Unique Organization of the Frontonasal Ectodermal Zone in Birds and Mammals

Preparation of embryos and engraftment of the mFEZ

Fertilized chicken eggs (Gallus gallus, Petaluma Farms, Petaluma, CA) were prepared for surgical manipulations as follows. Embryos were incubated to Hamburger and Hamilton stage 10 (HH 10 (Hamburger and Hamilton, 1951)) and then a small hole was made in the shell directly over the embryo after removing 1.0ml of albumin. Embryos were returned to the incubator until HH 21 or 25. At this time the ectoderm covering the dorsal region of the FNP was removed with a sharpened tungsten needle. Grafts were prepared from mouse embryos at e10 as described (Hu et al., 2003). Briefly, mouse embryos were dissected from uteri of euthanized dams and placed in ice cold PBS. Facial tissues were dissected from embryos and were placed in dispase (2.5U/ml in PBS) for 20 mins. Then putative FEZ ectoderm was removed, and the graft was transferred to the host and positioned to replace the removed ectoderm. The graft was secured with glass pins (Supplemental Figure 1, and see (Hu et al., 2003)). Mouse embryos were collected at e9.5, e10, and e10.5 for in situ hybridization analysis. Animal procedures were approved by the UCSF IACUC.

Histology

Twenty-four hours after engraftment and at days 9, 12, and 13 days of development embryos were collected, fixed in 4% paraformaldehyde, stained with ethidium bromide, and then photographed using epifluorescent or brightfield illumination on a Leica MZFLIII microscope connected to a computer. After documentation, chimeras were dehydrated, embedded in paraffin, and sectioned (10μm). Sections were stained with Safranin-O/Fast Green to visualize cartilage (Lu et al., 2005), modified Milligan's Trichrome to visualize bone (Lu et al., 2005). Sections were imaged using a Leica DM5000B and Adobe Photoshop.

In situ hybridization

Patterns of gene expression in chimeras and normal mouse, chick, and duck embryos were analyzed on tissue sections and/or in whole mount via in situ hybridization using radiolabeled or digoxigenin-labeled riboprobes as previously described (Lu et al., 2005). Subclones of mouse B2 SINE (Bollag et al., 1999), Shh, Fgf8, Bmp-2, Bmp4, and Bmp-7, and chick Bmp2, Bmp4, Bmp7, Shh, Fgf8, Msx1, and Msx2 were linearized to transcribe riboprobes. Images of in situ hybridization assays performed on tissue sections are pseudo-colored superimpositions of the in situ hybridization signal and a blue nuclear stain (bis-benzimide; Sigma). Whole mount in situ hybridizations were photographed using a Leica MFLZIII dissecting microscope.

Ontogeny of the murine FEZ

We initially identified the FEZ in chick embryos based on the presence of a boundary between Shh and Fgf8-expressing cells in the ectoderm covering the FNP at HH 20 (Hu et al., 2003). To compare the relationship between Shh and Fgf8 expression domains in the FEZ in birds and mammals, we performed whole mount in situ hybridization. In mice and chicks, Fgf8 transcripts were detected in the ectoderm that spans the medio-lateral axis of the developing middle and upper face (Fig. 1A,B). However, expression of Shh in these animals is unique. In chick embryos Shh and Fgf8 expressing cells form a boundary in the ectoderm covering the neural crest cells that comprise the avian FNP at Hamburger and Hamilton Stage 20 ((HH20) (Hamburger and Hamilton, 1951)), and Shh expression is continuous across the medio-lateral axis of the FNP (Fig. 1C,E (Hu et al., 2003)). This same pattern is also observed in duck embryos at (Supplemental Figure 2A). In contrast to the avian FEZ, the murine FEZ is not a single signaling center. Rather Shh is expressed in domains on the right and left side of the mouse face creating left and right FEZs (Fig. 1D,F). In these regions neural crest cells are present, but these cells are absent or greatly reduced at the midline. A bilateral pattern of Shh expression is also observed in human embryos in this region and is accompanied by reduced mesenchymal cells in this region (Odent et al., 1999) and the presence of median nasal processes rather than a Frontonasal Process. Thus, these unique patterns of Shh expression in the FEZ correlate with the morphology that distinguishes avian and mammalian embryos at these early times.

Our next step in characterizing the murine FEZ was to examine the ontogeny of Shh and Fgf8 expression prior to and after outgrowth of the median nasal processes had begun. At e9.5 (n=3) Shh expression was detected in the forebrain epithelium but not in the surface ectoderm in either medial or lateral domains (Fig. 2A,B and see supplemental Figure 2). In contrast, Fgf8 transcripts were detected in the forebrain, and in medial and lateral domains of the surface ectoderm (Fig. 2G,H). By e10.0 (n=6) Shh transcripts were not present in the medial region of the face (Fig. 2C, Supplemental Figure 2), and concomitantly there were no neural crest cells in this region of the face. In contrast, Shh transcripts were present in the ectoderm adjacent to neural crest cells that formed the median nasal processes (Fig. 2D), and this domain forms a boundary with cells expressing Fgf8 (Fig. 2I,J). The boundary between Shh and Fgf8-expressing cells persists in the left and right median nasal processes at e10.5 (Fig. 2F,L,K, n=10), but no Shh transcripts were detected in ectodermal cells located between the median nasal processes (Fig. 2E). Again, no neural crest cells were present in the midline of the mouse face.

In addition to Shh and Fg8, Bmps are also expressed in the FEZ of birds and mice. In chick embryos at HH 22 (n=10) Bmp-2 (Fig. 3A), Bmp-4 (Fig. 3B), and Bmp-7 (Fig. 3C) transcripts were present in the FEZ. Again, the expression of these genes spanned the entire medio-lateral axis of the FNP. In mice, Bmp-2 expression was not evident in the FEZ during the times we examined (e9-5,n=3; e10.0, n=6; e10.5,n=10, Fig. 3D,G and data not shown). At e10.5 Bmp-4 transcripts were detected in the lateral areas of the FEZ (Fig. 3H), but not in medial regions (Fig. 3E). Bmp-7 transcripts were present throughout the FEZ (Fig. 3F,I). Collectively, the results of this gene expression analysis suggest a FEZ is present in ectoderm covering the developing mouse face, but the spatial arrangement and some of the molecular constituents of the FEZ are different in mice and birds (Fig. 1G).

FEZ function is conserved across species

Based on our gene expression analysis, the FEZ appears to be present in mouse embryos. We assessed the functional capacity of the murine FEZ by replacing the chick FEZ at HH 21 with the equivalent ectoderm derived from e10.0 mouse embryos, or as a control, with flank ectoderm from the mouse. Transplantation of flank ectoderm to the avian FEZ did not produce morphologic alterations (Fig. 4D, n=5) which agrees with our previous observations (Hu et al., 2003). However, when we grafted the mouse FEZ onto the developing FNP of chicks, we observed morphologic alterations. The grafted mouse FEZ altered the morphology of the avian FNP even thought the grafted tissue was very small and occupied up to about one-third of the FNP (Supplemental Fig.1). In these mouse-chick chimeras we observed that the distal portion of the upper jaw was duplicated at day 9 (Fig. 4B). There were two egg teeth present on the upper beak, and there was a cleft between the two tips (n=5). We performed whole mount in situ hybridization to detect the mouse-specific repetitive DNA element B2SINE. As expected, we observed that one of the tips was associated with the transplanted mouse ectoderm while the other was associated with chick ectoderm (Fig. 4C). Thus, by transplanting the mouse FEZ onto the FNP of developing chick embryos, the avian FEZ was altered. In these chimeras, there were two FEZ's located side-by-side, and we observed a duplication of the upper beak.

In order to examine changes in the mesenchyme in response to the grafted ectoderm, we transplanted the putative FEZ ectoderm from mouse embryos at e10.0 to the dorsal surface of the FNP of HH 25 chick embryos (Hu et al., 2003). Using this approach we could position the small graft adjacent to responsive mesenchyme located at a distance from the autochthonous FEZ, and then we could observe changes exclusively due to the presence of the murine ectoderm. We examined mouse-chick chimeras at various times after engraftment. At day 9 (n=3, Fig. 5A-C), 12 (n= 3, Fig. 5D-F), and 13 (n=1, Fig. 5G-I) ectopic growths were evident on the upper beaks (Fig. 5A,D,G). These outgrowths were comprised of skeletal elements that were integrated with the host skeleton, and the bone and cartilage developed with the same dorso-ventral polarity as in the autochthonous beak (Fig. 5B,H). Furthermore, an egg tooth is normally formed in ectoderm covering the dorsal surface of the distal tip of the upper beak, and an egg tooth was also observed on the dorsal surface of each duplicated beak (Fig. 5E,H). Again, we used in situ hybridization to detect the mouse-specific repetitive DNA element B2SINE. The ectopic egg tooth was formed exclusively by host ectoderm (Fig. 5F); in this embryo, the mouse ectoderm was not apparent in sections containing the egg tooth. Nonetheless, these results indicate that in addition to patterning the skeleton, the FEZ induces differentiation of the egg tooth in birds. Importantly, we never observed hair, or hair follicles, in mouse-chick chimeras supporting the idea that specification of ectodermal appendages arises from signals provided by the mesenchyme (Schneider, 2005).

Response of neural crest mesenchyme to the FEZ

Our results demonstrate that mouse ectoderm covering the median nasal processes contains patterning information that regulates morphogenesis of the upper jaw. Next, we examined mechanisms responsible for patterning. Twenty-four hours after engraftment (n=5/5) Shh (Fig. 6B) and Fgf8 (Fig. 6C) transcripts were still present in the donor ectoderm (Fig. 6A). Additionally, using mouse-specific in situ hybridization probes we observed mBmp-4 (Fig. 6E), and mBmp-7 (Fig. 6F) expression in the transplanted ectoderm, but mBmp-2 expression was not induced in the grafted ectoderm (Fig.6D).

In our previous work, we demonstrated that Msx-1 was up-regulated in response to grafted avian ectoderm (Hu et al., 2003). We assessed whether the chick mesenchyme had a similar response to the grafted mouse ectoderm. Indeed, we observed ectopic expression of Msx-1 and Msx-2 in mesenchyme adjacent to the grafted mouse ectoderm (data not shown). We also examined expression of Bmps in mesenchyme beneath the grafted tissue, because these molecules regulate morphogenesis of the upper jaw (Abzhanov et al., 2004; Wu et al., 2006; Wu et al., 2004). Normally, at this time (∼HH28) Bmp-2, Bmp-4, and Bmp-7 expression patterns become restricted to the mesenchyme in the tip of the growing beak (Fig. 7A-H). Bmp-2 (Fig. 7B) was expressed in the dorsal region, Bmp-7 (Fig. 7D) was expressed in the ventral region, and Bmp-4 (Fig. 7C) was expressed throughout the growing FNP. In mesenchyme located in the region of the grafted ectoderm, Bmp-2 (Fig. 7F), Bmp-4 (Fig. 7G), and Bmp-7 (Fig. 7H) transcripts were not detected. Adjacent to grafted ectoderm in chimeras we observed strong up-regulation of Bmp-2 (Fig. 7I) and Bmp-7 (Fig. 7L), and weak up-regulation of Bmp-4 at this time in each chimera (n=5/5), Fig. 7K and Supplemental Figure 3). These results indicate that signals from the FEZ regulate expression of patterning genes in neural crest cells that will form the skeleton of the upper jaw.

Conservation of function between mFEZ and cFEZ

Our experiments suggest that mechanisms mediating function of the FEZ are highly conserved between mice and chicks. For example, when transplanted ectopically the FEZ from the mouse and the chick induce duplications of the distal portion of the upper beak that exhibit dorso-ventral polarity. Similarly, Msx-1 is induced in the mesenchyme in response to either murine (not shown) or avian (Hu et al., 2003) ectoderm. Additionally, ectodermal responses to the FEZ are similar; transplanted mouse and chick (Hu et al., 2003) ectoderm induced egg tooth formation in host ectoderm. These observations suggest that molecular signals emanating from the FEZ are shared between the signaling centers in these two species, and that these signals evoke intrinsic responses from underlying mesenchymal cells. While we are unable to replace the mouse FEZ with the equivalent avian tissue or remove the mouse FEZ to directly test the necessity of this signaling center for development of the mouse jaw, the similarity of function between the chick and mouse FEZ in our chimeric assay suggests that the FEZ is an important signaling center that patterns the developing primordia of the upper jaw in mammals.

How does the FEZ form?

Neural crest cells and the brain participate in regulating the onset of Shh expression in the FEZ. When duck neural crest cells were replaced with crest from the more precocious quail, Shh expression in the duck host was accelerated (Schneider and Helms, 2003). Similarly, Shh expression in the FEZ was delayed by ablating presumptive neural crest cells. In these experiments the cut edge of the neural tube regenerated neural crest cells, and Shh expression began once the regenerated cells reached the FNP (Marcucio et al., 2005). In zebrafish Shh expression in the oral ectoderm requires the presence of neural crest cells (Eberhart et al., 2008). Based on these observations, the establishment of unique organization of the FEZ in mice and chicks may reflect the intrinsic differences that are apparent in the distribution of neural crest cells in this region of the face. In chicks, neural crest cells are present across the entire mediolateral axis of the FNP, but in mice, these cells are located in lateral regions adjacent to the areas of Shh expression. Additionally, SHH signaling within the forebrain is required for Shh expression in the FEZ (Marcucio et al., 2005; Eberhart et al., 2006), and unique signals from this tissue may also regulate spatial organization of the FEZ.

Variation in FEZ organization is associated with early facial form

Our findings suggest that the initial growth patterns that generate Median Nasal Processes in mammals and Frontonasal Processes in birds result from distinct organization of the FEZ between these classes of animals. In chicks and ducks the FEZ is a single unit that spans the FNP, but in mice and humans, the FEZ is divided into left and right halves. Accordingly, the FNP initially extends as a unified entity, but growth in the homologous region in mammalian embryos occurs in the left and right median nasal processes which merge in the midline as development proceeds. We did not observe an avian-to-mammalian transformation when we replaced the avian FEZ with a transplanted murine FEZ. Rather, we observed a duplication of the distal tip of the upper beak. We determined that the transplanted mouse ectoderm was associated with one of the duplicated structures, while chick ectoderm was associated with the other. The grafted mouse ectoderm was very small and spanned less that one third of the avian FNP. Hence, any spatial patterning information contained in the mouse ectoderm was likely diluted in the presence of the large amount of mesenchyme in the host. In fact, the presence of neural crest cells in the midline of the chick FNP and their absence in the homologous region of the mouse indicates that the distribution of cells that can respond to the FEZ are also unique between these animals, and these differences are also likely to contribute to the distinct morphologies that are observed. Alternatively, the transplanted mouse FEZ may have been reprogrammed by the chick mesenchyme. For instance, when we examined Shh expression in the transplanted tissue we never observed a gap in expression of Shh. This observation suggests that the avian neural crest cells may have induced Shh expression in the grafted FEZ. However, this must be interpreted cautiously since we were not able to examine the transplanted mouse ectoderm in its entirety, and we may have missed the Shh-negative region.

The extent that the FEZ regulates species-specific morphology is unknown. However, our data suggest that the FEZ participates in the dialogue that patterns the developing uppe jaw. The FEZ appears to evoke species-specific responses from the adjacent mesenchyme. This agrees with previous data indicating that neural crest cells autonomously control shape and size of skeletal elements in the head (Schneider and Helms, 2003; Tucker and Lumsden, 2004). Unique signals from the brain and other adjacent signaling centers as well as information contained in the neural crest cells themselves are likely to elaborate species-specific morphologies. For instance, regulation of Bmp-4 expression by Fibroblast growth factors produced in the nasal pit has been suggested to regulate the widening of the bill that occurs in duck embryos (Szabo-Rogers et al., 2008). Nonetheless, the mouse and chick FEZ appear to use similar molecular signals to establish different growth zones in the upper jaw, and these growth zones correspond to the morphological features that distinguish the earliest developmental features of the avian and mammalian face (Fig. 8).

In addition to tissues that regulate morphogenesis of the upper jaw, the role of specific signaling molecules have been described. For instance, species-specific patterning of the upper jaw in birds has been shown to rely partly on signaling by BMP-2 and BMP-4 (Abzhanov et al., 2004). Further, BMP-4 has been shown to establish dual growth zones in the mesenchyme of the developing upper jaw mesenchyme of duck embryos (Wu et al., 2006; Wu et al., 2004). Our data indicate that the FEZ participates in regulating expression of Bmp-2, Bmp-4, and Bmp-7 in neural crest mesenchyme. Previous work has demonstrated that combinatorial signaling by FGF8 and SHH induces Bmp-4 expression in neural crest mesenchyme at late stages of avian development. In these embryos, ectopic cartilaginous outgrowths were observed (Abzhanov and Tabin, 2004). These observations support our conclusion that signals from the FEZ act upstream of molecules that regulate morphogenesis of the skeleton of the upper jaw.

Regulation of variation by the FEZ may have important clinical implications as well, because extremes in morphologic variation could create birth defects. Disruptions to SHH signaling (Ahlgren and Bronner-Fraser, 1999; Chiang et al., 1996; Jeong et al., 2004) or the FEZ (Cordero et al., 2004) produce catastrophic structural defects in the face. For example, altered SHH signaling creates defects in the middle of the upper jaw and face in individuals with Holoprosencephaly (HPE; reviewed in: (Dubourg et al., 2007)). A hallmark of HPE is variable presentation of the disease phenotype (McKusick, 2000; Ming and Muenke, 2002). At extremes, fetuses exhibit severe forebrain defects and cyclopia characterized by a single median eye located beneath a proboscis. Microforms of HPE include midfacial hypoplasia, a single central incisor, and hypotelorism, while some individuals with mutations in HPE genes are clinically unaffected. However, there is no clear phenotype-genotype correlation that adequately explains the spectrum of phenotypes in patients with HPE (Marini et al., 2003; Ming and Muenke, 2002; Nanni et al., 1999; Roessler et al., 1996) indicating the multifactorial etiology of the disease. These observations inspired studies to identify multiple mutations in patients with HPE and correlate these mutations with severity of phenotype (Ming and Muenke, 2002). Our results suggest that genes involved in regulating formation or function of the FEZ may be candidates for “second hit genes.” These molecules may participate in generating facial defects associated with HPE by contributing to variation in morphogenesis of the upper jaw and mid-face. Therefore, understanding how the FEZ forms, and in particular how the spatial domain of Shh expression is regulated, could provide significant insight into mechanisms of facial patterning and disease phenotypes.

In conclusion, results of our work illustrate a complex set of tissue interactions among forebrain, neural crest, and stomodeal ectoderm that regulate the earliest morphogenetic events during development of the upper jaw. Previously we showed that the brain is an important signaling center that participates in regulating facial development, in part by controlling the onset of FEZ activity (Marcucio et al., 2005). In addition, neural crest cells (Marcucio et al., 2005; Schneider and Helms, 2003) and Bmps (Foppiano et al., 2007) participate in regulating Shh expression in the FEZ. The FEZ then regulates expression of Bmps in neural crest mesenchyme and establishes growth patterns in the upper jaw anlagen. These mechanisms appear conserved between mice and birds, but our results illustrate that differences in the earliest patterns of growth within the upper jaw may be generated through a relatively simple mechanism. The spatial organization pattern of genes that are expressed in the FEZ correspond to either median or frontonasal processes. Signals from the brain or migration of neural crest patterns may be different between mammals and birds and could establish unique organization of the FEZ, but further work is required to identify the regulatory differences that control these unique patterns. Nonetheless, these results reconcile how similar molecular signals may generate morphological diversity in the face. These molecular differences are apparent during the earliest stages of facial development, and they presage outgrowth of facial primordia. Thus, distinct patterns of facial morphogenesis in birds and mammals is apparent before there is gross evidence of divergent facial morphology. In fact, formation of the distinct FEZ in mammals and birds is currently the earliest known feature of the developing upper jaw that distinguishes these organisms. Future exploration of the interactions that regulate formation and organization of the FEZ is likely to yield evidence of even earlier differences and will allow us to identify avenues by which variation in facial morphology can be influenced.