Head morphogenesis in embryonic avian chimeras: evidence for a segmental pattern in the ectoderm corresponding to the neuromeres

The development of the cephalic region of higher vertebrates involves extremely complicated morphogenetic processes. This part of the body has been the site of enormous evolutionary changes, most of which have involved the addition of structures related to the increasing complexity of the brain and associated sensory organs.

The mechanisms involved in assembling the various pieces of the puzzle during ontogeny of the head are still imperfectly understood. This is why we thought that it would be informative to draw up a precise map of the cephalic region at the early stages of neurogenesis and to combine the cartographical study with a topological and dynamic analysis of development. The quail-chick chimera system is ideally suited to such an investigation. The work done in our laboratory and elsewhere concerning the fate of the cephalic neural crest (Le Lièvre, 1974, 1978; Le Lièvre and Le Douarin, 1974, 1975; Johnston, 1966; Johnston and Hazelton, 1972; Noden, 1975, 1984) has already shown that the facial skeleton and dermis do not originate in situ but develop after a phase of dorsoventral migration of cells from the cephalic neural crest. In our previous studies (Couly and Le Douarin, 1985, 1987), we found that the anterior neural fold (located cranially with respect to the prospective neural crest) yields the superficial epidermis covering the forehead and nasofrontal regions. The morphogenetic movements affecting the more rostral parts of the neural fold could be followed from the presomitic stage until the cells derived from this zone were definitively positioned in the head structures.

The present work is essentially concerned with the development of the remaining cephalic ectoderm, from the neurula to the late embryonic stage, and with its relationships to cerebral and neural crest structures.

We report here a striking regionalization of the presumptive facial and hypobranchial ectoderm in areas, arranged in a metameric-like fashion, that we call ‘ectomeres’ and which parallel the segmentation of the neural anlage into individual neuromeres. This métamérisation is reflected in neural crest cell migration to specific arch structures, since it is possible to correlate the origin of mesectodermal cells, superficial ectoderm and that of the cranial nerves specifying transverse levels of the head. These ectomeres are oriented cranially at a 45° angle with respect to the embryonic axis at these early stages of neurogenesis.

Quail and chick eggs from commercial sources were used in these experiments. Microsurgery was performed on embryos at the 3-somite stage (after about 30 to 32 h of incubation in a humidified atmosphere at 38°C for chick and 28 to 30 h for quail eggs).

The microsurgical technique involved the selective removal, by means of a microscalpel, of the ectodermal germ layer from defined areas of the presumptive cephalic territory, including, in some series, the neural fold at the same level (for details, see Fig. 1 and Table 1). Microscalpels are made by sharpening thin needles on an Arkansas stone. An ocular micrometer in the dissecting microscope enables the size of the fragments of neural fold or of ectoderm involved in the experiment to be precisely measured.

Trypsin was not used to separate ectoderm from the underlying mesenchymal cells. However, at the stage considered, only a few of the latter were present in the areas included in the operations since the neural crest cells had not yet migrated.

In all cases, quail embryos were used as donors and chick embryos, from which the corresponding ectodermal area had been removed, as recipients. The operated embryos were killed at embryonic day 3.5 to 8 (E3.5-E8), according to the experimental series, and fixed in Zenker’s fluid.

Chimerism was analysed at the head and neck level on 5 μm serial sections stained according to the Feulgen-Rossenbeck method, which allows quail cells to be distinguished from chick cells (Le Douarin, 1969, 1973).

The neural fold was divided into anteroposterior segments 150 μm long. Zones A, B and C (Fig. la) were investigated previously in 0to 3-somite stage embryos (Couly and Le Douarin, 1985, 1987). The experiments described in the present article were performed exclusively on 3-somite embryos and involved areas of the head as defined by zones D, E and F in Fig. 1.

In a preliminary experimental series (experiments P_r), rectangular territories of lateral ectoderm perpendicular to the embryonic axis were delineated as indicated in Fig. 2a for levels D and E of the neural fold.

Numerous types of grafts, involving topographically distinct ectodermal areas lateral to areas D, E and F, were carried out. They will not be described in detail except for the two designated in Fig. 2a as P_rI and P_rII.

In experiment P_rI, involving the D level of the neural fold and ectoderm, the territories occupied by quail cells in the ectodermal area were located in a laterofacial region including the otic and posterior maxillomandibular areas (see Fig. 2b). Quail cells invaded the maxillary bud and the mandibular arch with a modest contribution to the second branchial arch derivatives, as seen in Fig. 2c.

In experiment P_rII, transplantation of the ectodermal territory (originating opposite zone E of the neural fold) resulted in the labelling of a small laterocervical region caudal to that labelled in experiment P_rI (Fig. 2b). Neural crest cells migrated only to the 2nd branchial arch (Fig. 2d). Consequently, there is an overlap of neural crest cell migration from zones D and E of the neural fold mainly in the rostral region of the second branchial arch. Similar results were obtained in experiments performed at later developmental stages by Le Lièvre and Le Douarin (1975). In these preliminary experiments, the dispersion of the grafted crest cells did not correlate with the distribution of quail cells from the epidermal area.

In the following series of experiments, we aimed at mapping the epidermis covering the maxillary buds and the branchial arches. This led us to identify strips of ectoderm superimposable on neural crest-derived mesenchymal cells carrying the quail marker.

The appropriate angle of the lateral ectodermal strips was found, on the basis of the results provided by several transplants of different orientations, to be about 45° with respect to the embryo axis (Fig. lb).

In the following study, two series of experiments, including or excluding the neural fold, were performed (Table 1). The embryos that were studied for chimerism in serial sections showed a normal morphology. Embryos with evident malformations were discarded. The number of embryos analyzed in each experiment is indicated in Table 1. The results described below were strikingly similar for all samples within each series, showing the reproducibility of the experimental design.

In this experiment, the neural fold corresponding to the prosencephalic (zones A,B,C in the experiments described by Couly and Le Douarin, 1985, 1987) and mesencephalic regions (D) were involved in the operation along with the corresponding lateral and rostroventral superficial ectoderm (Table 1).

Four embryos were killed at E6 (n=2) and E8 (n=2). The grafted quail cells were distributed within the superficial ectoderm of the right half of the head (Fig. 3a-c). In detail, regions affected by the labelling were the epidermis covering the calvarium, the nasofrontal area, the cornea, the lens, the upper beak as well as the ectoderm of the right nasal cavity along with the right olfactory placode and the right palate. Moreover, grafted ectoderm was also found covering the complex formed by the inferior beak and all but the proximal part of the tongue and the larynx.

Mesenchymal tissues derived from prosencephalic and mesencephalic neural crest were found evenly distributed underneath the ectoderm.

We already know from our previous experiments that the anterior neural fold of zones A, B and C contributes extensively to the epidermis of the nasofrontal ectoderm, which includes the nasal mucosa and the olfactory epithelium, the ectoderm of the upper beak and the ectoderm covering the premaxillary bones and the nasal septum. Therefore, it is not surprising that the lateral cephalic ectoderm of the 3-somite embryo forehead gives rise only to the ectoderm covering the secondary palate, the lateral wall and roof of the mouth and the ectoderm covering the mobile portion of the tongue.

This experiment investigated the precise origin of the ectoderm contributing to the floor and roof of the oral cavity and of the tongue and the position of the trigeminal placode (Fig. 4a-e).

The neural fold (Zone D, about 150 μm long) of the mesencephalic area, along with a strip of ectoderm of the same width, oriented at about 45° with respect to the dorsoventral axis, was replaced in the 3-somite chick embryo by it quail counterpart (Fig. lb). The mesenchyme arising from the mesencephalic neural crest was found in the first arch, in the maxillary and mandibulary processes including the primordium of the tongue and, to a lesser extent, in the second branchial arch (see also Le Lièvre and Le Douarin, 1974,1975 for experiments performed at later stages) (Fig. 4b).

The trigeminal ganglion was entirely of quail origin, (Fig. 4) indicating that the graft included both its neural crest and placodal components (see Narayanan and Narayanan, 1980; Ayer-Le Lièvre and Le Douarin, 1982; D’Amico-Martel and Noden, 1983).

The ectodermal area of graft origin was limited anteriorly as indicated in Fig. 4a, covering maxillary and mandibulary buds.

In experiment IIA, the superposition of mesectodermal and ectodermal cells of graft origin is very marked indicating that mesencephalic crest cells migrate laterally along a posteroanterior vector.

This is the same graft as in experiment IIA, without the neural fold (see Table 1).

This operation yielded embryos in which quail cells were totally absent from mesenchymal structures and from the trigeminal ganglion. Otherwise, as far as the superficial ectoderm is concerned, no differences were visible between these embryos and those of experimental series IIA (Figs 5-9). Some quail cells form mandibulary muscles. Experiments IIB demonstrate that the superficial epithelium covering the tongue is of ectodermal and not of endodermal origin. The tongue epithelium originates from the first branchial arch ectoderm, which, in addition, contributes to most of the oral epithelium. These results also indicate that the presumptive territory of the trigeminal placode is included in the neural fold itself.

The experimental design was similar to that described in experiment II except that it involved the neural fold at the presumptive metencephalic level (Zone E, about 150 um long) (see Fig. lb) together with the adjacent superficial ectoderm.

The mesectoderm arising from zone E of the neural fold was distributed primarily in the 2nd branchial arch with a few quail cells sparsely distributed in the 3rd arch (Figs 10a and lla,c). The geniculate ganglion was entirely of quail origin, indicating that precursors of both its neuronal (from the placode) and non-neuronal (from neural crest) contingents were included in the graft (Fig. lie), as was the case for the trigeminal ganglion in experiment II. Moreover, the cranial part of the auditory placode was of graft origin.

The superficial ectoderm arising from the graft was located caudally to the strip labelled in experiment II and covered the areas of the 2nd branchial arch ventrally (Fig. 10).

Only the superficial ectodermal component described above was found in embryos resulting from experiment IIIB (Figs 12, 13a).

The geniculate ganglion in this series was formed by host (chick) non-neuronal cells, while the neurons were entirely of quail type. Therefore, in contrast to the trigeminal, the placode is situated laterally and externally with respect to the neural fold at this stage (3somite).

The auditory placode and ganglion were partly of quail origin so indicating that they are located more laterally than the trigeminal placode at this early stage of neurogenesis.

The myelencephalic region, extending to the level of somite 3, included both the neural fold (Zone F) and the superficial ectoderm (Fig. lb, Table 1).

The fate of the ectodermal germ layer of this particular area was carefully analyzed for its possible contribution to the glandular derivatives of the 3rd and 4th pharyngeal pouches from which the thymic rudiments and the parathyroid glands originate (Figs 13b, 14).

The epithelial component of the parathyroid and thymus was of chick origin in all 8 embryos observed at E6 and E8, at which time these structures are fully identifiable. The thymic rudiment was composed only of chick cells except for mesenchymal layers surrounding the thymic glandular cord, which were derived from donor neural crest cells (see also Le Douarin and Jotereau, 1975, for experiments done at later developmental stages).

The parathyroid gland was entirely formed of chick cells on the operated side, with a few quail mesenchymal cells between the epithelial cords. The epidermis, of quail origin, formed a dorsolateral strip located at the level of the more caudal part of the neck and of the cervicothoracic junction, reaching ventrally the mediodorsal line.

Migration of the myelencephalic neural crest could be followed in these birds. Crest cells from this level contributed to the dermis in a territory corresponding roughly to the quail ectodermal area.

As expected (Le Douarin and Teillet, 1973), neural crest cells from this level also gave rise to enteric ganglia. The superior-jugular ganglionic complex (of neural crest origin) as well as the petrosal and nodose ganglia, the neurons of which originate from epibranchial placodes and the non-neuronal cells from the neural crest were all made up of quail cells in these birds (Fig. 14).

The structures containing the quail nuclear marker in this experimental series (4 embryos) were the same as in experiment IVA except that no mesectodermal cells of quail origin were present. Moreover, only the neurons and not the non-neuronal cells were of donor type in the petrosal and nodose ganglia.

In this work, we define segmentally distributed strips of ectodermal areas in the cephalic region at the late neurula stage of the avian embryo. These strips correspond to the first, second, third and fourth branchial arches. The ectoderm of these regions covers the following structures in a craniocaudal sequence.

(1) The first branchial arch, yielding the maxillary and the mandibulary buds plus the lateral wall and floor of the mouth and the mobile part of the tongue (experiment II, corresponding to area D of the neural fold) (Fig. 15). This region includes the presumptive territory of the trigeminal placode, which is located at the external aspect of the neural fold itself and extends on the lateral ectodermal area as indicated in Fig. 4.

The region occupied by quail crest cells originating from zone D of the neural fold corresponds to the extension of the region innervated by the trigeminal nerve (D’Amico-Martel and Noden, 1983).

(2) The second branchial arch, which is covered by epidermis arising from the ectoderm indicated as zone E in Figs 10-13 (experiment III).

(3) The third and fourth arches are included in experiments in which ectoderm corresponding to zone F of the neural fold is transplanted. Considering that this involves an embryonic area equivalent in surface to that of zones E and D, the contribution to superficial ectoderm is smaller. The posterior cervicolateral region of the epidermis is labelled, as is a small area of ventrolateral skin in the thoracic region.

As could be expected (D’Amico-Martel and Noden, 1983; Le Douarin et al. 1986), the territories of the 1st and 2nd epibranchial placodes are lateral and quite distinct from the neural fold at that stage.

The ectodermal strips that we have defined correspond strikingly well with the spatial distribution of neural crest cells arising from the corresponding levels of the neural folds. Moreover, these antero-posterior ‘segments’ of superficial ectoderm parallel the discrete métamérisation of the hindbrain described by pioneering embryological workers as rhombomeres (Orr, 1887; Neal, 1918; Vaage, 1969) and further documented recently (Lumsden and Keynes, 1989; Wilkinson et al. 1989). Fig. 16 indicates the correspondence that can be established between our ectodermal segments and the rhombomeres (see Vaage, 1969). This is of particular interest since the neuromeric métamérisation of the hindbrain seems to be genetically determined, as suggested by the restricted expression of genes of the Hox-2 cluster in well-defined segments of the embryonic mouse brain correlating with the previously defined rhombomeres and to the corresponding neural crest derivatives (i.e., cranial nerve ganglia, Holland and Hogan, 1988; Wilkinson et al. 1989).

It therefore appears that not only is the neural crest patterned according to its rhombomeric origin (Lumdsen and Keynes, 1989; Wilkinson et al. 1989) but that, in fact, the entire ectodermal layer at this level of the body, including the presumptive epidermis may form a genetically defined developmental unit. It is important to remember that cephalic neural crest cells migrate superficially as a multisheet of cells underneath the ectoderm (Johnson, 1966; Noden, 1975; Thiery et al. 1982). Interestingly, it has already been suggested that superficial ectoderm may provide directional guidance for crest migration (Lôfberg et al. 1985).

The neural crest cells originating from levels D, E and F migrate into well-defined facial and hypobranchial structures (i.e., the maxillary buds and the branchial arches) and not only form the skeleton and dermis of these regions (Johnston, 1966; Johnston et al. 1974; Johnston and Hazelton, 1972; Johnston and Listgarten, 1972; Le Lièvre and Le Douarin, 1975) but also provide their sensory and motor innervation (V, VII, IX, X). This is why we propose the term of ‘ectomere’ for the bands of ectoderm described above, whose orientations with respect to the embryonic axis correspond to a roughly 45° caudocranial angle at the late neurula stage (precisely at the 3-somite stage). This does not imply in our view that the superficial ectoderm constituting these ectomeres is committed to a specific spatial fate in the embryos. Such a possibility however should not be ruled out since it has been extensively documented that chondrogenesis of the Vertebrate skull is elicited by epitheliomesenchymal interactions (see Hall, 1988; Thorogood, 1988 for reviews). Whether the patterning of the facial and hypobranchial skeletal structures is specified by the neural crest cells, as suggested by Noden (1983), by the ectoderm or by both is now under investigation in our laboratory. Namely, experiments in which heterotopic grafting of the presumptive ectomeres are being performed to answer this question.

Finally, the work reported in our previous articles (Couly and Le Douarin, 1985, 1987; Le Douarin et al. 1986), together with the results described here, allow us to visualize the prospective fates of different regions of the superficial head ectoderm at the neurula stage (early somitic embryo). The anterior neural fold provides the adenohypophysis medially and the epithelium of the olfactory organ laterally (internal nasal ectoderm including the sensory olfactory placode) plus a large area of epidermis covering the anterior cephalic region (roughly corresponding to the area occupied by the telencephalon) the nose, the upper beak and primary palate.

Posteriorly, the neural fold cells located caudad to the epiphysis anlage (region C in Couly and Le Douarin, 1987) are endowed with migratory properties and become the neural crest, which later contributes to the skeleton and dermis of the face and the ventral aspect of the neck.

Our results, based on experiments carried out at an earlier developmental stage than those described previously (Johnston, 1966; Le Lièvre and Le Douarin, 1975), confirm the neural crest cell distribution but give a more detailed picture of the fate of different presumptive cephalic regions. In particular, we confirm that the nasofrontal neural crest arises principally from that part of the neural fold corresponding to the prosencephalon (in fact, the posterior region of the diencephalon).

In the construction of the cephalic region of the body, another important component is the mesoderm. Métamérisation of head mesoderm has been documented by Meier (1981) and Anderson and Meier (1981). It seems that at the early 3-somite stage studied by Meier (1981) the putative somitomeres that he describes at the rhombencephalic level (somitomeres 4 to 7) correspond roughly to the ectomeres that we have defined (Fig. 16).

It would be interesting to perform chimerism experiments on the cephalic mesoderm to collect more detailed information on these points. A comprehensive picture of the developmental dynamics of head morphogenesis will then emerge.

The authors are grateful to Drs Julian Smith and Evan Balaban for their critical reading of the manuscript. They thank Monique Le Thierry for her skilful technical assistance, Evelyne Bourson for preparing the manuscript, Sophie Gournet for the illustrations and Yann Rantier and Bernard Henri for the photographs.

Financial support was obtained from the Centre National de la Recherche Scientifique, the Association pour la Recherche contre le Cancer, the Fondation pour la Recherche Médicale Française, the Ligue Nationale Française contre le Cancer, the Institut National de la Santé et de la Recherche Médicale and by Basic Research Grant no. 1-866 from March of Dimes Birth Defect Foundation.