ABSTRACT
Retinoic acid (RA) is thought to play a role in embryonic pattern formation in vertebrates. A naturally occurring gradient of endogenous RA has been demonstrated in the developing chick limb bud, while local application of RA leads to the formation of additional digits. In mammals, a well-defined spectrum of birth defects has been reported as a result of fetal exposure to excess RA. In analogy to the chick limb bud, it may be speculated that these malformations are the result of disturbance of morphogenetic RA concentration gradients.
A candidate gene involved in the regulation of endogenous RA concentrations is the gene encoding cellular RA binding protein (CRABP). We have isolated a partial cDNA clone corresponding to the chicken homolog of CRABP, and performed in situ hybridization experiments on sections of embryos at various stages of development. CRABP expression was detected in the CNS, the craniofacial mesenchyme, ganglia of the peripheral nervous system, the limb bud, and the visceral arch area. Our results indicate that the spatiotemporally specified expression pattern displayed by the CRABP gene exhibits a striking correspondence to the tissues that are affected by exposure of avian or mammalian embryos to RA. We hypothesize that CRABP plays an important role in normal embryogenesis and that embryonic tissues showing high CRABP expression are susceptible to the adverse effects of excess RA.
Introduction
All-trans retinoic acid (RA), a biologically active metabolite of retinol (vitamin A), plays an important role in cellular differentiation and pattern formation during embryonic development of vertebrate animals. In chicken embryos, RA was found to be present in physiological concentrations similar to those known to be sufficient for the induction of EC cell differentiation in vitro (Thaller and Eichele, 1987). In the same study, Thaller and Eichele also demonstrated that RA is present in a higher concentration in the posterior part of the developing limb bud as compared to the anterior part. Local application of RA to the anterior margin of the limb bud results in dose-dependent changes in the skeletal pattern (Tickle et al. 1982; Eichele, 1989; Tickle et al. 1989). Local application of RA to a chick blastoderm grown in vitro was shown to interfere with axis formation (Mitrani and Shimoni, 1989). In Xenopus laevis, RA was found to act on differentiation of the central nervous system (CNS), causing transformation of anterior neural tissue to a posterior neural specification (Durston et al. 1989). Thus, the conclusion is warranted that RA acts as a morphogen in various developmental processes.
RA has long been known to be teratogenic in humans (Kochar, 1967), and has been reported to be the cause of severe birth defects when administered to pregnant women (Lammer et al. 1985). RA-induced malformations include defects of the CNS (mainly hydrocephalus), cleft lip and cleft palate, and congenital heart defects. In rodents, the same spectrum of birth defects was observed, while high doses of RA have also been reported to result in limb deformations (Kochar, 1973). Also in chicken embryos, the tissues affected by RA treatment seem to be similar to those reported for mammals (Jelinek and Kistler, 1981). Excess RA obviously disturbs differentiation of the CNS in Xenopus laevis (Durston et al. 1989). From these observations it seems likely that RA-induced malformations represent a phenomenon general to all vertebrates.
If RA gradients influence morphogenetic processes during embryonic development, it is important to know which mechanisms regulate RA concentrations. In addition to regulation of the synthesis and degradation of RA, fine-tuning of RA levels may be achieved at the cellular level. In this respect, a possible role could be reserved for the cellular RA-binding protein CRABP. This 15.5x103MR protein, which displays a highly specific binding affinity for RA (Ong and Chytil, 1975; Jetten and Jetten, 1979), shows a structural similarity to the P2 family of proteins (Eriksson et al. 1981; Sundelin et al. 1985). Members of this family, which also includes cellular retinol binding protein (CRBP), are small, cytoplasmic proteins that have been implicated in the transport of specific small hydrophobic molecules.
CRABP exhibits a spatiotemporally restricted expression pattern in the mouse embryo (Vaessen et al. 1989a). Using in situ hybridization techniques, we detected a high level of CRABP transcripts in a subpopulation of cells in the CNS, as well as in the craniofacial mesenchyme. These results were later confirmed by others (Perez-Castro et al. 1989). It was also shown that CRABP is differentially expressed in the developing limbs of the mouse (Perez-Castro et al. 1989; Dollé et al. 1989). Maden et al. (1989) reported on the immunocytochemical localization of CRABP in the chicken embryo. Apart from particular cells of the neural tube, CRABP was also found in various neural crest derivatives, including dorsal root ganglia and enteric ganglia. Earlier, these authors showed that a gradient of CRABP is present at the tip of the chick limb bud, with its maximum concentration in the anterior part (Maden et al. 1988).
Apart from the limb bud, all major CRABP expression sites reported are of neurectodermal origin, and notably include many neural crest derivatives. Because of this observation, and also because neural crest cells have been mentioned as candidate targets for RA-induced malformations (Poswillo, 1975; Lammer et al. 1985), it would be of interest to know whether other neural crest cell derivatives also show high levels of CRABP expression. In order to study CRABP expression during embryogenesis, we performed in situ hybridization experiments on chicken embryos at various stages of development. Use of the chicken embryo for expression studies facilitated the acquisition of well-standardized material of early developmental stages. Furthermore, neural crest cell migration has been most thoroughly studied in avian embryos, and several markers for early neural crest cells have become available (reviewed by Anderson, 1989). We chose to use the monoclonal antibody HNK-1 (Abo and Balch, 1981), which recognizes a sulfated glucuronic acid present on several cell adhesion molecules (Kruse et al. 1984, 1986; Pesheva et al. 1987; Rathjen et al. 1987; Hoffman and Edelman, 1988). In the chicken embryo HNK-1 binds to most premigratory neural crest cells and neuronal neural crest cell derivatives (Vincent et al. 1983; Vincent and Thiery, 1984; Tucker et al. 1984). In addition, HNK-1 is an early marker for neural differentiation induced by RA in murine embryonal carcinoma cells (McBumey et al. 1988).
Our results show that the CRABP gene exhibits a spatiotemporally specified expression pattern, which offers a striking correlation to parts of the embryo that correspond to the tissues commonly affected by RA-induced malformations. The .observed correlation is highly suggestive for a role of CRABP in RA-mediated morphogenetic processes, and may help to understand the underlying molecular mechanisms.
Materials and methods
Isolation of a chicken CRABP cDNA clone
A chicken embryo cDNA library, consisting of oligo(dT)-primed cDNAs inserted into the EcoRI site of bacteriophage lambda gtll (Clontech), was screened with a mouse CRABP cDNA probe. This probe, a 170bp Taql-Taql fragment isolated from clone MoT-CAll and containing part of the CRABP-coding region (Vaessen et al. 1989a), was labeled with æP using random priming (Feinberg and Vogelstein, 1983). The filters were hybridized at 56°C, in a buffer containing 6xSSC and 9% dextran sulphate, and washed at the same temperature, twice in 3xSSC, 0.1% NaDodSO4, and twice in lxSSC, 0.1% NaDodSO4.
Screening of approximately lx106 bacteriophage plaques resulted in the isolation of a cDNA clone, designated C4. DNA isolated from clone C4 was digested with EcoRI, and ligated to EcoRI plasmid vector pTZ18R (Pharmacia). Transformation of E. coli DH5αF’ yielded the subclone ChCRABP C4.5, which contains a cDNA insert of approximately 300 bp.
Sequence analysis
For determination of the nucleotide sequence, the ChCRABP C4.5 cDNA insert was subcloned into bacteriophage M13 mpl9. The cDNA insert was cloned in both orientations, allowing both strands to be read. After isolation of single-stranded DNA, the nucleotide sequence was determined via the dideoxy chain termination method (Sanger et al. 1977) using Sequenase (United States Biochemical) according to instructions by the manufacturer.
Computer-assisted analysis of the DNA sequence was carried out with a Microgenie software package (Beckman).
Chicken embryos
Fertilized eggs from the White Leghorn (Gallus domesticas) were incubated at 38 °C in a forced-draft incubator at a relative humidity of 80%. Staging of the embryos was performed according to Hamburger and Hamilton (1951).
RNA isolation and blot hybridization
Total RNA was isolated from whole chicken embryos using the LiCl/Urea method described by Auffray and Rougeon (1980). For RNA blot analysis, RNA was electrophoresed on 1 % agarose gels in the presence of formaldehyde. Prior to electrophoresis, ethidium bromide was added to the RNA samples in order to allow visualisation of the ribosomal RNAs in the gel. In this way, it was ascertained that the amounts of RNA in the different lanes were approximately the same.
Following electrophoresis, the RNA was transferred to nitrocellulose filters (Maniatis et al. 1982). DNA probes were labeled as described above. The RNA blots were hybridized at 42°C in a buffer containing 6xSSC, 9 % dextran sulphate, and 50% formamide, and washed at 56°C, twice in 3xSSC, 0.1% NaDodSO4, and twice in lxSSC, 0.1% NaDodSO4.
Preparation of embryo sections
Chicken embryos were fixed for 24 h in 4 % paraformaldehyde in phosphate-buffered saline at 4°C. After fixation, the embryos were embedded in paraffin. Sections were cut at 5 μm, placed on chromium(III)potassium sulphate coated slides, and air dried.
Prior to in situ hybridization or immunocytochemical treatment, the sections were deparaffinized and hydrated.
In situ hybridization techniques
We used the cDNA insert of ChCRABP C4.5, labeled with 35S via nick translation, and treated with DNAsel to obtain fragments of approximately 50-100 bp as a probe for in situ hybridization to the chicken embryo sections.
Section pretreatment and in situ hybridization procedures were based on standard methods (Akam, 1983; Hafen et al. 1983). For autoradiography, slides were dipped in Kodak NBT-2 emulsion, and developed after 5-10 days of exposure. Counterstaining with haematoxylin was carried out as required.
Immunocytochemistry
The HNK-1 hybridoma cell fine was purchased from the American Tissue Culture Collection (ATCC HU 200). HNK-1 immunoperoxidase staining was performed using undiluted supernatant. Rabbit anti-mouse peroxidase-conjugated immunoglobulins (Dako, Denmark) were used in a dilution of 1:100. In order to reduce background staining, 2% chick serum was added to the conjugate. Peroxidase was visualized by 0.1% 3,3’diaminobenzidine.4HCl (Serva) and 0.02% hydrogen peroxide. All rinsing and diluting was done in phosphate-buffered saline with 0.1 % Tween 20.
Results
Isolation of a chicken CRABP cDNA clone
We obtained a chicken CRABP probe by screening a chicken embryo cDNA library with a Taqi-TaqI fragment containing nucleotides 147-315 of mouse CRABP cDNA clone MoT-CAll. This resulted in the isolation of bacteriophage clone C4, from which an EcoRI subclone in plasmid vector pTZ18R was derived. This subclone, which was named ChCRABP C4.5, contains a 310 bp cDNA insert, which corresponds to nucleotides 129-438 of mouse CRABP clone MoT-CA11. The sequence of ChCRABP C4.5 and its comparison with the mouse CRABP sequence is shown in Fig. 1. The chicken cDNA clone contains part of the CRABP-coding sequence. Homology with mouse CRABP is 87 % at the nucleotide level and 94 % at the amino acid level.
Sequence of the partial cDNA clone encoding chicken CRABP. The cDNA sequence and deduced amino acid sequence of ChCRABP C4.5 are shown in bold face, with underneath the corresponding sequence of mouse CRABP cDNA clone MoT-CAll. Non-homologous nucleotides are indicated by asterisks, and non-homologous amino acids are underlined. The EcoRI sites that flank the C4.5 cDNA insert are indicated.
Sequence of the partial cDNA clone encoding chicken CRABP. The cDNA sequence and deduced amino acid sequence of ChCRABP C4.5 are shown in bold face, with underneath the corresponding sequence of mouse CRABP cDNA clone MoT-CAll. Non-homologous nucleotides are indicated by asterisks, and non-homologous amino acids are underlined. The EcoRI sites that flank the C4.5 cDNA insert are indicated.
The 32P-labeled ChCRABP C4.5 insert DNA was hybridized to nitrocellulose blots containing total RNA isolated from chicken embryos at various stages of development. This resulted in detection of one single transcript of approximately 1 kb (see Fig. 2), which is in agreement with the size of the mouse CRABP mRNA (Vaessen et al. 1989a). Thus, ChCRABP C4.5 specifically hybridizes to chicken CRABP mRNA and can be used to study CRABP expression patterns in the chick embryo.
Hybridization of ChCRABP C4.5 to RNA samples isolated from (a) day 2; (b) day 3; (c) day 4; (d) day 5; and (e) day 6 chicken embryos. The position of the 28S and 18S ribosomal RNAs is indicated. The 18S ribosomal RNAs, stained with ethidium bromide, are shown underneath the corresponding lanes.
Hybridization of ChCRABP C4.5 to RNA samples isolated from (a) day 2; (b) day 3; (c) day 4; (d) day 5; and (e) day 6 chicken embryos. The position of the 28S and 18S ribosomal RNAs is indicated. The 18S ribosomal RNAs, stained with ethidium bromide, are shown underneath the corresponding lanes.
Localization of CRABP transcripts in chicken embryos
We performed in situ hybridization experiments with the radiolabeled cDNA insert of ChCRABP C4.5 on sections obtained from chicken embryos at various stages of development. The CRABP gene exhibits a strongly restricted expression pattern in day 2, day 3 and day 4 embryos, which tends to become more diffuse from day 5 onwards. High CRABP expression was found in the CNS, in the craniofacial mesenchyme, in the visceral arches, and in the ganglia of the peripheral nervous system. CRABP transcripts were also detected in the limb bud, where expression was predominantly found in mesenchymal cells located in the anterior part.
In order to establish a possible relationship between expression sites of CRABP and the occurrence of migratory neural crest cells, we performed immunostaining experiments with the neural crest cell marker HNK-1 on sections serial to those used for the in situ hybridizations.
The following results were obtained:
(A) CRABP expression in stage 12 embryos
Figs 3 and 4 show serial sections of a stage 12 embryo (day 2) that were subjected to in situ hybridization with a CRABP probe (Fig. 3) and immunostaining with HNK-1 (Fig. 4). Occasional CRABP-positive cells were observed in the outer cell layer of the mesencephalon, rhombencephalon and spinal cord, but not in the prosencephalon. In the cells of the auditory pit, which is deep and wide open at this stage, CRABP transcripts were also detected. These cells exhibit strong expression of the HNK-1 epitope, while the cells of the neural tube are negative. CRABP expression was also observed in the craniofacial mesenchyme anterior to the telencephalon and diencephalon. These mesenchymal ceils are also positive for HNK-1.
Localisation of CRABP transcripts in a longitudinal section from a stage 12 (day 2) embryo. CRABP-positive celts, indicated by arrows, appear in the CNS. the craniofacial mesenchyme, and the auditory pit (A). An elevated hybridization signal is also observed in the presumptive neural crest cells that are shown to be HNK-1 positive in Fig. 4 (open arrow).
Localisation of CRABP transcripts in a longitudinal section from a stage 12 (day 2) embryo. CRABP-positive celts, indicated by arrows, appear in the CNS. the craniofacial mesenchyme, and the auditory pit (A). An elevated hybridization signal is also observed in the presumptive neural crest cells that are shown to be HNK-1 positive in Fig. 4 (open arrow).
HNK-1 immunoreactivity in a section serial to the one shown in Fig. 3. HNK-1 positive ceils appear in the craniofacial mesenchyme, (he auditory pit (A). A group of HNK-1 positive cells, presumably neural crest cells, is observed anterior to the auditor)’ pit (open arrow).
(B) CRABP expression in stage 22 embryos
In stage 22 embryos (day 3), cells that show a strong hybridization to the CRABP probe were detected in the outer cell layers of mesencephalon, rhombencephalon and spinal cord. CRABP-positive cells are lying apart or in small clusters, but are more abundant than in stage 12 embryos (see Fig. 5). The distribution of CRABP transcripts coincides with HNK-1 expression, as illustrated by Fig. 6. As can be seen in Fig. 7, CRABP expression is also observed in the neuroepithelial cells of the auditory vesicle, which is HNK-1 positive. The craniofacial mesenchyme, located anterior to telencephalon and diencephalon, and posteriorly confined by the eyes, exhibits a high level of CRABP transcripts (see Fig. 5). These cells no longer express the HNK-1 epitope at this stage.
Transverse section from the head region of a stage 22 (day 3) embryo, with CRABP-specific hybridization in the CNS and the craniofacial mesenchyme (arrows). D, diencephalon; M, mesencephalon.
HNK-1 immunoreactivity in a transverse section from a stage 22 (day 3) embryo, serial to the section shown in Fig. 5. Positive cells are present in the CNS, and in the lateral parts of the head mesenchyme, mainly in the area surrounding the eyes. D, diencephalon; M, mesencephalon.
CRABP-specific hybridization in a longitudinal section from a stage 22 (day 3) embryo, showing CRABP expression in the visceral arch area, and in the auditory vesicle (A). 1, 11, 111, first, second, and third visceral arch.
In addition to the CNS and the craniofacial mesenchyme, CRABP-positive cells were detected in the visceral arches, as illustrated by Figs 7 and 8. The CRABP-positive region extends from the craniofacial mesenchyme of the head into the first or mandibular arch, where an even higher level of CRABP transcripts was observed. High CRABP expression continues into the second, third and fourth arch. The hybridization signal appears to be strongest in the mesenchymal cells bordering the overlying epithelium of pharynx, visceral clefts, and pericardial cavity, while the epithelium itself is CRABP negative. In the viscera! arch area, only occasional HNK-1 positive cells were observed.
Transverse section from a stage 22 (day 3) embryo, demonstrating CRABP expression in the mantle layer of the myelencephalon (My), in the first viscera] arch, and in the fifth cranial ganglion (arrow). P, pharynx.
CRABP expression was also detected in cranial and dorsal root ganglia, which are positive for HNK-1.
(C) CRABP expression in stage 24 embryos
In stage 24 embryos (day 4), a high level of CRABP expression was observed in the dorsolateral part of mesencephalon and metencephalon, extending to lateral and ventral, and continuing into myelencephalon and spinal cord. CRABP-positive cells were also observed in the ventral part of telencephalon and diencephalon, posteriorly confined by the olfactory pit, which is CRABP negative. The strong CRABP-specific hybridization signal has extended to nearly the whole outer neural cell layer, as shown in Fig. 9. In embryos of this stage, HNK-1 expression was observed all along the neural tissue.
CRABP-specific hybridization in a transversal section from a stage 24 (day 4) embryo, with positive cells in the mantle layer of the myelencephalon (My).
In the craniofacial mesenchyme, CRABP expression is still detectable, but reduced as compared to stage 22 embryos. Expression in the visceral arch area is similarly diminished, with the strongest hybridization signal appearing at the ventral side of the arches. At this stage, neither the craniofacial mesenchyme nor the visceral arches showed any immunoreactivity with HNK-1. An elevated level of CRABP transcripts was still detected in the spinal cord, and in the dorsal root ganglia and cranial ganglia, which showed a continuously strong expression of the HNK-1 epitope. Fig. 10 shows a transverse section through the spinal cord, with a characteristic distribution of CRABP transcripts on the dorsolateral sides and in two cel! groups located vent rally, near the notochord.
Transversal section from a stage 24 (day 4) embryo, showing CRABP expression in the neural tube and in a dorsal root ganglion (open arrow). The CRABP-positive cells located on the ventral side of the neural lube are indicated by solid arrows. N= notochord.
Discussion
We have isolated a cDNA clone encoding part of chicken CRABP and used it for in situ hybridization studies of CRABP expression during chicken embryogenesis. After determination of the nucleotide sequence of our partial cDNA clone ChCRABP C4.5, we compared the deduced amino acid sequence with mouse CRABP (Vaessen et al. 1989a). Only 6 out of 100 amino acids were different, with most substitutions concerning amino acids with similar physicochemical properties. The high degree of conservation observed for CRABP in cow (Sundelin et al. 1985), mouse and chicken suggests that the biological function of the protein does not allow important structural variation. The protein sequence of ChCRABP C4.5 is in agreement with the partial sequence of chicken CRABP type I as reported by Kitamoto et al. (1988). These authors state that two types of CRABP containing 6 amino acid replacements in the NH2 terminal region are present in the chick embryo: a major one, type I, and a minor one, type II. It would be of interest to know whether CRABP I and II are encoded by different loci, or are the result of e.g. differential mRNA splicing. However, Southern blotting experiments performed on mouse and chicken DNA as well as on DNA isolated from mouse-hamster and human-hamster somatic cell hybrids did not, sofar, give any indication for the existence of more than one CRABP gene (Vaessen et al. 1989a,b).
Our in situ data demonstrate that CRABP expression in the CNS is strictly limited to a subpopulation of neural cells in day 2, day 3 and day 4 embryos. CRABP expression begins in a few single cells, gradually expanding to the whole outer cell layer of mesencephalon, rhombencephalon and spinal cord. In agreement with the results reported earlier for the mouse embryo, we also detected a high level of CRABP transcripts in cells of the craniofacial mesenchyme. In addition, we here report that CRABP is highly expressed in the visceral arches. As a matter of fact, CRABP expression in the visceral arch area also occurs in the mouse embryo (Vaessen, unpublished results). Thus, CRABP expression patterns in chicken and in mouse embryos show a strong similarity.
The major CRABP expression sites reported here exhibit a striking correspondence to tissues commonly affected by exposure of mammalian embryos to RA. Clinical reports include malformations of the CNS - mainly hydrocephalus; of structures derived from the craniofacial and mandibular arch mesenchyme - microtia/anotia, micrognathia and cleft palate; congenital heart defects - predominantly conotruncal or branchial-arch mesenchymal tissue defects, occasionally combined with thymic defects (Lammer et al. 1985). Although the adverse effects of RA on the chicken embryo are less well documented, the target tissues seem to be the same (Jelinek and Kistler, 1981). Several reports implicate Vitamin A in the inhibition of cranial neural crest cell development in the chick embryo (Hassell et al. 1977; Keith, 1977). In addition, recent results obtained in our own group demonstrate that RA treatment of chicken embryos may give rise to congenital heart defects and craniofacial malformations as well as to limb deformations (M. Broekhuyzen, pers. comm.). Apart from leading to digit duplications, local application of RA to the chick wing bud may also result in upper beak defects (Tickle et al. 1982; Tamarin et al. 1984; Wedden and Tickle, 1986). Interestingly, after treatment with RA we also obtained an embryo with a cleft lower beak (Vaessen, unpublished results), suggesting that in the chicken embryo the first arch - the lower beak primordium - is susceptible to the teratogenic effects of RA, just as in the mammahan embryo. The fact that Wedden and Tickle did not observe effects of RA on the lower beak must probably be ascribed to their using local application of FLA in the chick wing bud as a means to generate malformations. In conclusion, the resemblance of CRABP expression patterns between chicken and mouse embryos, and the fact that in both species the CRABP-positive tissues appear to be susceptible to the adverse effects of RA, leads to the assumption that similar RA-sensitive morphogenetic processess take place in both species, and that CRABP expression has an important function in these processes.
The tissues that are frequently affected by RA-induced malformations share a common embryonic origin in that they have all received contributions from the cephalic neural crest. It has been suggested that excess RA has an adverse influence on cephalic neural crest cells, possibly by interfering with normal neural crest cell migration (Thorogood et al. 1982; Pratt et al. 1987). This theory is supported by the observation that RA interferes with the cell - substratum adhesion of neural crest cells in vitro (Smith-Thomas et al. 1987). If endogenous RA concentrations also determine the migratory behaviour of neural crest cells in vivo, differential expression of CRABP could play a role in the regulation of this procès. Maden et al. (1989) described the occurrence of single CRABP-positive cells in a fine from the dorsal neural tube to the lateral edge of the dorsal aorta and suggest that these cells may be neural crest cells in the progress of migration. The fact that these CRABP-expressing cells have escaped our attention can be explained by immunocytochemistry being better suited for detection of single positive cells than in situ hybridization, due to the higher background levels obtained with the latter technique. However, we observed two additional CRABP expression sites that may be related to neural crest cell migration: the craniofacial mesenchyme and the visceral arch area.
For a more detailed investigation of a possible relationship between CRABP expression sites and neural crest cell migration, we employed the monoclonal antibody HNK-1 (Abo and Balch, 1981), which recognizes an epitope present on several cell adhesion molecules. Canning and Stern (1988) showed that HNK-1 identifies tissues involved in mesoderm formation in the chick embryo. Prior to mesoderm induction, HNK-1 binds to the inducing tissue (hypoblast) and reveals a mosaic pattern in the responding tissue (epiblast). After primitive streak formation, the epiblast displays an anteroposterior gradient of HNK-1 expression. At the end of gastrulation, the primitive streak region loses its HNK-1 reactivity. HNK-1 expression is next seen in cells of the forming notochord and in cranial neural crest cells (Canning and Stem, 1988; Stern and Canning, 1990). In later stages of chicken embryogenesis, HNK-1 recognizes most pre-migratory neural crest cells and neuronal neural crest cell derivatives (Vincent et al. 1983; Vincent and Thiery, 1984; Tucker et al. 1984). However, thé absence of HNK-1 immunoreactivity does not rule out the posibility that migratory neural crest cells are present since disappearance of the HNK-1 epitope from cranial neural crest cells at certain stages of migration has been reported (Vincent and Thiery, 1984; Bronner-Fraser, 1987).
In view of the existence of multiple HNK-1 antigens during development and adult life, it might seem surprising that migrating neural crest cells in the avian embryo can be visualized specifically with HNK-1 or related antibodies. Although reactivity is not restricted to neural crest cells, use of HNK-1 as a marker for neural crest cells is possible because the staining of other antigenic lineages does not overlap topographically or temporally with the distibution of crest cells. For one thing, neural crest cells do not become HNK-1 positive until they leave the neuroepithelium and start to migrate. Additional proof for the presence of the HNK-1 epitope on migrating neural crest cells is the fact that injection of HNK-1 antibodies lateral to the mesencephalic neural tube perturbs cranial neural crest cell migration (Bronner-Fraser, 1987).
In our investigation of a possible relationship between CRABP expression and neural crest cells we found that the cranial and dorsal root ganglia are positive for both CRABP and HNK-1. In the CNS, a partial coincidence of CRABP and HNK-1 expression is observed, with a remarkable colocalization at stage 22. In stage 24 embryos, CRABP continues to be differentially expressed, while HNK-1 reactivity occurs all along the neural tissue. However, the observed expression of the HNK-1 epitope in the CNS is probably unrelated to neural crest cell migration.
The cells of the craniofacial mesenchyme are initially (stage 12) positive for HNK-1 and also express CRABP. Later during development (stage 22) they start to differentiate into cartilage, muscle and bone, and lose the HNK-1 epitope. In contrast, CRABP expression in the craniofacial mesenchyme continues until stage 24. The visceral arches are filled with cells of neural crest origin (Le Lièvre and Le Douarin, 1975) which contribute to the development of the heart and arch arteries (Bockman et al. 1987; Philips et al. 1987). While showing a high level of CRABP expression, these neural crest derived cells do not express the HNK-1 epitope anymore. After administration of RA to chicken embryos Jelinek and Kistler (1981) showed that treatment on day 3 frequently gave rise to heart defects. Administration on day 4 resulted in a high incidence of heart defects and craniofacial malformations, as well as limb deformations. A tentative suggestion is evoked that the period of sensitivity to RA treatment of a particular morphogenetic system coincides with CRABP expression and loss of HNK-1 reactivity.
The role of CRABP in RA-mediated morphogenetic processes is still poorly understood. It has been suggested that CRABP has a function as a transport protein, mediating transfer of RA to the nucleus, where RA is thought to exert its biological activity (Takase et al. 1986; Shubeita et al. 1987). On the other hand, the human myelocytic leukemia cell line HL60, which is deficient in CRABP (Breitman et al. 1982; Douer and Koeffler, 1982), is still able to differentiate in response to RA, indicating that binding to CRABP is not obligatory for transport of RA to its nuclear receptor sites.
A different model, proposed by Hirschel-Scholz et al. suggests that CRABP has a buffer function, protecting the cell from the deleterious effects of unbound RA (Hirschel-Scholz et al. 1989). Maden et al. reason along the same lines, proposing a function for CRABP in the sequestering of RA in the cytoplasm (Maden et al. 1989). The existence of reciprocating concentration gradients for RA and CRABP in the developing chick limb bud (Thaller and Eichele, 1987; Maden et al. 1988) is consistent with these models. While the concentration of RA itself is highest in the posterior part, CRABP is present in a higher concentration in the anterior part. Thus, CRABP could be effective in reducing the concentration of free RA in the anterior part.
Recently, a nuclear receptor for RA (RARα) was identified, which was shown to be related to the steroidμhyroid hormone receptor family (Petkovich et al. 1987; Giguere et al. 1987). In addition, three more RA receptors were identified, designated RARβ, RARγ, and RARδ, which were related to RAR α but were obviously encoded by different genes (Brand et al. 1988; Krust et al. 1989; Ragsdale et al. 1989).
It is assumed that the different RA receptors act as transcription factors, mediating expression of specific sets of genes. This suggests that RA-induced malformations occur as the result of aberrant gene expression. Evidently, genes known from in vitro experiments to be susceptible to RA induction, such as the gene encoding Growth Hormone (Bedo et al. 1989), and the homeo-box-containing genes (Colberg-Poley et al. 1985a,b; Breier et al. 1986; Deschamps et al. 1987; Mavilio et al. 1988), are candidate target genes involved in RA-induced malformations.
We propose that a high level of CRABP expression in certain tissues reflects a particular sensitivity to RA. CRABP would be instrumental in protecting cells from the developmentally important action of RA by prohibiting aberrant activation of RA responsive gene sequences during critical stages. This would explain why fetal exposure to excess RA predominantly affects those tissues that exhibit a high level of CRABP expression during certain stages of embryogenesis.
ACKNOWLEDGEMENTS
The authors thank R. Beekhuizen for technical assistance. This work was supported by the Netherlands Cancer Society (Koningin Wilhelmina Fonds).
‘The nucleotide sequence data reported here will appear in the EMBL, GenBank and DDBJ Nucleotide Sequence Databases under the accession number X53701 CHICKEN CRABP.’