Abstract

Endogenous retinoids are important for patterning many aspects of the embryo including the branchial arches and frontonasal region of the embryonic face. The nasal placodes express retinaldehyde dehydrogenase-3 (RALDH3) and thus retinoids from the placode are a potential patterning influence on the developing face. We have carried out experiments that have used Citral, a RALDH antagonist, to address the function of retinoid signaling from the nasal pit in a whole embryo model. When Citral-soaked beads were implanted into the nasal pit of stage 20 chicken embryos, the result was a specific loss of derivatives from the lateral nasal prominences. Providing exogenous retinoic acid residue development of the beak demonstrating that most Citral-induced defects were produced by the specific blocking of RA synthesis. The mechanism of Citral effects was a specific increase in programmed cell death on the lateral (lateral nasal prominence) but not the medial side (frontonasal mass) of the nasal pit. Gene expression studies were focused on the Bone Morphogenetic Protein (BMP) pathway, which has a well-established role in programmed cell death. Unexpectedly, blocking RA synthesis decreased rather than increased Msx1, Msx2, and Bmp4 expression. We also examined cell survival genes, the most relevant of which was Fgf8, which is expressed around the nasal pit and in the frontonasal mass. We found that Fgf8 was not initially expressed along the lateral side of the nasal pit at the start of our experiments, whereas it was expressed on the medial side. Citral prevented upregulation of Fgf8 along the lateral edge and this may have contributed to the specific increase in programmed cell death in the lateral nasal prominence. Consistent with this idea, exogenous FGF8 was able to prevent cell death, rescue most of the morphological defects and was able to prevent a decrease in retinoic acid receptorβ (Rarβ) expression caused by Citral. Together, our results demonstrate that endogenous retinoids act upstream of FGF8 and the balance of these two factors is critical for regulating programmed cell death and morphogenesis in the face. In addition, our data suggest a novel role for endogenous retinoids from the nasal pit in controlling the precise downregulation of FGF in the center of the frontonasal mass observed during normal vertebrate development.

Keywords: Chicken embryo; Retinoid synthesis; Retinaldehyde dehydrogenase; FGF8; Rarβ; Lateral nasal prominence; Face; Nasal pit; Citral

Article Outline

Introduction

Methods

Bead implants

Skeletal staining and analysis

Whole-mount in situ hybridization and analysis of gene expression patterns

Whole-mount Nile Blue Sulfate and TUNEL staining for programmed cell death

Results

Dose–response data

Early effects on nasal pit morphogenesis

Effects on skeletal development

Position-specific effects

Rescue of the Citral phenotype with RA

Temporal effects of Citral on upper beak morphogenesis

Inhibition of RA synthesis leads to increased programmed cell death

Endogenous RA is required to maintain expression of cell survival gene Fgf8 and genes in the BMP pathway

Fgf8 is sufficient to rescue most of the skeletal defects caused by Citral

FGF8 prevents a Citral-induced decrease in Rarβ expression

Discussion

Functions of retinoid signaling emanating from the nasal pit

Retinoids induce and maintain FGFs and thereby promote cell survival

Mechanism of the FGF8 rescue

A progressive model for patterning the frontonasal region with retinoids and FGFs

Acknowledgements

References

Introduction

Vitamin A derivatives or retinoids are critical for many aspects of embryo development as amply demonstrated in the offspring of vitamin A-deficient animals (Clagett-Dame and DeLuca, 2002, Ross et al., 2000 and Zile, 2001). Vitamin A is taken in through the maternal diet and locally metabolized to active metabolites in the tissues of the embryo. The enzymes required for conversion of retinol to retinoic acid include 4 members of the retinaldehyde dehydrogenase enzyme family, RALDH1-4 (ALDH1a1-4) (Duester et al., 2003 and Lin et al., 2003). Once metabolized in the cytosol, retinoic acid binds to three types of nuclear retinoic acid receptors (RARα, β, γ). The RARs may dimerize with RXR receptors leading to activation of downstream genes, however, unliganded retinoic acid receptors can repress transcription by recruitment of other cofactors (Weston et al., 2003). Targeted deletion of RALDH2 (Niederreither et al., 1999) followed by partial rescue through administration of retinoids permits development to advanced stages revealing a requirement for patterning the heart (Niederreither et al., 2001), posterior branchial arches (Niederreither et al., 2003), limbs (Niederreither et al., 2002), and brain (Niederreither et al., 2002), however, no effects on craniofacial development were described. This may be because several RALDH genes are expressed in the embryo and there may be some compensation by other family members. Blocking slightly higher up the pathway, recent studies have shown that retinal conversion from retinol has specific effects on the craniofacial skeleton (Lampert et al., 2003). Blocking the pathway at the lowest point by deleting nuclear retinoic acid receptors leads to defects in fusion of the primary palate (Lohnes et al., 1994). Thus, it is likely that retinoid signaling is required for some aspects of craniofacial morphogenesis; however, global knockout models are not precise enough to tell us where critical levels of retinoids are required.

Our previous work has shown that levels of retinoids are one signal required for establishing frontonasal mass identity (Lee et al., 2001). Increased retinoid acid levels in combination with blocking of BMP signaling lead to transformation of the side of the beak (derived from the maxillary prominence) into a second set of midline structures (derived from the frontonasal mass). These experiments suggested it is necessary to have a source of endogenous retinoids to pattern the frontonasal mass and its derivatives, the prenasal and premaxillary skeletal elements. Two of the retinoic acid synthesis enzymes are expressed at high levels in the nasal placodes, RALDH2 which is expressed in the mesenchyme under the placode (Berggren et al., 1999 and Bhasin et al., 2003) and RALDH3 which is expressed in the epithelium (Blentic et al., 2003 and Mic et al., 2000). Furthermore, expression of a reporter gene under the control of a retinoic acid response element (RARE) is very strong in the nasal placodes of transgenic mice (Niederreither et al., 2002). We hypothesized that RA synthesized in the placode and nasal pit could be used in patterning both the lateral nasal prominence and frontonasal mass.

The blocking of retinoid synthesis or signaling in chicken embryos during neural crest cell migration stages has shown that retinoids are required for forebrain development and the lack of brain growth secondarily affects upper beak formation (Schneider et al., 2001). In the present study, we work at later stages and specifically target the nasal pit with the chemical Citral, an inhibitor of retinaldehyde dehydrogenases (Gagnon et al., 2002 and Kikonyogo et al., 1999). We examined the effects of blocking retinoid synthesis on craniofacial development and conducted rescue experiments with RA to test the specificity of Citral on retinaldehyde dehydrogenase. We then looked at gene expression changes between stages 20 and 24 for several genes that are in the retinoid pathway and could provide a mechanism for the development of the skeletal phenotype. Finally, we tested whether one of the proteins downstream of endogenous retinoids-FGF8-could rescue development. Our results support the idea that endogenous retinoids from the nasal pit play dynamic and important roles in regulating Fgf8 expression and subsequent morphogenesis in the embryo.

Methods

Bead implants

SM2 beads (200 μm diameter, BioRad) were soaked in several concentrations of Citral (Fluka, stock concentration = 0.888 g/ml) for 10 min as described in Schneider et al. (2001). AG1X-2 beads (BioRad, format form, 200 μm diameter) were soaked in various concentrations of RA dissolved in DMSO. Both types of control beads were soaked in DMSO. Heparin Acrylic beads (Sigma) were soaked in 1 mg/ml FGF8b (human, Peprotech) for a minimum of 1 h at room temperature. White leghorn eggs were incubated at 38°C until stage 20. For Citral bead implantation, a small incision was made in the ectoderm of the right nasal pit or center of frontonasal mass, through which beads were pushed into the mesenchyme. FGF8b beads were also implanted in the mesenchyme. RA beads were placed in contact with the ectodermal lining of the right nasal pit. The day following implantation, the treated embryos were observed to confirm the presence of beads in the face.

Skeletal staining and analysis

Chicken embryos were fixed at stage 38 (after 12 days of incubation) and kept in 100% ethanol for 4 days. Next, embryos were immersed in 100% acetone for 4 days, rinsed in water and then stained for up to 10 days in alizarin red/alcian blue solution (Plant et al., 2000). Specimens were cleared in 20% glycerol/2% KOH solution followed by a series of glycerol/H₂O solutions with increasing amount of glycerol (50%, 100%).

Morphology of each skeletal element was assessed and described as normal, absent, or defective. The category defective includes all morphology changes such as overall reduction in size and loss of processes.

Whole-mount in situ hybridization and analysis of gene expression patterns

Whole-mount in situ was performed as described in Shen et al. (1997). Due to the dynamic expression patterns of many of the genes, we compared expression of 3 h, 6, 16, and 24 h embryos to non-treated embryos of a similar stage. In this way, we could determine whether there were possibly bilateral effects of the Citral. Hybridizations and washes were carried out in an Intavis In situ Hybridization Robot (InTavis, CA). Individuals who generously provided cDNAs for this work include: Bmp4 (P. Francis-West), Fgf8 (J.C. Izpisua-Belmonte), Msx1, and Msx2 (S. Wedden), Rarβ (P. Brickell).

Whole-mount Nile Blue Sulfate and TUNEL staining for programmed cell death

Embryos were treated with Citral for defined time periods of 1, 3, 6, 16, and 24 h and removed from the egg. Embryos were transferred to a plastic tube containing 10 ml PBS with 30 μl of Nile Blue Sulphate (1 mg/ml filtered Nile Blue Sulphate in ddH₂O (Jeffs et al., 1992). Embryos were rocked gently for 30–40 min at room temperature, washed in PBS for 30–60 min, and then photographed immediately.

Terminal uridine nick-end labeling (TUNEL) technique was used to detect cell death in tissue sections as described (Shen et al., 1997). Embryos were fixed at 3, 6, and 16 h post-treatment with beads soaked in 0.01 or 0.1 g/ml Citral. Control embryos were fixed at the same time points. Two sections from each treated and control embryo were placed on the same slide and 4–5 slides were examined. Controls for the TUNEL reaction consisted of sections treated with DNAse I before TUNEL detection and sections in which no enzyme was added. Sections were photographed under bright field. Approximate numbers of dead cells were recorded for different regions of the embryonic face on treated and untreated sides. Each area was scored as having 0, 1–5, 5–10, 10–25, 25–50 apoptotic cells. At least four different embryos were scored for each time point.

Results

Dose–response data

Early effects on nasal pit morphogenesis

The nasal placode is first observed as a thickening of the ectoderm at either side of the telencephalon at stage 15. Invagination of the placode occurs to form a shallow pit at stage 20, a nasal slit by stage 24, and finally the nasal orifice at the sides of the beak by stage 32. The nasal pit lies in between the lateral nasal prominence and frontonasal mass and could potentially provide signals to both parts of the face. The frontonasal mass makes the prenasal cartilage, premaxilla, and nasal septum (Lee et al., 2001, Richman and Tickle, 1989, Richman and Tickle, 1992 and Wedden, 1987); the lateral nasal prominences give rise to the nasal conchae (MacDonald et al., 2004) and nasal bone. To test the role of retinoids as possible patterning signals, we applied beads soaked in Citral to the nasal pit of stage 20 chicken embryos, a time when there is very strong expression of Raldh3 in the nasal pit (Blentic et al., 2003). Different soaking concentrations were tested, ranging from 0.001 to 0.888 g/ml (stock concentration) since these were expected to release different amounts of the compound. Embryos treated with beads soaked in 0.01 g/ml had normal invagination of the nasal pit and normal lateral nasal prominence morphology at all time points examined (n = 20 for 1, 6, 16, and 24 h). Embryos treated with 0.1 g/ml of Citral showed loss of nasal pit morphology visible shortly after beads were implanted (n = 9, 1 h); however, the lateral nasal prominence was normal in shape until 16 h (n = 5/5, Figs. 3I, J). Undiluted Citral or 0.5 g/ml led to complete loss of the lateral nasal prominence and nasal pit on the treated side almost immediately after placing the bead (n = 9, 1 h). The main effects appeared to be on the lateral nasal prominence rather than the adjacent frontonasal mass and maxillary prominences.

Effects on skeletal development

The effects of Citral beads on skeletal morphogenesis were consistent with the early changes in morphology of the nasal pit. Beads soaked in undiluted Citral or a concentration of 0.5 g/ml ablated the entire right side of the upper beak including all lateral nasal prominence derivatives and most maxillary prominence derivatives (n = 13, data not shown). As the concentration was decreased the effects became more localized. Consistent morphological defects were observed with 0.1 g/ml of Citral (Table 1A, Figs. 1C, D). The major defects were seen in the derivatives of the lateral nasal prominence, consistent with the early phenotype. The nasal bone and nasal conchae were absent in the majority of specimens (Table 1A). There were also significant abnormalities in the maxillary prominence derivatives including the maxillary, palatine, and jugal bones (Table 1A, Figs. 1C, D). The only frontonasal mass derivative affected was the maxillary process of the premaxillary bone and frequently this process was truncated on both sides of the beak (Fig. 1D). At concentrations of 0.01 g/ml, a proportion of embryos developed normally (n = 11/28) and others had similar defects to those treated with 0.1 g/ml. Therefore, with this soaking concentration, it was not possible to predict which embryos would recover from the Citral and go on to develop normally. Embryos treated with 0.001 g/ml of Citral showed normal morphology of the skull at stage 38 (n = 7/7). DMSO-treated embryos always developed normally and therefore cutting into the nasal placode had no effect on development (Figs. 1A, B). We also treated a subset of embryos with Citral-soaked beads resting on the epithelium of the nasal placode and these had identical effects on the skeleton (n = 13/13). Therefore, Citral was capable of passing freely through the epithelium.

Position-specific effects

We hypothesized that the nasal placode is the main source of retinoids in the frontonasal region and that if we placed beads further from this source, the effects on morphogenesis would be different. A Citral-soaked bead in the center of the frontonasal mass led to decreased upper beak length and deviations to the treated side, however, the morphology of skeletal structures was mainly normal (n = 12/12, Figs. 1E, F). Since the effects of blocking retinoid synthesis were less pronounced than when the nasal placode was targeted, it suggests that the frontonasal mass is less dependent on RA synthesis at stage 20. Similar normal morphology was obtained by blocking RA signaling at the receptor level in stage 18 embryos (Schneider et al., 2001).

Rescue of the Citral phenotype with RA

Citral may affect other aldehyde dehydrogenases besides those involved in RA synthesis. To determine how much of the phenotype was due to inhibition of retinoid synthesis, we applied exogenous RA sequentially after Citral beads were implanted. While we and others had published the effects of RA on outgrowth of the upper beak (Richman and Delgado, 1995, Tamarin et al., 1984 and Wedden and Tickle, 1986), the effects on the skeleton had not previously been analyzed. Treatment with 0.75 mg/ml exogenous all trans-RA at a similar time point (stage 20 plus 1–2 h) always resulted in the loss of the prenasal cartilage and premaxillary bones (15/15; Figs. 2A, B). The nasal conchae were always present in embryos treated with only RA, though reduced in size and abnormal in shape (15/15, Figs. 2A, B). We found that the majority of dual-treated embryos (Citral 0.1 g/ml + RA 0.75 mg/ml) were almost normal (Figs. 2C, D) when compared to DMSO-treated controls (Figs. 1A, B). The skeletal elements rescued most frequently were the nasal conchae (100%) and nasal bone (84%; Table 1B; Figs. 2C, D). Chi² analysis showed that there was a highly significant increase in the presence of the nasal conchae and bones as compared to Citral-treated embryos (P < 0.0001; Table 1A; Figs. 1C, D). In addition, fewer maxillary derivatives (palatine, jugal and maxillary bone) were defective in retinoic acid rescued-embryos as compared to Citral-treated control embryos (Table 1B, Figs. 2C, D). There was only one embryo out of 19 that had the typical RA phenotype even though Citral was also present (Table 1B). Thus, we did not see the effects of exogenous RA overwhelm the effects of Citral. Rather, the RA rescued all aspects of the Citral phenotype, suggesting that it is mainly through inhibition of RA synthesis that the defects arise.

Temporal effects of Citral on upper beak morphogenesis

To see whether effects of Citral were reversible, we first treated embryos with Citral for a defined period and then removed the beads. Even beads removed after only 3 h lead to defects in the upper beak similar to those produced by the beads being left in place. All embryos were missing the right nasal conchae (n = 8/8) and most had abnormally shaped maxillary and palatine bone (n = 7/8). These results suggest that most of the effects of Citral occur very rapidly or that sufficient amounts of Citral remain in the tissue to sustain the inhibition of RA synthesis. We therefore did a time course of rescue experiments with RA to determine whether there was a point at which RA could no longer reverse the effects of Citral. RA Beads implanted after an initial Citral exposure of 6 h rescued the nasal conchae and nasal bone (n = 9/9). As time between the Citral and RA implants increased, the rescue was not as effective. A 16-h treatment with Citral could not be rescued by RA (10/11 missing right nasal conchae and nasal bone). This suggests that irreversible patterning changes have already taken place by this time (Figs. 2E, F). A similar time frame has been reported for irreversible changes to be induced by exogenous RA (Tickle et al., 1985 and Wedden and Tickle, 1986).

Inhibition of RA synthesis leads to increased programmed cell death

We suspected that the early changes in nasal pit morphology and the loss of lateral nasal prominence derivatives may be linked to a specific induction of cell death in these regions. Two possibilities needed to be investigated, one was that Citral-induced necrosis or that loss of RA lead to an increase in programmed cell death. If necrosis was the mechanism, then dead or dying cells would be equally distributed around the bead and these changes would be clearly visible very soon after bead placement. However, if apoptosis was occurring then there would be a delay in the response while a specific downstream molecular cascade was induced. In addition, there may not be equal responses of all tissues to the Citral depending on the importance of elevated RA levels in that area of the embryo. We determined that embryos treated with 0.1 g/ml Citral-treated embryos showed no detectable increase in programmed cell death at 3 h either with Nile Blue Sulfate staining (n = 10/10, Fig. 3B) or in TUNEL-stained sections (n = 4/6, Figs. 3C, D) in the lateral nasal prominence, frontonasal mass or maxillary prominence. By 6 h, there was detectable cell death in the lateral nasal prominence visible in whole embryos (n = 2/7, Fig. 3F) and mesenchymal apoptosis was obvious in tissue sections, primarily restricted to the lateral nasal prominence (10–25 cells per section, n = 6/6 Figs. 3G, H) and anterior maxillary prominence (5–10 cells per section, n = 4/6). At 16 h, there were similar numbers of TUNEL positive cells in the lateral nasal prominence mesenchyme (10–25 cells per section, n = 5/5, Figs. 3K, L). It was also clear that apoptosis was not induced in the adjacent frontonasal mass tissue, even though the bead was also contacting this tissue (n = 11/11 of TUNEL-stained specimens; Figs. 3G, K). In addition to induced cell death in the lateral nasal prominence, we observed normal patterns of apoptosis in the trigeminal ganglion, in the nasolacrimal groove, the midline of the mandibular prominence and second branchial arch (Figs. 3A, B, E, F, I, J). Thus, there is a specific response in the lateral nasal prominence mesenchyme that sets it apart from other facial mesenchyme and suggests that the effects of Citral are not merely induction of cellular necrosis. In addition, apoptosis was induced in tissue that will give rise to the frontonasal mass by interfering with retinoic acid receptor signaling at stage 10 (Schneider et al., 2001). Thus, RA ligand or receptor signaling in relation to cell survival, is spatially and temporally restricted during development.

Endogenous RA is required to maintain expression of cell survival gene Fgf8 and genes in the BMP pathway

The mechanism of Citral action may have been to increase expression of genes normally associated with the induction of the programmed cell death pathway in the face and elsewhere in the embryo, Bmp4 (Bone morphogenetic protein), Msx1 (Muscle segment homeobox), Msx2 (Ashique et al., 2002, Barlow and Francis-West, 1997 and Graham et al., 1994). In addition, decreased RA synthesis may affect genes in the RA pathway such as the retinoic acid receptors and these in turn may regulate expression of genes involved with cell survival such as Fgf8 (Abu-Issa et al., 2002, Brondani and Hamy, 2000, Brondani et al., 2002 and Trumpp et al., 1999) and Sonic Hedgehog (Shh) (Ahlgren and Bronner-Fraser, 1999 and Sanz-Ezquerro and Tickle, 2000). Rarβ (retinoic acid receptor β) was previously shown to be expressed in the developing lateral nasal and maxillary prominence (Rowe et al., 1991 and Rowe et al., 1992) and is known to be directly activated by retinoid acid (Rossant et al., 1991).

We examined expression at 3 h (stage 20.5, before increases in programmed cell death), 6 h (stage 21, initiation of apoptosis in the lateral nasal prominence), and 16 h (stage 22.5, irreversible patterning changes have taken place, increased apoptosis) following bead implantation. Expression levels were referenced to DMSO or non-treated, stage-matched controls rather than to the contralateral side since we wanted to determine if Citral had bilateral effects on gene expression (Table 2). Msx1 and Bmp4, but not Msx2, were expressed at the start of our experiment (Figs. 4A, D, G). Expression of Msx1 and Bmp4 was decreased in the lateral nasal prominence following Citral treatment (Figs. 4B, C, I–K) and Msx2 failed to upregulate in the globular process (Figs. 4E, F). There were no discernable effects on maxillary prominence expression of Msx1 and Bmp4 (Figs. 4B, C, J, K). There were also short-term decreases in expression of Msx1 and Msx2 in the globular process of the contralateral side that were correlated with effects on the maxillary process of the premaxilla (Table 2; Figs. 4B, E). In contrast, Bmp4 transcripts were not reduced in the globular process on the untreated side at any time point (Figs. 4I–K; Table 2). Thus, it is conceivable that the effects on expression of Msx1 and Msx2 in the globular process are connected to the failure of fusion between the frontonasal mass and maxillary prominences on the contralateral side. These expression data refute the idea that the increase in cell death associated with RALDH inhibition was caused by an activation of the BMP–Msx-programmed cell death pathway.