Common precursors for neural and mesectodermal derivatives in the cephalic neural crest

The neural crest (NC) of vertebrate embryos is a transitory structure that arises from the joining neural folds as the neural tube closes. All along the neuraxis, it gives rise to cell types as diverse as neurons and glial cells of the peripheral nervous system, melanocytes and neuroendocrine cells (see Le Douarin, 1982 for a review). In addition to these phenotypes, the head NC (but not the trunk NC in higher vertebrates) yields mesectoderm, from which the craniofacial mesenchyme including cartilage, bone, muscles and connective tissue develop (Hammond and Yntema, 1964; Le Lièvre and Le Douarin, 1975; Le Douarin, 1982 for a review).

On the basis of experiments involving heterotopic transplantations of fragments of avian cephalic NC, it has been previously proposed that the mesectodermal lineage could segregate early from the neural, melanocyte and endocrine cell lineages (Le Douarin and Teillet, 1974). The problem arises as to whether mesectodermal precursors are generated early in the neurectoderm or whether they diverge later during cephalic NC ontogenesis. Recent studies have shown that both the trunk and cephalic NC of the avian embryo contain, besides a few fully committed precursors (Cohen and Königsberg, 1975; Ziller et al. 1983; Baraid, 1988; Sieber-Blum, 1989), a large majority of pluripotent cells able to give rise to a diversified progeny including neurons, glia and pigment cells (Bronner-Fraser and Fraser, 1988, 1989; Baroffio et al 1988; Sieber-Blum, 1989; Dupin et al. 1990). We have recently developed an experimental approach in order to clarify the developmental potentialities of quail migrating cephalic NC cells. This issue can only be achieved in vitro by using a technology ensuring that a single cell is seeded and by cultivating these cells in a highly permissive and constant environment (Baroffio et al. 1988). This led us to search for highly multipotent progenitors in the cephalic NC; all types of derivatives including mesectodermal cells could anse from these progenitors The alternative would be that mesectodermal precursors segregate early in the ectodermal germ layer from those able to give rise to neural denvatives and melanocytes.

We show here that the cephalic NC of quail embryos contains cells which, when grown as single cells in culture, generate colonies containing mesectodermal derivatives associated with neuronal and glial cell types. In addition, we provide evidence for a multipotent progenitor giving rise to melanocytes in addition to these cell types, suggesting the existence of a NC stem cell.

The migrating cephalic NC of a 9- to 13-somite quail embryo were bilaterally dissected out and dissociated to single cells according to a procedure described elsewhere (Baroffio et al. 1988). Individual cells were aspirated from this suspension with a micropipet and under microscopic control (Olympus CK2. ×100 magnification). This procedure ensured that a single cell was seeded per culture. Each cell was then placed on a feeder layer of growth-inhibited mouse Swiss 3T3 cells in a growth medium containing fetal calf serum (10%), chick embryo extract (2 %), hormones and growth factors (Baroffio et al 1988) Cultures were fixed after 13 to 16 days in 4 % paraformaldehyde in phosphate buffer and stained with Hoechst bisbenzimide (Serva) in order to differentiate quail and mouse cell nuclei.

A first immunocytochemical labelling was aimed at detecting Schwann cells using the anti-Schwann cell myelin protein (SMP) monoclonal antibody (mAb) (Dulac et al 1988) (ascites fluid diluted 1:500) followed by a fluorescein-coupled goat anti-mouse Ig (GAM-FITC) (Cappel; dilution 1·50). After observation by epifluorescence, the cultures were permeabilised with 0·25% Tnton X-100 for 20 min, and immunostained with a polyclonal antibody against tyrosine hydroxylase (TOH) (Eugene Tech, 1:120) recognised by a rhodamine-coupled (TRITC) goat anti-rabbit Ig (Nordic; 1·50) to detect adrenergic cells Finally a third immunolabelling was done with a mixture of mAbs against the three neurofilament proteins (NF) (Immunotech, 1:50) followed by GAM-FITC, in order to identify the neurons, the fibers of which displayed a pattern unambiguously different from the SMP-labelled cells. After observation by epifluorescence, only those cultures that were devoid of neurons and adrenergic cells, or that contained either no or a few Schwann cells could be stained with the HNK1 mAb (Abo and Balch, 1981, Tucker et al. 1984) followed by GAM-TRTTC in order to reveal ‘non-Schwann’ glial cells (i.e. cells with the following phenotype: NF^-TOH^-SMP-HNK1⁺: Dupin et al 1990). Melanocytes were recognised by their own marker, melanin.

Because of the three immunocytochemical labellings that were necessary to identify neurons, adrenergic cells and Schwann cells (see above), cartilage was the mesectodermal derivative that we chose to monitor in the clones, since it does not need immunocytochemical techniques in order to be identified. After observation by phase contrast, the identity of cartilage was routinely confirmed with an histochemical reaction Thus, following immunocytochemistry, cultures were stained for 15 min with a solution of toluidine blue, pH 4·0, after which cartilage appeared pink; the other cells in the culture remained colourless In certain cases, we confirmed the presence of cartilage in the cultures by using a rabbit antiserum against chick-cartilage-specific proteoglycan (gift of Dr Pacifici; data not shown).

The aim of this investigation was to see whether cells able to produce both neural and mesectodermal derivatives exist in the cephalic NC of the quail embryo. This was performed by cultivating single quail cephalic NC cells in identical conditions and analyzing the cell types differentiating in the clones obtained.

In a previous series of experiments (Baroffio et al. 1988) in which single quail mesencephalic NC cells were grown clonally in vitro for 10 days, 7 of the 173 colonies obtained contained islets of cartilage cells. Cartilage was not found associated with neurons, but, in some cases, appeared together with non-neuronal cells. These cells possessed an epitope recognised by the mAb HNK1, which is mainly (but not exclusively) expressed by NC cells and the majority of their neural derivatives (Abo and Balch, 1981; Tucker et al. 1984). This suggested that pluripotent precursors endowed with both mesectodermal and neural potentialities could exist in the crest.

In order to confirm this finding, we prepared and analysed a larger series of clonal cultures and took advantage of the specific marker for Schwann cells, SMP (Dulac et al. 1988). In these clones, we looked for the presence of cartilage, melanocytes, NF⁺-neurons, TOH-containing adrenergic cells, SMP-immuno-reactive Schwann cells and in some cases HNK1⁺ SMP^- non-neuronal cells. From 834 single cells put into culture, we obtained 305 clones (37% clonal efficiency). The size of these clones varied from 20 to 50000 cells. Various combinations of the differentiated cell types were found, similar to those that have been previously published (Baroffio et al. 1988; Dupin et al. 1990). 10 out of the 305 clones (3·3%) were found to include at least one islet of cartilage (Table 1). They were of variable size, counting from a few cells to large masses of several hundreds of cells. Among these ten clones, two contained no other identifiable derivative (clones 1 and 2). Each of the remaining eight clones(2.6%) included at least one neural derivative. Six relatively small clones were composed, in addition to cartilage, of non-neuronal cells, i.e. of HNK1⁺ presumably glial satellite cells in clone 3, of Schwann cells in clone 4 and of both types of glial cells in clones 5, 6, 7 and 8. In the two larger remaining clones, neuronal as well as glial cells were identified. Clone 9 included Schwann, neuronal and adrenergic cells and clone 10, Schwann, neuronal, adrenergic and pigment cells. A large part of the latter clone is presented in Fig. 1A, showing numerous small NC cells that have grown mainly around the large 3T3 cells, sizeable formations of cartilage stained with toluidine blue as well as a black melanocyte (shown at higher magnification in Fig. 1C). Cartilage nodules were also recognizable before toluidine blue staining as aggregates of large rounded cells (Fig. IB). Successive immunocytochemical labellings performed on this clone revealed the presence of TOH-containing adrenergic cells (Fig. ID), of NF-immunoreactive fibers (Fig. IE) and of SMP-positive Schwann cells (Fig. IF). Since this clone was undoubtedly obtained from a single cephalic NC cell, it identifies a posteriori a highly multipotent progenitor.

Since cell types as different as neurons, glia, melanocytes and cartilage develop from a common structure, i.e. the NC, it can be questioned as to whether they arise from common progenitors, or whether each lineage has its own precursor. The analysis of the progeny yielded by single NC cells has recently shown that the neuronal, ghal and pigmented cell types can differentiate from the same progenitor (Bronner-Fraser and Fraser, 1988, 1989; Baroffio et al. 1988, SieberBlum, 1989; Dupin et al. 1990).

The present work shows that a few clones (2.6%) obtained from cephalic NC cells contain both neural (glia and/or neurons and adrenergic cells) and mesectodermal (cartilage) derivatives. This is the first demonstration that neurons, glial and mesenchymal cells can originate from the same precursor in a vertebrate embryo at the postgastrulation stage. This is a significant finding since it shows that, in the avian embryo, the potentialities to produce both ectodermal and mesenchymal cells are not completely segregated during the gastrulation process. Gastrulation results in the sorting out from the primitive ectoblast, of cells constituting the mesoderm and the definitive endoderm on the one hand from the so-called definitive ectoderm on the other hand. In a recent study, it was shown that presumptive endomesodermal cells are readily identifiable prior to their invagination through the primitive streak and that their immunoablation results in the failure of formation of the endodermal and mesodermal germ layers (Stem and Canning, 1990). However, this experiment did not allow them to determine whether the development of the head mesectoderm was also impaired by the same experimental situation. The data presented here confirm that, until completion of the neurula stage, the ectoderm of the vertebrate avian embryo, and particularly the cephalic neural fold, still contains mesenchymal components destined to join the mesodermal layer to form bones, cartilage and connective tissues of the face. More precisely, we demonstrate that these mesenchymal components can originate from multipotent cells with dual mesectodermal and neural potentialities. The segregation of these two lineages probably occurs during the migration process of NC cells, since the same study reveals that the NC contains clonogenic cells yielding only cartilage (see clones 1 and 2 in Table 1). The frequency of all cartilage progenitors in the cephalic NC revealed by this study is very low (3.3%). The actual proportion of mesectodermal precursors could be higher, since about 20 % of the 305 clones studied included cells of undefined phenotype,i.e. that were negative for all markers and which assumed a fibroblastic morphology compatible with an ectomesenchymal phenotype.

In this study, we found a single clone out of 305 (see clone 10 in Table 1 and Fig. 1) that contained the entire spectrum of phenotypes identifiable by the available markers, i.e. neurons, adrenergic cells, Schwann cells, melanocytes and cartilage. The corresponding founder cell was thus able to give rise to cells belonging to the major NC-derived lineages. Such multipotent cells, which occur at low frequency, are similar to the stem cells found in the hematopoietic system, which are able to give rise to all types of blood cells. The term ‘stem cell’ refers (a) to its totipotency and (b) to its ability for self-renewal through asymmetric divisions generating similar multipotent stem cells and daughter cells committed to differentiation (Lajtha, 1979). This concept, which has now been extended to several other systems (Hall and Watt, 1989), might also apply to the NC. Indeed the presence of highly multipotent cells, although not formally identified as totipotent, suggests the existence of stem cells in the NC. It should be noted, nevertheless, that the self-renewal capacity of putative NC stem cells remains to be demonstrated.

Our previous (Baroffio et al. 1988; Dupin et al. 1990) and present analyses were carried out during a 4h-period of development when NC cells are migrating away from the cephalic neural pnmordium underneath the superficial ectoderm. We analyzed the cell types differentiating in 533 clones obtained from single cells provided with identical culture conditions. Fig. 2 summarizes the various types of progenitors (and their relative frequency) that are ranked according to the number of phenotypic markers expressed in their progeny. Twenty-one types of progenitors were thus identified. In addition to scarce multipotent cells giving rise to cartilage, neurons, adrenergic cells, Schwann cells and melanocytes (see clone 10 in Table 1), the large majority of clonogenic crest cells (80%) were pluripotent cells. They could be organized as 16 classes of progenitors with progeny from 2 to 4 differentiated cell types. Finally 4 types of monopotent precursors generated the remaining 20 % of the clones composed of a single cell type. This demonstrates the presence in the migrating NC of a cascade of precursors with a variable number of potentialities during this developmental window of time. This observation suggests the existence of filiations between these progenitors.

These results agree well with the model that subpopulations of NC cells become rapidly but not synchronously committed to different developmental fates as they migrate and divide (Le Douarin, 1986). A mathematical analysis of the combinations of potentialities intermediate between the multipotent progenitor and the monopotent precursors suggests that these filiations do not proceed in a sequential ordered fashion but that random events contribute to the segregation of cell lineages (Baroffio and Blot, 1991). Interestingly, these observations are similar to those reported for hemopoietic precursors developing in single cell cultures (Suda et al. 1983; Ogawa et al. 1983). This reinforces a possible analogy between the NC and the hematopoietic system (Anderson, 1989).

We propose that, in vivo, the active proliferation of pluripotent NC cells during the migration phase generates a large and heterogeneous population of oligopotent precursor cells which, with time, become progressively restricted to distinct lineages. Environmental factors that have previously been demonstrated to play a decisive role in the fate of NC cells (see Le Douarin, 1982; Le Douarin and Smith, 1988 for a review) have to exert a selection among these differently committed precursors since, in each type of NC derivative, only some of them proceed to a final differentiation state. It is important to note, however, that the results reported here do not rule out the possibility that environmental influences may play a role, during or after completion of the migration, in the commitment of NC cells. If it is true that, at least for some cells, restrictions of potentialities already take place when they are migrating (Cohen and Königsberg, 1975; Ziller et al. 1983; Baraid, 1988; Sieber-Blum, 1989), many others are still pluripotent (Bronner-Fraser and Fraser, 1988, 1989; Baroffio et al. 1988; SieberBlum, 1989; Dupin et al. 1990) and may retain multiple potentialities when they arrive at the sites of gangliogenesis. In particular, the pluripotent cells that are still present in the migrating NC might be the target of factors that would be able to drive them into a particular differentiation pathway.

In conclusion, we demonstrated the presence of common precursors for component cells of both the peripheral nervous system and the mesectodermal derivatives in the cephalic NC by using clonal cultures. The existence of multipotent cells from which the major NC-derived cell types differentiate, suggests the presence of stem cells in the NC. In addition, the whole range of progenitors analysed in this study shows that many NC cells are still pluripotent during the migration phase whereas others have already undergone restrictions in their developmental potentialities. This implies that NC cells do not reach the sites of differentiation in an identical state of commitment. As a consequence, the role of the microenvironment at the sites in which differentiation occurs may be in directing multipotent progenitors towards a specific differentiation pathway as well as selectively promoting the survival of appropriate subsets of fully determined NC cells.

We thank Drs C. R. Bader, A. Kato, J. Smith and E. Tribollet for their comments on the manuscript, B. Henri and G. vanKaenel for preparing the illustrations and G. Gateau for technical assistance.

This work was supported by the Centre National de la Recherche Scientifique, the Fondation pour la Recherche Médicale Française, the Association pour la Recherche contre le Cancer and the Fonds National Suisse pour la Recherche Scientifique.