a likely predominance of non-canonical signalling in palatal fusion.
In summary, there is much to learn about WNT signalling during facial morphogenesis. Our studies have helped to identify the main WNT ligands that are likely to be involved in facial patterning, lip fusion, and nasal morphogenesis. Future work will test the hypotheses we have raised and thereby build a more complete picture of molecular signalling during craniofacial development.
In Situ Hybridization
Fertilized chicken eggs were incubated at 38°C. Embryos were fixed overnight at 4°C in 4% paraformaldehyde. Embryos were washed twice in phosphate buffered saline with 0.1% Tween-20, dehydrated in methanol, and stored at −20°C. Whole mount in situ hybridization was performed, as previously described, but using an Intavis in situ hybridization robot (Song et al.,2004). Selected stained embryos were processed through isopropanol, isopropanol:paraffin (50:50) and then embedded in paraffin. Embryos were sectioned at a thickness of 7μm and counterstained with eosin. Section in situ hybridization was also performed with antisense 35S-labeled antisense probes using published protocols (Wilke et al.,1997).
The following individuals provided gallus cDNAs for this study: C. Marcelle, WNT1, 3A 5B, 11, CTNNB1 (Marcelle et al.,1997; Geetha-Loganathan et al.,2005); C. Hurle, DKK1 (Grotewold and Ruther,2002), C.J. Tabin, LEF1 (Gavin et al.,1990; Bergstein et al.,1997; Kengaku et al.,1998; Jasoni et al.,1999); S. Chapman, WNT8B (Chapman et al.,2004); K.G. Storey, WNT8C (Olivera-Martinez and Storey,2007); E. Frolova, WNT5A, WNT9B (Fokina and Frolova,2006); P.A. Krieg, WNT11B (Hardy et al.,2008); P. Francis-West, FRZB1 (Ladher et al.,2000); T. Nohno, WNT4, FRZD 1, 2, 3, 4, 6, 7, 8. 10 (Kengaku et al.,1997; Kawakami et al.,2000); and L. Burrus, FRZD5 (Cauthen et al.,2001). DKK2 was purchased from MRC Geneservice (833 bp, chEST339h24). We independently cloned a 758-bp fragment of WNT16 using published sequence (GenBank accession number: AY753296). For non-radioactive in situs, we used WNT6 (1,500 bp) as described (Rodriguez-Niedenfuhr et al.,2003) and for radioactive in situs we used the WNT6 probe from A. McMahon (Hollyday et al.,1995).
This work was funded through grants from the CIHR to J.M.R. and BBSRC to P.F.W. P.G.-L. and S.N. are supported by MSFHR post-doctoral fellowships. We also thank GR. Handrigan for helping in cloning WNT16. We extend our gratitude to all those who provided us with in situ hybridization probes.