Abstract
An antibody that recognizes tyrosine hydroxylase can be used as a marker for hatching gland cells in Xenopus embryos. Using this marker, we have shown that hatching gland cells are induced at the end of gastrulation and that presumptive hatching gland cells are localized to the anterior neural folds in Xenopus. The movements of neurulation bring the hatching gland cells together to form a characteristic Y pattern on the dorsoanterior surface of the head. The Y pattern delineates several zones of surface ectoderm which can be visualized by the presence or absence of ciliated cells. As development proceeds the hatching gland pattern is altered, demonstrating the active changes involved in forming the face. Lithium, UV irradiation and retinoic acid can be used to alter the hatching gland pattern in specific ways which help to understand the underlying mechanisms of ectodermal patterning.
Introduction
The surface ectoderm of the Xenopus embryo contains several transitory cell types, which are among the first to differentiate in development. These cell types are cement gland cells, hatching gland cells and ciliated cells. Study of these cells presents several advantages in determining the mechanisms of embryonic induction and patterning of tissues. First, the cells appear on the surface of the embryo making them easily viewed in whole mount and immediately accessible to surgical manipulation. Second, the cement gland and hatching gland comprise only a few cells and are located in small, precise regions of the embryo, making them useful as dorsoanterior markers. Third, these three cell types are positioned in a way that allows the entire surface of the Xenopus embryo to be defined, as will be shown in the present work.
The hatching gland aids in the release of the embryo from the fertilization envelope and jelly by releasing proteolytic enzymes which partially digest these barriers. The proteolysis coupled with early muscular movements of the embryo causes the jelly and fertilization envelopes to rupture releasing the tadpole to the environment (Bles, 1905; Carroll and Hedrick, 1974). The cells can be detected after hatching but disappear later in development (Nieuwkoop et al. 1985). Recently, a cDNA clone, UVS.2, and an antibody raised against its protein product have been identified and found to be specific to hatching gland (Sato and Sargent, 1990). UVS.2 RNA is detectable in the anterior neural folds, shortly before neural fold closure, by in situ hybridization, and the antibody against UVS.2 detects the protein shortly after neural fold closure (Sato and Sargent, 1990). The cDNA clone XA-1 (Sive et al. 1989) is another hatching-gland-specific marker (Hemmati-Brivanlou et al. 1990), which can be detected at the end of gastrulation.
We have made the fortuitous discovery that an antibody raised against tyrosine hydroxylase (TH), the rate-limiting enzyme in the synthesis of dopamine, can act as a marker for hatching gland cells on the surface of the Xenopus embryo. This has allowed us to answer questions about the development of hatching gland cells and to define some of the factors that determine their final pattern on the embryo.
Materials and methods
Embryos
Mature Xenopus laevis females were primed for ovulation with 50 IU of pregnant mare serum gonadotrophin 24 –72 h before ovulation was induced by injection of 500–700 IU of human chorionic gonadotrophin. Eggs were fertilized using minced testis in 80% Steinberg’s solution. Fertilized eggs were flooded with 20 % Steinberg’s solution and then dejellied with 2.5% cysteine, pH8.0. Embryos that failed to cleave normally were removed at the 2-to 4-cell stage. Embryos were raised at temperatures varying from 13 to 18°C, depending on the experiment and stage of embryo needed, and embryos were staged according to Nieuwkoop and Faber (1967).
Embryo manipulations
Explant experiments
Operations on the embryos were performed in modified Barth’s solution, pH 7.8 (MBS) (Gurdon, 1977). Explants were removed from embryos ranging in age from stage 10 to 15. In all cases, the explants were a single layer of surface cells removed using fine forceps and a sharpened needle. The thickness of a few explants was checked by histological examination after fixing in Smith’s fixative (Smith, 1912) before they began to curl. Operated embryos were allowed to heal in MBS and were transferred to 20 % Steinberg’s solution with antibiotics (10 mg 1-1 penicillin, 10 mg 1-1 streptomycin sulphate) for further development. Explants were allowed to develop in MBS.
For gastrula explants, a square of about 25 % of the embryo diameter was taken lateral to the dorsal midline to catch one side of the presumptive anterior neural fold zone. The unoperated side acted as a control. Explants from the neural folds-were made from the surface cells and consisted primarily of cells directly above the neural fold although some cells lateral to the folds were likely also taken. Except for the transverse fold explants, one side of the embryo was left untouched as a control for anti-TH-staining. All explants were allowed to develop to the equivalent of stage 28 before they were examined for TH-positive cells on their surface. The donor embryos developed normally and were assayed for any damage to the hatching gland pattern at stage 28.
Inhibition of neural fold closure
Closure of the anterior neural folds was inhibited by putting stage 13/14 embryos in either Danilchik’s medium (Keller et al. 1985) or MBS. Under these conditions, a variable number of embryos failed to completely close their neural folds. Despite the unfused neural tubes, embryos appeared to develop normally but had to be maintained in these solutions or the exposed neural tissue would disintegrate.
Alteration of dorsoanterior axis
Generation of embryos with alterations in the dorsoanterior axis by ultraviolet (UV) irradiation and lithium followed the procedures of Kao and Elinson (1988). The dose of UV was varied by varying the exposure time between 30 – 120 s The exposure to 0.3 M lithium was varied from 4 – 12 min. The various morphologies generated by these treatments were scored according to the Dorsoanterior Index (DAI) of Kao and Elinson (1988). ‘
Retinoic acid treatment
Retinoic acid was dissolved in dimethyl sulphoxide at 100mM. This was further diluted to working concentrations of 100 μM and 10 μ M in 20% Steinberg’s. Aliquots from the working concentrations were further diluted to give the 1/un and 0.1 gM solutions. Since some retinoic acid came out of solution upon dilution, the listed concentrations overestimated the actual concentrations. Embryos were treated with retinoic acid by incubation in the solutions for 30 min at stage 10.
Immunocytochemistry
Staining against tyrosine hydroxylase (TH) was performed using a mouse monoclonal antibody raised against TH from rat PC12 cells (antibody KTHM 788, Incstar Corp.). The anti-TH antibody was diluted 1:500 in a blocking buffer that consisted of 3.8% bovine serum albumin in phosphate-buffered saline (PBS) (110mM NaCl, 1.9ITIM KC1, 8rrtM Na2HPO4, 2mM KH2PO4, pH 7.4) with 0.02 % sodium azide.
Embryos were fixed with 2 % trichloroacetic acid in PBS and stained whole. The staining of hatching gland cells with the anti-TH antibody was highly dependent upon the fixation used. Neither cold methanol nor 4% paraformaldehyde in PBS preserved staining in hatching gland cells. Fixed embryos were washed three times in PBS for 5-10 min each and permeabilized by three washes in 0.1 % Triton X-100 in PBS for 15-30 min each. Permeabilized embryos were then washed three times in PBS for 5 min each. The entire wash routine was repeated after treatment with primary antibodies and secondary antibodies. Embryos were incubated in the anti-TH antibody overnight at 4°C. Visualization was done using a fluorescein-labelled goat anti-mouse IgG/IgM antibody (Jackson Laboratories), in which the embryos were incubated for 2 – 3 h at room temperature or overnight at 4 °C. Control staining consisted of applying blocking buffer instead of the primary antibody.
For whole-mount observations, embryos were dissected so that when squashed, both the front of the face and the back of the head were in the same focal plane. Explants were fixed, stained and viewed as whole mounts. When sections were needed to view internal fluorescent structures, embryos were stained as usual and then dehydrated by two 5 min washes in 50 % ethanol followed by two 5 min washes in 100% ethanol. The embryos were embedded in polyester wax (Steedman, 1957), and 20 pm sections were cut. Sections were floated in 0.05 % Triton X-100 on subbed slides. The solution used for subbing slides consisted of 72 ml distilled H2O, 30 ml 95% ethanol, 7 ml acetic acid, 1.5 g gelatin and 0.1 g chrome alum. Slides with sections were dried at room temperature, and then dewaxed with 100% ethanol. All embryos, explants and sections were mounted in 1.5% n-propyl gallate, 50% glycerol in PBS pH 8.1 and viewed under epifluorescence with a Zeiss Orthoplan fluorescence microscope.
Ciliated cells were stained with a monoclonal antibody against γ -tubulin (N.357, Amersham) diluted 1:500 in blocking buffer using the same protocol as for anti-TH.
Histology
Embryos were fixed in Smith’s fixative, rinsed twice with distilled water and stored in 4% formalin. The fixed embryos were embedded in paraplast (Sentein, 1976), and 10gm sections were cut. The sections were stained with the periodic acid – Schiff (PAS) method (Lillie, 1965) without counterstain. Sections of explants were fixed in the same way but stained with the Feulgen’s reaction and light green.
Results
Antibodies raised against tyrosine hydroxylase recognized Xenopus hatching gland cells
The earliest and most prevalent anti-TH staining in the Xenopus embryo was found in a distinct group of cells on the dorsoanterior surface of the head (Fig. 1). The staining appeared weak and punctate when first visible at stage 18/19 along the newly formed dorsal midline. By stage 24, the staining was so dense that a discrete pattern of cells was easily seen. Staining could be detected as late as stage 42, but the cells were no longer contiguous with each other and appeared quite small.
The pattern of cells, best observed in whole mounts, can be described as an inverted Y. The stem of the Y ran along the dorsal midline of the embryo and reached posterior to the level of the ear vesicle. The anterior arms of the Y diverged where the back bent to become the front of the head. These arms often extended to the lateral edges of the cement gland (Fig. 1).
Several lines of evidence suggested that the immunoreactive cells on the top of the head were hatching gland cells. First, the Y pattern matched descriptions of the hatching gland distribution in Xenopus by Yoshizaki (1973) and recently by Sato and Sargent (1990). Second, in section, immunoreactive cells had a distinct, bottleshaped morphology, a clear apical cytoplasm and an accumulation of pigment granules in the basal cytoplasm (Fig. 2). Immunoreactivity is concentrated in the clear apical cytoplasm. This matches previous cytological descriptions of Xenopus hatching gland cell morphology (Yoshizaki, 1973). Third, when sections were PAS stained, the cells thought to be hatching gland cells based on morphology had greater PAS staining, a characteristic of hatching gland cells (Yoshizaki, 1973).
In living embryos, the hatching gland pattern could usually be seen as lines of dark pigmentation. Although this heavily pigmented area appears quite thin in the intact embryo, the cells that react with the anti-TH antibody are actually a thinner subset of the pattern. The cells that appear darkly pigmented form a border around the hatching gland cells.
Staining of tissues, known to express TH in other organisms, provided evidence that the antibody was capable of cross reacting with Xenopus TH. These tissues included brain (Messenger and Warner, 1989), spinal cord and retina (Reh and Tully, 1986). Immunoreactivity was found in the brain of the Xenopus embryo starting at stage 34, with staining in the spinal cord starting shortly after. Staining of cells in the retina was found after stage 42 (Fig. 1). This indicates that the antibody is capable of recognizing embryonic Xenopus TH.
Attempts were made to use Western blotting to determine whether the antigen detected by the anti-TH antibody was the same molecular weight as the mammalian homologues. Although TH from mouse adrenal gland could be detected with this antibody, we have not yet been able to detect a band from embryonic or adult Xenopus tissues. Thus we cannot rule out the possibility that the antibody recognized a protein other than TH in the hatching gland cells.
Commitment and distribution of hatching gland cells
Explants were made from the surface layer of gastrulae (stages 10 – 12.5) to determine when the hatching gland cells were committed to their fate, and of neurulae (stages 14/15) to localize presumptive hatching gland cells within the neural plate and folds. Explants with surface cells immunoreactive to the anti-TH antibody were only found when the explants were removed after stage 12 (Table 1). Donor embryos, which were operated on before stage 12, had a morphologically normal hatching gland pattern, while donor embryos, which had explants removed after stage 12, showed loss of hatching gland pattern (Table 1). In general, this operation resulted in disruption of the hatching gland pattern on one side of the embryo.
When operations were done on neurulae, explants from the anterior neural fold area produced TH-positive cells, and the donor embryos, from which the explants were removed, had a damaged pattern on the operated side (Figs 3, 4). This damage usually consisted of one anterior tip having far fewer cells and, in some cases, none at all (Fig. 3). Other than these defects in the hatching gland pattern, the embryos appeared to develop normally with no obvious malformations. Explants from the posterior or transverse neural folds never had TH-positive surface cells, and generally led to no loss of pattern in the donor embryos (Fig. 4). Occasional alterations in the angle of divergence at the arms of the Y were observed.
This assay was only able to assess the origin of the arms of the Y and not the stem, because any missing portions of the stem would be masked by contributions from the other side of the embryo. The coming together of the hatching gland cells from both sides of the embryo to form the stem was demonstrated in embryos in which neural tube closure was inhibited. When neural folds failed to close in the regions where the hatching gland cells arose, fluorescent cells were found on either side of the exposed neural tissue (Fig. 5). Fusion of the neural folds was not required for differentiation of the hatching gland cells.
Formation of the hatching gland pattern and patterning of the face
The sharp pattern of hatching gland cells defined several boundaries on the embryo. The zone surrounded by cement gland and the arms of the hatching gland cells was different from the rest of the epidermis because it contained few ciliated cells (Fig. 6). The nasal pits and stomatodeum developed within the non-ciliated zone.
The divergent anterior tips of the hatching gland appeared to be stretched as the embryo developed because the tips were always adjacent to the cement gland as the cement gland moved further away from the edge of the neural fold. As would be expected if the arms were stretched, the pattern width was thin, often only one cell thick. In addition, the cells of the arms were rectangular, with the long axis aligned in the anterior–posterior direction suggesting that they were being stretched. This also occurred in the most caudal cells of the posterior tip suggesting that these were also stretched. In contrast, cells were generally cuboidal at the junction of the arms and at the stem of the Y (Fig-1).
Retinoic acid, lithium and UV light alter the hatching gland pattern
In order to determine which events in the formation of the body plan of the embryo were important in determining the pattern of hatching gland cells, three treatments, known to alter the body plan in different ways, were given to Xenopus embryos. These were UV irradiation, treatment with lithium and treatment with retinoic acid.
Dorsoanterior reduced embryos, resulting from UV irradiation, had altered or absent hatching glands depending upon the degree of irradiation. In DAI 4 embryos, the pattern was normal except that the angle of divergence for the anterior tips was decreased (Fig. 7C). With lower DAI embryos, this angle was reduced to zero and appeared as a line down the dorsoanterior midline (Fig. 7A). In some DAI 2 embryos and all DAI 1 or 0 embryos, the hatching gland cells were lost entirely.
Dorsoanterior enhanced embryos, resulting from lithium treatment, had a deterioration of the Y pattern (Fig. 7E and F). The midline cells of the stem appeared to be unzipped beginning at the posterior tip. In DAI 8 embryos, this unzipping reached the point where the arms normally diverged. In these extreme embryos, the posterior portion became very disorganized, but the predictability of the pattern was poor. In DAI 9 or 10 embryos, the pattern tended to break down into a large patch of cells over the anterior surface. The progressive nature of this pattern deterioration is shown in Fig. 8.
Retinoic acid caused a dose-dependent loss of hatching gland cells (Fig. 9). At concentrations of 100μm, retinoic acid resulted in a complete loss of hatching gland. These embryos lacked all anterior morphological structures and, in some cases, did not complete neural fold fusion. Loss of these structures did not appear to be due to deterioration of the embryo. At 10JUM, retinoic acid treatment resulted in a 10-fold reduction in the number of cells and the cells were loosely distributed on the dorsoanterior surface. However, even with severe reductions in the number of cells, the Y pattern was roughly maintained (Fig. 7G and H). The graded loss of hatching gland cells appeared to correspond to a graded loss in cement gland and eye tissue.
Discussion
Hatching gland as a marker for face morphogenesis
Hatching gland cells are a small group of cells on the surface of the Xenopus embryo, which can serve as excellent markers for early dorsoanterior development within the ectoderm. In Xenopus, hatching gland cells are morphologically distinguishable by the appearance of apical microvilli at stage 22 and by the presence of apical granules and extensive endoplasmic reticulum at stage 24 (Yoshizaki, 1973). We have shown that the anti-TH antibody can recognize cells that match the morphological description of hatching gland cells. It can also recognize these cells before the morphological markers appear. The clone UVS-2 can also detect hatching gland cells at stage 15/16 (Sato and Sargent, 1990), and the clone XA-1 marks hatching gland cells at the end of gastrulation (Hemmati-Brivanlou et al. 1990). Both of these markers appear before staining with the anti-TH antibody can be observed.
Using anti-TH antibody staining, we have found that hatching gland cells are determined at stage 12 to 12.5 in the surface ectoderm and that these induced cells come to lie above the anterior neural folds, similiar to the hatching gland cells of Rana japonica (Yoshizaki, 1976). Sive et al. (1990) obtained expression of XA-1, a hatching gland marker, in explants from stage 11.5 embryos. The apparent difference between this result and ours may be due to our removal of surface cells only for explants, whereas Sive et al. (1990) removed surface and deep cells. Alternatively, induction of these two markers may be independent events.
Since hatching gland cells arise on the surface of the neural folds, their later distribution is determined by the movements of those folds. The transverse fold moves posteriorly during neural fold closure, and this acts as a barrier, preventing the most anterior, lateral neural folds from joining at the midline. As these anterior folds are split, the hatching gland cells are separated at their anterior tip, creating the Y pattern visible later in development (Fig. 10). The transverse fold then becomes the non-ciliated zone bounded by the hatching and cement glands. This zone can also be visualized as an EPI-1 antigen-positive zone surrounded by cells lacking the EPI-1 antigen (Akers et al. 1986). Differentiation within this zone gives rise to both the nasal placodes and the stomatodeum. These organs are induced at the end of gastrulation (Balinsky, 1981), as are hatching gland cells, indicating that most ectodermal patterning is laid down at this time. Chu and Klymkowsky (1989) have described a non-ciliated zone on the dorsal midline of the embryo (Fig. 6B). In the cephalic region, this non-ciliated zone consists of hatching gland cells while the trunk midline is devoid of both ciliated and hatching gland cells. As development proceeds, the hatching gland cells can be used to visualize the movements of facial morphogenesis (Fig-10).
Mechanisms of hatching gland patterning as shown by alterations in embryonic morphology
We have been able to modify the normal Y-shaped pattern of hatching gland cells using three treatments that cause specific defects in the embryo: lithium, UV and retinoic acid.
UV and lithium both act to alter the character of the mesoderm (Kao and Elinson, 1988), and both caused alterations in the hatching gland pattern. Although the exact numbers of hatching gland cells were not counted, it was apparent that alterations in the pattern could occur without appreciable changes in cell numbers in both UV- and lithium-treated embryos. In DAI 2 embryos, with obviously reduced heads, there could be apparently normal numbers of hatching gland cells (Fig. 7A).
Changes in the anterior hatching gland pattern in UV-irradiated embryos can be best visualized as a narrowing of the Y pattern (Fig. 7B). The transverse neural fold, which causes the spreading of the anterior hatching gland cells, is the most anterior neural structure and is reduced by UV treatment. As the transverse neural fold is reduced, the Y narrows to a straight line (Fig. 7A). When all cephalic structures are lost (DAI 1), all hatching gland cells are lost.
In lithium-treated embryos, the pattern is more variable. In these embryos, increased levels of anterior neural tissues seem to inhibit the ability of neural folds to close properly. The surface ectoderm may heal over areas that failed to dose, and this causes variability in the final surface pattern. If the area does not heal over, the exposed neural tissue begins to degenerate and the embryo soon dies. When embryos are kept in MBS or Danilchik’s medium, the neural tissue does not degenerate and the hatching gland cells can be seen at the edge of the neural structures. Despite the variability, posterior parts of the hatching gland pattern are altered by shorter treatments of lithium than are the anterior parts of the pattern (Fig. 7E and F).
Alterations in the pattern of hatching gland cells by UV and lithium can be summarized as follows: alterations that reduce dorsal mesoderm caused a reduction in anterior hatching gland pattern and alterations that reduce ventral mesoderm caused a reduction in posterior hatching gland pattern. Thus anterior-posterior patterning of the hatching gland cells is determined by the earlier underlying dorsal-ventral pattern of the mesoderm.
Retinoic acid caused reductions in hatching gland cell number (Fig. 9) but did not alter the basic pattern (Fig. 7G). The changes in hatching gland cell number can be explained by random suppression of the hatching gland phenotype in cells that have already been patterned. As the distal tips of the pattern have fewer cells, this part of the pattern would be lost before the more densely populated center of the pattern. This suggests that retinoic acid acts after the neural patterning step, but before the hatching gland differentiates.
If retinoic acid acts as a transforming factor (Durston et al. 1989), it might be expected to cause a shift of the intact hatching gland pattern forward without narrowing the Y pattern, as it supplements the endogenous transforming factor gradient hypothesized by Nieuwkoop et al. (1952). Retinoic acid treatment has been shown to result in a forward shifting of the anterior nèural marker engrailed in Xenopus (Sive et al. 1990). Since this shifting was not observed for hatching gland cells, retinoic acid does not appear to act with any established gradient in the non-neural ectoderm; rather it suppresses anterior differentiation regardless of position in the anterior–posterior axis.
Surface versus deep axis
Although the surface and deep cells of Xenopus give essentially the same tissues when exposed to neural inducing substances in vitro (Asashima and Grunz, 1983), there are differences in the fates of these two layers in the embryo. Lineage tracing of the two layers in the neural plate has shown that surface cells give rise predominantly to secondary neurons, whereas the deep layer becomes predominantly primary neurons (Har-tenstein, 1989). Hatching gland cells at the surface of the neural plate boundary are in the same area where deep cells give rise to neural crest cells (Sadaghiani and Thiebaud, 1987).
Both our observations and those of Yoshizaki (1976) suggest that hatching gland cells arise only from surface cells. Sato and Sargent (1990) mention that the UVS.2 protein, which gives the same staining in whole mounts as the anti-TH antibody, is found in both deep and superficial ectodermal cells. However, we find that removal of surface cells, prior to the stage at which UVS.2 protein can be detected, eliminates hatching gland cells. Yanai (1951) also suggests that the hatching gland cells are thrust from the deep layer into the surface layer during neural fold closure. Although our observations indicate that hatching gland cells come from the surface only, we cannot rule out the possibility that hatching gland cells move into the surface layer from below at stage 12, a possibility that can be tested by lineage tracing.
The non-neural ectoderm provides a relatively simple system with which to investigate patterning in the Xenopus embryo. With the anti-TH marker for hatching gland cells, the surface of the embryo can be defined in terms of molecular markers and the simple patterns that are formed can be specifically manipulated. We hope to exploit this system to define the steps involved in normal patterning of the Xenopus embryo.
ACKNOWLEDGEMENTS
We thank Evelyn Houliston, Yoshio Masui, Andris Briedis, Michael Crawford, and an anonymous reviewer for helpful criticism, discussion and advice. We thank Ed Knapp for photographic work, Elizabeth Campolin for the drawings, and Richard Harland and Hazel Sive for communicating results prior to publication. This work was supported by a grant from NSERC, Canada to RPE.