A SHH-responsive signaling center in the forebrain regulates craniofacial morphogenesis via the facial ectoderm

Previous research has examined the role of various epithelia during development of the craniofacial complex. For example, signals from endoderm and ectoderm are required for skeletogenesis in the face(Brito et al., 2006; Couly et al., 2002; Graham et al., 2005; Ruhin et al., 2003; Tyler and Hall, 1977). Our research focuses on mechanisms regulating development of the upper jaw. The skeleton in this region is formed from neural crest-derived cells generated along the dorsal surface of the forebrain and midbrain(Couly et al., 1993; Evans and Noden, 2006; Noden, 1978). These cells migrate adjacent to the forebrain, reside in the frontonasal (FNP) and maxillary processes (MXP), and are encased by ectodermal and neuroectodermal epithelia. Subsequent interactions among these tissues control morphogenesis of the skeleton (Hu et al.,2003; Marcucio et al.,2005; Schneider and Helms,2003; Schneider et al.,2001).

Patterning and growth of the upper jaw is controlled by discrete signaling centers located in the ectoderm. The frontonasal ectodermal zone (FEZ)regulates development of the distal tip of the upper jaw(Hu et al., 2003). In chicks,the FEZ forms at approximately stage 20 [HH20(Hamburger and Hamilton,1951)] when Shh transcripts are detected in ectodermal cells comprising the roof of the stomodeum(Marcucio et al., 2005). At this time, Shh- and Fgf8-expressing cells form a boundary that spans the mediolateral axis of the FNP. Shortly after this boundary forms, Fgf8 is downregulated and becomes restricted to epithelium near the nasal pit, whereas Shh expression is maintained for a longer period of time. Retroviral mapping studies revealed that the boundary presages the distal tip of the upper beak, and our studies demonstrate that the FEZ regulates dorsoventral patterning within the distal tip of the upper jaw(Hu et al., 2003). Mice have a similar signaling center, but Shh transcripts do not span the width of the upper jaw. Rather, distinct domains of Shh expression are associated with the left and right median nasal processes(Hu and Marcucio, 2008). In addition to Shh and Fgf8, multiple genes encoding bone morphogenetic proteins (Bmp2, Bmp4 and Bmp7) are expressed in the FEZ (Abzhanov and Tabin,2004; Foppiano et al.,2007; Hu et al.,2003; Marcucio et al.,2005), but the role of these molecules is unknown. However, our observations indicate that the FEZ functions, in part, by regulating expression of Bmps in neural crest mesenchyme(Hu and Marcucio, 2008). In turn, BMP signaling regulates growth of the upper jaw(Abzhanov et al., 2004; Wu et al., 2006; Wu et al., 2004). Recently,the nasal pit has been shown to pattern the jaw skeleton. FGF signaling in this region controls cell proliferation and antagonizes Bmp4expression (Szabo-Rogers et al.,2008). Thus, interactions among the brain, ectoderm and neural crest sculpt this region of the head.

SHH signaling within the forebrain is required for induction of Shh in the FEZ (Marcucio et al.,2005), and differences in Shh expression correlate with morphological variation in mice and chicks. Therefore, we hypothesized that the forebrain may generate morphological variation by regulating signals from the ectoderm that control development of the upper jaw. To test this, we activated SHH signaling within the forebrain after emigration of neural crest cells was complete. Our results demonstrate that a SHH-responsive center in the brain imprints information on the ectoderm that controls morphogenesis of the upper jaw.

Affi-gel blue beads (50-100 mesh, 200-250 μm diameter; BioRad) were soaked in recombinant SHH-N protein [800 μg/ml in PBS 0.1% bovine serum albumin (BSA), Ontogeny] for 1 hour at 37°C; beads were placed into the forebrain of HH11 embryos (see Fig. S1D in the supplementary material). Alternatively, beads soaked in SU5402 (3 mg/ml) were implanted in the FNP of∼HH17 embryos.

Mouse and chick embryos were fixed in 4% paraformaldehyde (4°C overnight), transferred to PBS containing 0.01% ethidium bromide and photographed using a Leica MFLZIII dissecting microscope with epifluorescence. Embryos were dehydrated, embedded in paraffin and sectioned (8 μm). Mouse embryos were collected at E10.5, E11 and E12. The day we observed a plug was E0. No further staging was performed.

In situ hybridization was performed on sections or in whole mount as described (Albrecht et al.,1997). Subclones of Shh, Nkx2.1, Fgf8, Bmp2, Bmp4, Bmp7, Wnt4,Wnt6, Wnt9b, Wnt14, Raldh2 and Zfhxb1 were linearized for transcription of ³⁵S- or dig-labeled antisense riboprobes. Sections were counterstained with bis-benzimide, and images are superimpositions of fluorescent and pseudo-colored dark-field image.

Cartilage (Red) was visualized on sections with Safranin O/Fast Green(SO/FG) staining (Lu et al.,2005). Alternatively, embryos were stained with Alcian Blue and Alizarin Red (Wassersug,1976).

Twenty minutes before sacrifice, 1 μl of BrdU (Zymed, South San Francisco, CA) was injected into the vitelline vein. Embryos were processed as above. BrdU incorporation was assessed by immunohistochemistry using diaminobenzidine (DAB) followed by counterstaining with hematoxylin (Zymed). The number of proliferating cells was determined on images captured using Adobe Photoshop of the Olympus CAST system. Analysis of Variance (ANOVA) was performed on medial and lateral sections separately to determine statistical significance, and on treated and control sides of embryos 24 hours after implantation of SU5402 beads.

DNA fragmentation was examined using a TUNEL kit following the manufacturer's instructions (Apoptag Plus, Intergen).

To visualize activation of the Wnt pathway we injected the Top-Gal plasmid DNA (200 nl; 2 μg/μl) into the space between the ectoderm and the forebrain on the right side of embryos. Then, electroporation [five pulses (10 volts) 20 mseconds apart] was used to transfect cells. Twenty-four hours later, embryos were fixed in 0.4% gluteraldehyde (15 minutes at room temperature), washed and then standard X-Gal reaction was performed.

Prior to experimentation we analyzed the timing of neural crest emigration from the brain by immunohistochemical detection of the neural crest marker HNK-1. At HH9-, emigration of neural crest cells from the neural tube was under way (see Fig. S1A in the supplementary material), and cells had moved laterally and rostrally by HH9 (see Fig. S1B in the supplementary material). At HH10, the HNK-1-positive cells reached the lateral and rostral edge of the dorsal surface of the embryo (see Fig. S1C in the supplementary material). We activated the SHH pathway by placing a bead soaked in the N terminus of SHH into the lumen of the forebrain at HH11. Thus, generation of neural crest cells destined for the upper jaw was not affected, and molecular, cellular and morphological changes resulted from alterations in the interactions among the forebrain, neural crest cells and surface ectoderm.

SHH signaling specifies the dorsoventral axis of the brain (reviewed by Lupo et al., 2006). Therefore,we assessed the extent to which the forebrain was altered by examining expression of the ventral markers Nkx2.1 and Shh, and the dorsal marker Pax6 near the sagittal midline of embryos. Within 24 hours (∼HH15), we observed an expansion of Nkx2.1 and Shh expression domains (Fig. 1A,B,D,E), and a downregulation of Pax6(Fig. 1C,F). At 48 and 72 hours, we observed similar changes in gene expression in the brain (not shown). Additionally, we observed changes in Fgf8 expression; 24 hours after bead placement, controls exhibited Fgf8 expression in the ectoderm and in the brain (Fig. 1G). However, in the brain of treated embryos Fgf8expression was mosaic (Fig. 1J). At 60 hours (∼HH20/21), Fgf8 transcripts were detected in the forebrain, optic recess and ectoderm of control embryos(Fig. 1H), but in treated embryos expression in the optic recess and facial ectoderm was absent(Fig. 1K). In control and treated embryos, Fgf8 expression was downregulated in the facial ectoderm at 72 hours (Fig. 1I,L). At each time point we never observed Fgf8transcripts in the optic recess of treated embryos, but we did observe a continuous domain of Shh and Nkx2.1 expression in the floor of the brain, which is normally disrupted by the optic recess. Thus, our treatment disrupted the normal appearance of the optic recess.

We confirmed that these changes resulted from altered specification of the brain rather than alterations in cell survival. We only observed TUNEL-positive cells in basal regions of the brain in control and treated embryos at 24, 48 and 60 hours after bead implantation. No significant signs of cell death were detected in dorsal regions of the brain, in the surface ectoderm or in the mesenchyme of treated embryos (see Fig. S2 in the supplementary material).

The changes in the brain agree with studies illustrating the role of SHH in patterning neural tissues (reviewed by Lupo et al., 2006). However,our goal was to examine the relationship between patterning the forebrain and development of the facial skeleton. Therefore, we analyzed the ontogeny of the malformations in treated embryos. At 48 hours (∼HH19), mild defects were apparent (n=8). The relationship between the nasal pits and the MXP were different in control (bead soaked in PBS; n=7) and experimental embryos (Fig. 2A,B). At this time, neural crest cells have arrived in the FNP but growth has not begun yet. Similarly, in mice, neural crest cells have begun populating the upper jaw anlagen, but there has not been much growth by this time(Fig. 2C). However, at this time, the chick and mouse faces appear distinct. The chick has well-developed nasal pits whereas these are small in mice. Thus, unique attributes characterize these regions of the face from the beginning of development. By 72 hours (∼HH23), treated embryos exhibited severe signs of malformation. Normally, the FNP has begun to grow at this time(Fig. 2D). In treated embryos the forebrain was small, the eyes were small, the nasal pits were malformed,and the FNP and LNP exhibited aberrant growth, but the MXPs appeared unaffected (Fig. 2E)(n=14). Therefore, we focused on morphological changes within the FNP. In treated embryos, growth was occurring in lateral regions and converging toward the middle part of the FNP. These embryos resembled E11.0 mouse embryos (Fig. 2F). At this time in mice, growth was occurring in lateral regions near the nasal pit,and was converging towards the middle of the upper jaw anlagen. Hence, in mice, the midline of the upper jaw anlagen is filled in by lateral to medial growth of the embryonic primordia. By 96 hours (∼HH28) after bead implantation, the phenotype had become more severe. In controls, growth was centered in the middle of the FNP (Fig. 2G). However, in treated embryos growth was concentrated in lateral regions of the FNP, and these developing primordia had converged to the midline to touch each other (Fig. 2H). In mice at this time, the right and left median nasal processes had expanded and abutted each other in the midline where a furrow formed (Fig. 2I). Interestingly, in severe cases (n=8/17 at 96 hours, and 1 at day 13),some of the treated chicks had midfacial clefts (see Fig. S3 in the supplementary material), a malformation observed in a number of mouse mutants,including the Fgf8 hypomorph (D.H. and R.S.M., unpublished). Owing to changes in the growth characteristics within the FNP, we assessed the extent to which the FEZ in the treated chicks was altered.

Spatial organization of the FEZ correlates with growth patterns observed in chicks and mice (Hu and Marcucio,2008). In normal chicks, Shh expression was evident along the tip of the growing FNP. The Shh expression domain in these embryos was small and was centered within the medial ectoderm of the expanding FNP (Fig. 2J). At 72 (data not shown) and 96 hours after treatment, Shh expression in the FEZ was downregulated in the midline (Fig. 2K). This expression pattern resembles that observed in mice at E12. Shh transcripts were not detected in the midline(Fig. 2L). At 72 hours after implantation of beads, the total number of proliferating cells was reduced in the midline, but unaltered in lateral regions, of the FNP in treated embryos compared with controls (Fig. 3). Thus, the FEZ controls growth in the medial region of the upper jaw.

Embryos were allowed to develop to later stages to examine formation of the upper beak. Eleven (n=5; not shown), 12 (n=5)(Fig. 4A-D) and 13(n=6) (Fig. 4E,F) days after bead implantation, the eyes were small, the upper beak was shortened,the nasal capsule was underdeveloped and bones comprising the proximal portion of the upper jaw were small and malformed. The premaxillary bone located in the distal part of the upper jaw appeared smaller, but was not as severely affected as the more proximal skeletal elements. As changes in the FEZ corresponded to the reductions in the premaxillary bone, we focused on malformations in the proximal region of the jaw and nasal capsule.

We identified molecular changes in other regions of the face that may generate the phenotype that we observed. A number of genes in, or adjacent to,the nasal pit were altered within 72 hours of bead placement. Wnt4was normally observed in the nasal pit epithelium(Fig. 5A), but in treated embryos Wnt4 transcripts were not detected(Fig. 5E). Similarly, Wnt6 was normally expressed in two domains in the nasal pit(Fig. 5B). In treated embryos, Wnt6 was reduced in these areas(Fig. 5F). Wnt9b was normally expressed in the ectoderm covering the lateral region of the FNP, the lateral nasal processes and the MXPs, but was excluded from the epithelium of the nasal pit (Fig. 5C). In treated embryos, we observed malformations in this region of the face that distorted expression of Wnt9b. The epithelium of the nasal pit normally intersects the surface ectoderm at right angles, but in treated embryos, the nasal pit was sloped. Therefore, when the face of the embryos were viewed from the front, the nasal pit epithelium was visible and this tissue did not express Wnt9b (Fig. 5G).

To examine the extent to which canonical WNT signaling was altered in treated embryos, we electroporated the TOP-Gal plasmid(DasGupta and Fuchs, 1999)into the face 48 hours after treatment. Twenty-four hours later we observed strong expression of β-galactosidase throughout the FNP and MXPs in control embryos (Fig. 5D). Similarly, we observed strong expression of β-galactosidase in treated embryos (Fig. 5E). Although we could not quantify activation of the WNT pathway owing to variation in electroporation efficiency among embryos, our results indicate that the canonical WNT pathway was active in medial and lateral regions of the FNP in all embryos.

Additionally, 72 hours after bead placement, Raldh2 was expressed in mesenchyme of the maxillary and lateral nasal process(Fig. 5I). After treatment, we observed expansion of Raldh2 in the mesenchyme of the FNP(Fig. 5M). By contrast, Fgf8 was normally expressed in medial and lateral epithelium of the nasal pit (Fig. 5J). After activation of SHH signaling in the brain, we observed that the nasal pit was malformed and Fgf8 expression was not apparent in the medial epithelium (Fig. 5N). In the developing neural plate, Zfhx1b (Zeb-2/Sip1) is positively regulated by FGF signaling(Sheng et al., 2003). Therefore, we wanted to determine whether expression of Zfhx1b was altered in treated embryos. Normally, Zfhx1b is expressed in the mesenchyme of the FNP adjacent to the ectoderm(Fig. 5K), and we observed downregulation of Zfhx1b in mesenchyme of the FNP(Fig. 5O). However, whether this change was a direct result of downregulation of FGF signaling is unknown. Finally, FGF signaling represses expression of Bmp4 in lateral regions of the FNP (Szabo-Rogers et al.,2008; Wu et al.,2006), and we observed the expected upregulation of Bmp4expression in lateral regions of the ectoderm covering FNP in treated embryos(Fig. 5L,P). However, we did not detect changes in mesenchymal expression of Bmp4, Bmp2 and Noggin (see Fig. S4 in the supplementary material). Hence, multiple signaling pathways were altered after activation of SHH signaling within the brain.

Activation of SHH signaling in the brain created defects in the proximal region of the upper jaw and nasal capsule(Fig. 4), and these alterations were accompanied by changes in expression of Fgf8 (Figs 1, 5). To further explore these changes, we performed whole-mount in situ hybridization to compare expression of Fgf8 in control and treated embryos. Within 48 and 72 hours of treatment, the nasal pits were malformed and Fgf8 was downregulated near the nasal pit (Fig. 5Q-T).

Next, we blocked FGF signaling by implanting beads soaked in the FGF receptor inhibitor SU5402 (Eblaghie et al., 2003; Szabo-Rogers et al., 2008) into the mesenchyme comprising the right side of the FNP at HH17. We examined the skeletons at day 13, and we observed a similar phenotype in SHH- and SU5402-treated embryos (Figs 4, 6). The nasal capsule was reduced and proximal skeletal elements were small or absent in treated embryos(Fig. 6C-F). At 24 hours after bead placement, we did not observe evidence of increased cell death (see Fig. S5 in the supplementary material), but we detected reduced cell proliferation in mesenchyme adjacent to the bead (Fig. 7). Thus, blockade of FGF signaling generates a phenotype that resembles that produced when we activated the SHH pathway in the brain. However, we are unable to determine the extent to which blockade of signaling by specific Fgf ligands produce the phenotype.

The brain participates in regulating morphogenesis of the face. Our current work revealed that a SHH-responsive signaling center communicates patterning information to the ectoderm that covers the developing upper jaw. The brain regulates expression of signaling molecules in the ectoderm that then control morphogenesis of the upper jaw. Morphogenesis of the distal region of the upper jaw is regulated by the FEZ (Hu et al., 2003), whereas the proximal part of the upper jaw is controlled by signals near the nasal pit(Szabo-Rogers et al., 2008). Our results indicate that development of these modules is integrated by signals coming from the brain. Previously, we demonstrated that the forebrain regulates the onset of Shh expression in the FEZ(Marcucio et al., 2005). Now,we demonstrate that the brain regulates the spatial expression domain of Shh in the FEZ and expression of other signaling molecules near the nasal pit. Thus, the brain directs development of the middle and upper jaw by establishing discrete signaling centers in the ectoderm that covers the developing upper jaw. These signaling centers are then responsible for coordinating growth and patterning of the upper jaw anlagen.

Interactions between the brain, ectoderm and neural crest occur as part of a set of signaling interactions that pattern the facial skeleton. During early stages of facial development, tissue interactions establish regional identity within ingressing neural crest cells. Epithelial-mesenchymal interactions mediated by BMP4 and FGF8 define the maxillary and mandibular regions of the first pharyngeal arch (Shigetani et al.,2000). These signaling interactions may act to establish unique patterns of gene expression in the adjacent mesenchyme that control regional characteristics within the skeleton. In part, nested expression of Dlx genes in mesenchymal cells comprising the upper jaw anlagen pattern the developing skeleton (Depew et al., 2002). Expression of Dlx genes is regulated by signals from the epithelium covering the pharyngeal arches, and homeotic transformations occur in the jaws of animals with mutations in various Dlx genes (reviewed by Depew et al., 2005). Finally,signals from the pharyngeal endoderm appear necessary for development of the cartilages that comprise the ventromedial part of the upper jaw, but the nature and timing of these interactions are unknown(Benouaiche et al., 2008). At later stages, the ectoderm controls proximodistal extension and dorsoventral polarity of the upper jaw, but does not contain information that specifies upper or lower jaw structures. For example, when the FEZ was transplanted to the mandible, the lower jaw was duplicated rather than transformed into an FNP-like structure (Hu et al.,2003). Hence, a series of interactions occur among tissues that comprise the facial primordia. In our work, we have focused on identifying the role of the brain in these signaling interactions.

Signaling by SHH controls dorsoventral patterning of the forebrain(reviewed by (Bertrand and Dahmane,2006). Therefore, we examined dorsoventral polarity in treated and control embryos. We observed expansion of Shh and Nkx2.1 and a repression of Pax6 expression. These changes are consistent with other reports and indicate a shift in dorsoventral patterning of the forebrain(Goodrich et al., 1997; Manuel and Price, 2005).

The regulatory mechanisms controlling Shh expression in the neural tube are complex and involve multiple enhancers(Epstein et al., 2000; Epstein et al., 1999; Jeong et al., 2006; Jeong and Epstein, 2003) and signals (reviewed by Takahashi and Liu,2006). However, published data indicate that a cis-regulatory element controlling Shh expression in the telencephalon contains an Nkx2.1-binding site (Jeong et al., 2006), and Shh expression is absent from the telencephalon in the Nkx2.1^-/- mouse(Sussel et al., 1999). In the telencephalon, Nkx2.1 expression appears to be regulated by a combination of signals, including FGF8 and SHH. The allelic series of Fgf8 mutant mice (Meyers et al.,1998) have a progressive loss of Nkx2.1 and Shhexpression in the telencephalon that corresponds to decreasing gene dose(Storm et al., 2006), and our unpublished observations using SU5402 to block FGF signaling in the forebrain produce identical results (data not shown). However, we also observed similar changes in the telencephalon after blocking SHH signaling within the brain(Marcucio et al., 2005). Collectively, these data indicate that FGF8 and SHH signaling regulate expression of Shh via NKX2.1 in the telencephalon. In our current experiments, increased SHH signaling reorganized expression of Nkx2.1and Shh and shifted the anterior boundary of the ventral telencephalon.

Our objective was to examine the effect of the brain on development of the face. We used a loss-of-function approach to demonstrate the requirement for SHH signaling within the brain for the onset of FEZ function(Marcucio et al., 2005). Now,we complemented our previous research by performing a gain-of-function experiment. Investigators have reported that activation of SHH in the brain alters the dorsoventral polarity of the forebrain, inhibits neural crest generation and produces defects resembling holoprosencephaly (HPE)(Nasrallah and Golden, 2001). Therefore, we initiated our experiments after neural crest cells left the neural tube. We determined that activation of SHH signaling in the brain altered expression of signaling molecules in the ectoderm covering the upper jaw, and these changes created defects in the face.

In this work, we observed changes in skeletal elements that comprise the proximal part of the upper jaw and the nasal capsule. These elements are derived from lateral nasal and MXPs (Lee et al., 2004), and the malformations hinted at the presence of a signaling center located proximally in the jaw. Signals from the nasal pit,and in particular FGFs, have been shown to control development of proximal regions of the upper beak (Szabo-Rogers et al., 2008). We discovered that a number of genes were misexpressed in this region of the face after activating SHH signaling in the brain. We focused on the changes that we observed in the BMP, WNT and FGF signaling pathways owing to their importance for facial development(Ashique et al., 2002; Brugmann et al., 2007; Foppiano et al., 2007; Lan et al., 2006; MacDonald et al., 2004; Richman et al., 1997; Song et al., 2004; Storm et al., 2006; Szabo-Rogers et al.,2008).

We examined several candidate genes in and around the nasal pit to determine the extent to which the brain regulates expression of genes in this area. We observed upregulation of Bmp4 in the ectoderm but not mesenchyme near the nasal pit. A previous report demonstrated that increased expression of Bmp4 in the mesenchyme, but not in the ectoderm, causes expansion of the upper beak (Abzhanov et al., 2004). Hence, the reduction in the upper beak in conjunction with expanded Bmp4 expression in the ectoderm is consistent with this earlier report. In addition to the expansion of Bmp4, we observed other misregulated genes in treated embryos.

One of the genes that was misexpressed in treated embryos was the transcriptional co-factor Zfhx1b. This molecule was nearly undetectable in the FNP of treated embryos. Interestingly, Zfhx1b is thought to act primarily as a transcriptional repressor of Tgfβ/Bmp signaling (Postigo, 2003; Postigo et al., 2003), and coordinated Bmp signaling is required for development of this region of the face (e.g. Wu et al., 2006). Mutations in Zfhx1b cause a variety of developmental disorders,including Hirschsprungs disease and Mowat-Wilson Syndrome in humans(Mowat et al., 2003; Yamada et al., 2001; Zweier et al., 2002). Mice that lack exon 7 of Zfhx1b have deficiencies in neural crest generation and migration that may explain the malformations observed in humans. However, the role of this molecule during later stages of development is unknown.

The TOP-Gal plasmid has been used to identify sites of WNT signal activation in response to stabilized to β-catenin and TCF/Lef(DasGupta and Fuchs, 1999),and we used this plasmid to visualize Wnt activity in treated and control embryos. Our results revealed strong activation of the Wnt pathway in all embryos. These results lead us to conclude that activation of the SHH pathway in the forebrain did not impair canonical Wnt signaling in the developing upper jaw. The Wnt ligands that were downregulated, Wnt4 and Wnt6, are members of the non-canonical family of Wnt ligands that function through the planar cell polarity pathway rather than theβ-catenin pathway (Chang et al.,2007; Schmidt et al.,2007). Therefore, effects of changes in these molecules would not be detected by the TOP-Gal assay.

Additionally, we detected modest expansion of Raldh2 in the mesenchyme near the nasal pit. Mutations that destabilize and reduce protein levels of retinol dehydrogenase 10 create defects in the nasal region of mouse embryos, including clefting and malformations of the nasal septum(Sandell et al., 2007). Interestingly, increased signaling by retinoic acid has been show to downregulate Fgf8 expression in the epithelium of the first pharyngeal arch (Vieux-Rochas et al.,2007), and we observed a similar downregulation of Fgf8. Whether the changes in Raldh2 and Fgf8 are related remains to be determined.

The downregulation of Fgf8 in the nasal pit was one of the most notable changes in gene expression that we observed after ventralizing the brain. In our final experiment, we blocked the FGF pathway by inhibiting the ability of FGF ligands to activate Fgf receptors, and we observed a phenotype that was similar to that observed when the SHH pathway was activated in the brain. Hence, we conclude that alterations in FGF signaling from regions near the nasal pit appear to contribute to the morphological defects in the upper jaw. However, the extent to which the changes can be attributed to the absence of signaling by a specific ligand, such as FGF8, is unknown. Mice that express reduced levels of FGF8 have median clefts of the face that resemble the most severe cases that we observed (see Fig. S4 in the supplementary material), but a variety of FGF ligands are expressed throughout the face(Francis-West et al., 1998)and may have changed in our treated embryos. Although we are not able to attribute our results to the loss of signaling by specific FGF ligands, we(Hu et al., 2003), and others(Song et al., 2004; Szabo-Rogers et al., 2008; Wu et al., 2006), have demonstrated a requirement for FGF signaling throughout development of the FNP. The FGF signaling pathway establishes proliferative zones in the mesenchyme by positioning the expression domains of Bmp4 in the FNP(Szabo-Rogers et al., 2008; Wu et al., 2006), and these authors suggest that this may contribute to species-specific widening of the upper beak in birds.

In addition to the malformations evident in the proximal region of the upper jaw, we also observed changes in the FNP of treated embryos. These changes reflected a reorganization of gene expression patterns in the FEZ that altered growth of the FNP. As mentioned above, the establishment of species-specific growth zones has been studied recently. Signaling by BMPs is thought to underlie phenotypic variation in the developing upper jaw in a variety of birds (Abzhanov et al.,2004; Wu et al.,2006; Wu et al.,2004). We determined that the FEZ positively regulates expression of Bmps in the neural crest mesenchyme of the upper jaw(Hu and Marcucio, 2008). In this research, we observed an expansion of Bmp4 expression in lateral ectoderm of the FNP. When taken together, these studies indicate that signals from the FEZ induce or maintain expression of Bmps in the mesenchyme whereas antagonistic activity of FGF signaling restricts the expression domain of Bmp4 in the ectoderm (Hu and Marcucio, 2008) (Szabo-Rogers et al., 2008; Wu et al.,2006). Thus, an intricate signaling network establishes regionalized areas of gene expression and growth in the facial primordia.

Given the role of the FEZ in regulating expression of patterning genes in the FNP, we predicted that changes in gene expression patterns in the FEZ would alter growth of the FNP. After activating SHH signaling within the brain, chick embryos exhibited morphological alterations. The eyes were smaller and the FNP was divided into right and left `median nasal processes'. Similarly, mice have small eyes and they exhibit well-defined median nasal processes. In blind cavefish, increased SHH signaling is responsible for atrophy of the optic vesicles (Yamamoto et al., 2004), suggesting that the reduced eye size we observed was due to atrophy rather than to a transformation to a mammalian morphology. The reduced size of the eyes in treated chicks was not likely to alter the growth zones within the FNP. We have observed similar changes in the eyes during previous experiments, but changes in the FNP did not resemble those observed here (Marcucio et al., 2005). Furthermore, the morphological transformation of the FNP into left and right median nasal processes was directly related to the pattern of Shhexpression in the FEZ. In chicks, a single domain of Shh spans the mediolateral axis of the FNP, whereas in mice two lateral domains of Shh are apparent (Hu and Marcucio, 2008). After activation of SHH signaling within the forebrain, Shh expression domains in the FEZ resembled the murine pattern. These similarities, together with our observations on the role of the FEZ in regulating Bmp expression in the mesenchyme, lead us to conclude that spatial organization of the FEZ establishes growth zones that produce divergent morphologies distinguishing the early stages of development of the upper jaws of avian and mammalian embryos.

A recent report suggested that differential signaling by the canonical WNT pathway distinguishes the unique growth characteristics observed in mammals and birds. WNT responsiveness was reported to be exclusively in the midline of the developing avian FNP, whereas, in mice, WNT signaling occurred in lateral regions (Brugmann et al.,2007). However, our current work demonstrate that WNT signaling is active throughout the avian FNP, lateral nasal processes and MXPs. These results are reinforced by expression of Wnt ligands in ectoderm covering upper jaw in birds. The reason for the discrepancies observed in these studies is unknown but could reflect differences resulting from the mode of delivery of the WNT-reporter constructs. We used electroporation to deliver the TOP-Gal or control plasmids. We achieved widespread transfection and observed a large area of activation of the reporter construct. Hence, our data suggest that WNT signaling participates in regulating growth throughout the developing upper jaw anlagen in both chicks and mice.

Together, these data indicate that a highly orchestrated set of tissue interactions converge to regulate morphogenesis of the upper jaw. Signals from the brain establish gene expression patterns within the ectoderm. The ectoderm then signals to adjacent mesenchymal cells to establish areas of cell proliferation and create growth zones within the developing upper jaw. Interestingly, we determined that some genes, like Wnt9b, are regulated independently of forebrain signals, suggesting that intrinsic ectodermal properties or interactions between other tissues participate in patterning ectodermal signaling centers that control morphogenesis of the upper jaw. For example, Fgf8 expression is evident in the developing facial ectoderm prior to emigration of neural crest cells from the neural tube(Marcucio et al., 2005), and interactions between the neural crest cells and ectoderm have been shown to regulate expression of genes in the ectoderm(Eberhart et al., 2006; Marcucio et al., 2005; Schneider and Helms, 2003). Furthermore, signals from the pharyngeal endoderm appear to be required for development of FNP derivatives in chick embryos, but the exact nature of these interactions remain to be deduced(Benouaiche et al., 2008). By understanding the intrinsic molecular properties of the ectoderm, and the interactions that occur among the endoderm, forebrain, neural crest and ectoderm, we will define the signaling network that coordinates patterned growth of the upper jaw. This will allow us to determine the extent to which morphogenesis of these structures co-vary during development and will lead to an understanding of mechanisms that generate normal variation, disease phenotypes and evolutionary divergence in this region of the skull.