We show that mouse neural crest cells cultured in a serum-deprived chemically defined medium on appropriate culture substrata can be induced to express a neuronal phenotype. The uncommitted neural crest cells express a mesenchymal intermediate filament protein such as vimentin, but not the usual neuronal markers such as receptor sites for tetanus toxin or neurofilaments. In the chemically defined medium, receptor sites for tetanus toxin or neurofilaments can be characterized after a few hours in culture. Furthermore, these cells acquire tetrodotoxin-sensitive voltage-dependent Na+ channels and can generate action potentials. Such an in vitro system should allow us to analyze and manipulate early stages of neuronal differentiation in a mammalian embryo, at a level so far restricted to lower vertebrate embryos.

The ontogeny of the mammalian nervous system remains largely unknown, mainly because of the lack of experimental systems that will allow the isolation of identified progenitor cells and their in vitro manipulation into neuronal derivatives. Such systems have been of great value for the analysis of lineages for cells as diverse as haemopoietic (Metcalf & Moore, 1971) and glial progenitors (Raff et al. 1983b; Temple & Raff, 1986).

In spite of considerable advances towards understanding lineages derived from neural crest cells using amphibian (Holtfreter, 1968) and avian embryos (Le Douarin, 1982; 1986), little is known of the mechanisms of neuronal differentiation either at a cellular or at a molecular level.

As a variety of molecular genetic strategies are now available with the mouse to study embryogenesis (Hogan et al. 1986; Rossant & Pederson, 1986; Jaenisch, 1988), we chose to develop an in vitro model suitable for the study of early steps of neuronal differentiation from mouse identified progenitor cells.

From the experimental work on amphibian and avian embryos (Le Douarin, 1986), neural crest cells represent a unique system with which to approach this question as these progenitor cells give rise to a wide variety of cell types, ranging from sensory, autonomic ganglia neurons and Schwann cells to facial bone and cartilaginous structures, stroma of various secretory glands and melanocytes. In vitro differentiation of avian neural crest cells using defined culture medium was observed recently (Ziller et al. 1983; Ziller, 1986) and makes it possible to follow the transformation of undifferentiated neural crest cells into derivatives with different phenotypes. However, analogous experiments with mammalian embryos have not yet been successful.

In the present study, we addressed two specific questions: (i) is an early expression of a neuronal phenotype obtainable using an in vitro differentiation system from mouse neural crest cells?; (ii) is this phenotype comparable to that obtained in vivo?

We report here that mouse neural crest cells cultured in a fully defined medium, containing hormones, growth factors and transferrin (Basic Brazeau Medium, BBM) (Ziller et al. 1983), can be differentiated into cells expressing a neuronal phenotype after a few hours in culture. Under these conditions, these cells develop neuronal outgrowths, express tetanus-toxin-binding sites and neurofilament proteins. Furthermore, electrophysiological recordings show that those cells that have a neuronal morphology are able to generate action potentials resulting from a voltage-dependent activation of tetrodotoxin-sensitive Na+ channels.

Tissue culture methods

Preparation of neural tube segments

Swiss mouse (Iffa Credo, Lyon, France) embryos were isolated at day 9 of gestation (day of vaginal plug = day 0 of gestation). The embryos were freed of placenta and fetal membranes using fine forceps under a dissecting microscope. They were incubated for 15 min in PBS medium with 400 μg ml-1 dispase (Neutral protease; EC 3.4.24.4; Boehringer Mannheim) added. The enzyme action was stopped by transferring the embryos to phosphate-buffered saline (PBS; Gibco) supplemented with 10% fetal calf serum (FCS; Boehringer). Using fine tungsten needles, somites and surrounding tissues were removed to obtain neural tube segments of approximately 1 to 2 mm, as described in Ito & Takeuchi (1984) (Fig. 1A). If not otherwise indicated, these segments were washed in a serum-free medium and transferred with a sterile Pasteur pipette to fibronectin-coated Primaria (Falcon) culture dishes.

Fig. 1.

Primary neural tube culture technique. (A) Trunk fragments were isolated from E9 embryos as indicated in Methods, incubated in presence of dispase, and microdissected to separate neural tube (NT) from ectoderm (Ec), somites (Som), endoderm (End) and chord. Crest cells migrate from the neural tube as shown on phase-contrast photomicrographs of an expiant at 6 h (B) and 48 h (C) after explantation. Culture was done in BBM on a fibronectin-coated culture dish. Calibration bar: 200 μm.

Fig. 1.

Primary neural tube culture technique. (A) Trunk fragments were isolated from E9 embryos as indicated in Methods, incubated in presence of dispase, and microdissected to separate neural tube (NT) from ectoderm (Ec), somites (Som), endoderm (End) and chord. Crest cells migrate from the neural tube as shown on phase-contrast photomicrographs of an expiant at 6 h (B) and 48 h (C) after explantation. Culture was done in BBM on a fibronectin-coated culture dish. Calibration bar: 200 μm.

Removal of neural tube

Using sterile conditions, the neural tube was separated from the culture, 24 h after explantation, under microscope, with a fine glass pipette needle and removed with a 10 μ1 cone of a Gibson Pipetman.

Culture substrata

Cells were cultured either on 35 mm Primaria culture dishes (Falcon) or 24 mm glass coverslips (Proscience) for immunostaining experiments. In order to improve adhesion and migration of neural crest cells, we tested various substrata: type I collagen (rat tail collagen, Sigma type VII; 2·5 mg ml-1 collagen in 1 % acetic acid in distilled water), fibronectin (Jacques Boy Institut, 10 μg ml1) and laminin (Collaborative Research; 20 μgm11).

Culture media

The defined medium used was that described in Ziller et al. (1983) as Basic Brazeau Medium (BBM). It consists of Ham’s F-12 nutrient mixture, DMEM, BGjb Fitton-Jackson modification (6:3:1; Gibco), 2 g I-1 bovine serum albumin (BSA; Sigma), 0·0lM-Hepes buffer (Gibco), 100i.u.ml-1 streptopenicillin (Gibco), 100 μg 1—1 hydrocortisone (Sigma), 1 μg11 insulin (Sigma), 0·4 μg 1-1 tri-iodotyronine (Sigma), 10 ng I-1 glucagon (Sigma), 0·2 μg11 parathyroid hormone (Sigma), 10mgl-1 transferrin (Sigma), 0·lμgl-1 epithelial growth factor (EGF; Collaborative Research), 0·2 μg1-1 fibroblast growth factor (FGF; Collaborative Research). In some experiments, we used BBM with 15% FCS. About five expiants were pooled in a culture dish and incubated at 37 °C in 5% CO2.

When indicated, Nerve Growth Factor (NGF; 7S; Collaborative Research) was added at a final concentration of 10 μg ml-1.

DNA synthesis studies

[Methyl-3H]thymidine (3H-TdR; lμCiml-1; spec. act. 25Cimm-1; Amersham) was added for 24 h to the culture medium at 24 h after explantation. The cells were fixed at 48 h using 4% paraformaldehyde, washed twice in PBS and in distilled water. A 1:1 solution of Kodak NTB2 autoradiographic emulsion in distilled water was added to culture dishes and exposed for 8 days. Preparations were then revealed for 2 min in Kodak Microdol-X and fixed in Kodak Unifix. After several washes in distilled water, cells were stained for 1 min with a solution of hematoxylin-eosin and mounted in PBS-Glycerol (1:1) medium. Cells considered as labeled were cells that incorporated at least 10 grains per nucleus.

Immunocytochemical methods

All the cultures used for immunostaining were done on human fibronectin-coated glass coverslips.

Characterization of intermediate filament proteins Antibodies

To detect the presence of immunoreactivity to neurofilaments, we used a mixture of mouse monoclonal antibodies against neurofilament proteins of 70, 160 and 200x103Mr (a gift from Dr D. Paulin, Institut Pasteur; 1:10) visualized with a goat anti-mouse Ig conjugated with tetramethylrhodamine-isothiocyanate (TRITC, Nordic; 1:100).

The 70 and 160 X103Mr neurofilament proteins were separately detected using either, respectively, a monoclonal antibody against the 70x103Mr protein (mouse IgGi; Boehringer; 1:20) or a monoclonal antibody against the 160 X103Mr protein (mouse IgG1; Boehringer; 1:20). In both cases, second antibody was a goat anti-mouse lgG1 conjugated with fluorescein-isothiocyanate (FTTC, Nordic; 1:100).

The detection of vimentin was done using a rabbit polyclonal antibody against vimentin (a gift from Dr A. M. Hill, Université d’Orsay; 1:500). The second antibody was a goat anti-rabbit IgG conjugated either to FITC or TRITC (Nordic; 1:100).

Indirect immunofluorescence microscopy

After fixation with methanol at —20°C for 10 min and three washes in PBS, the cells were incubated with 25 μl of the first antibody for 1 h at room temperature, in a humidified atmosphere, washed three times with PBS, incubated with 25 μl of the appropriate second antibody for 45 min at room temperature, washed again in PBS, mounted in Mowiol 4-88 as in Ziller et al. (1983) and observed with a fluorescence Nikon Diaphot TMD microscope.

Characterization of tetanus toxin receptors

Tetanus toxin receptors were visualized as described in Mirsky et al. (1978). lie tetanus toxin fragment and antitetanic serum were donated by Dr B. Bizzini, Institut Pasteur. Living cultured cells were incubated with 25 μl of Ilc-tetanus toxin (lμg ml-1) for 20min at 4°C, washed with PBS and fixed with 4 % paraformaldehyde for 10 min at room temperature. After a wash with NH4CI, 0·5 mm in Tyrode’s solution, cells were incubated with rabbit anti-tetanic serum (25 μ1; 1:100) overnight at 4 °C and washed with PBS and 25 μ1 of rhodamine-coupled goat anti-rabbit IgG (Nordic, 1:50) was applied for 45 min at room temperature. Coverslips were added and specimens were observed as above.

Electrophysiological studies

Electrophysiological studies were made by the patch-clamp method (Hamill et al. 1981). For these studies, cells were bathed in a recording solution containing (in mm): NaCl 155, KC1 5·5, CaCl2 2·5, MgCl2 1, 4-(2-hydroxyethyl)-l-piperazineethanesulphonic acid (Hepes)-NaOH 10, pH 7·4. Patch pipettes (2-10 MΩ resistance) were filled with (mm) KC1 140, MgCl2 2, N-tris(hydroxymethyl) methyl-2-aminoethanesulphonic acid (TES) buffer 2, pH7·4 with 8·8 μM of CaCl2 and 80 μM of EGTA to have 10−8M as a free calcium concentration. Temperature, 20-22°C.

Choice of culture conditions allowing an early neuronal differentiation

Choice of a substratum appropriate for crest cell adhesion and migration

After one or two hours in culture in fibronectin-coated dishes in BBM medium, crest cells can be identified along the neural tube. Fig. 1 shows crest cells migrating out of a neural tube segment (B) and 48 h (C) after its transfer in culture.

Adhesion and migration of neural crest cells appeared to be dependent on interactions with attachment proteins. These properties have been tested in serumfree conditions as fetal calf serum contains such attachment proteins, i.e. fibronectin. Migration did not occur on collagen-coated culture dishes but was obtained using chemically treated plastic (Primaria dishes, Falcon) and with laminin and fibronectin substrata (Fig. 2). Fibronectin attachment protein appeared to give the best adhesion and migration of mouse neural crest cells, confirming results chained for avian crest cells (Rovasio et al. 1983). This fibronectin coating was used for all other experiments described in this paper.

Fig. 2.

Adhesion and migration of neural crest cells on different extracellular matrix components. Phase-contrast photomicrograph of expiants cultured for 48 h in BBM medium, without fetal calf serum. Fibronectin was the best substratum for migration as illustrated in A. Treated plastic (Primaria; Falcon) (B) as laminin coating (not shown) permitted migration but to a lesser extent compared with fibronectin substratum (A). In contrast, no migration occurred on type I collagen (C). Calibration bar: 200 μm.

Fig. 2.

Adhesion and migration of neural crest cells on different extracellular matrix components. Phase-contrast photomicrograph of expiants cultured for 48 h in BBM medium, without fetal calf serum. Fibronectin was the best substratum for migration as illustrated in A. Treated plastic (Primaria; Falcon) (B) as laminin coating (not shown) permitted migration but to a lesser extent compared with fibronectin substratum (A). In contrast, no migration occurred on type I collagen (C). Calibration bar: 200 μm.

Choice of a chemically defined medium

Experiments carried out with a variety of cell lines and primary cultures (Bottenstein & Sato, 1979; Darmon et al. 1981; Pfeiffer et al. 1981; Ziller et al. 1983) showed that neuronal differentiation can be elicited by removing serum from the culture medium. When cultured in serum-containing medium, the neural crest cells did not exhibit a morphologically identifiable neuronal phenotype. Furthermore, under these conditions, no binding was found for antibodies against the tetanus toxin receptor or neurofilaments. Chemically defined media developed by Bottenstein & Sato (1979), containing insulin, transferrin, selenium and attachment factors, which have been used to provoke differentiation of embryonal carcinoma cells into neurons (Darmon et al. 1981; Simonneau et al. 1985a), did not induce neuronal differentiation of neural crest cells. In contrast, when cultured in the fully chemically defined BBM medium, mouse neural crest cells differentiated into neurons, based on immunological and electrophysiological criteria.

Characteristics of the early neuronal phenotype

The neuronal differentiation obtained in BBM medium was assessed using morphologic criteria and using various markers (neurofilaments, tetanus toxin receptors, Na+ channel expression). As early as 24h of culture in BBM medium, process-bearing cells presenting tetanus toxin receptors can be readily identified using polyclonal antibodies against tetanus toxin. However, a number of observations using tetanus toxin antibodies indicate that tetanus toxin receptors are not specific for neurons but are found also in glial cells (Raff et al. 1983a,b). To determine the characteristics of the early neuronal phenotype, we analyzed the expression of the other markers.

Coexpression of vimentin and neurofilaments

Vimentin (Vim), an intermediate filament of 55 to 58x103,Mr (Bignami et al. 1982), has been observed in vivo at the level of neural crest migration pathways in E9 mouse embryos (Cochard & Paulin, 1984). In the presence of FCS in the culture medium, antibodies directed against Vim labelled all the cells presenting a neural crest cell morphology but no neurofilaments (NF) were identified using monoclonal antibodies against the three NF proteins (Fig. 3A,B). When the cells were cultured in BBM, neurofilament-positive (NF+) cells can be visualized. Furthermore, NF+ cells coexpressed vimentin as shown using double immunostaining (Fig. 3C,D). The appearance of Vim+ cells in serum-containing medium is drastically different from that of Vim+ cells in BBM: in the presence of serum, most cells are flat and polygonal; in the chemically defined medium, cells are smaller and start to extend processes. At 72 h in BBM medium, some cells presented neurites that were labelled by NF antibodies but not by Vim antibodies, with a residual Vim immunostaining in the soma. It should also be pointed out that Ziller et al. (1983) found a similar coexpression with the same antibody markers, suggesting a strong similarity between avian and mammalian systems. However, no staining was obtained with A2B5 monoclonal antibody, which is known to label an early neuronal population obtained in serum-containing medium from avian neural crest cells (Girdlestone & Weston, 1985; Vogel & Weston, 1988; Weston et al. 1988).

Fig. 3.

Expression of intermediate filaments. Simultaneous detection of vimentin (Vim) (A,C) and neurofilaments (NF) (B,D) at 48 h in culture in BBM added with 15% FCS (A,B) and 48 h in BBM (C,D). Indirect immunofluorescent staining with (i) fluorescein isothiocyanate (FITC)-conjugated second antibodies to detect Vim immunoreactivity and with (ii) tetramethylrhodamine isothiocyanate (TRITC)-conjugated second antibodies to detect NF immunoreactivity. Calibration bar: 50 μm.

Fig. 3.

Expression of intermediate filaments. Simultaneous detection of vimentin (Vim) (A,C) and neurofilaments (NF) (B,D) at 48 h in culture in BBM added with 15% FCS (A,B) and 48 h in BBM (C,D). Indirect immunofluorescent staining with (i) fluorescein isothiocyanate (FITC)-conjugated second antibodies to detect Vim immunoreactivity and with (ii) tetramethylrhodamine isothiocyanate (TRITC)-conjugated second antibodies to detect NF immunoreactivity. Calibration bar: 50 μm.

Coexpression of 70 and 160×103Mr neurofilament proteins

As observed on sections of E9 mouse embryos by Cochard & Paulin (1984), we found a simultaneous expression of the 70x103Mr (NF-L) and 160 x103Mr (NF-M) neurofilament proteins. A faint NF-L staining was detectable after 24 h of culture. Fig. 4 illustrates the simultaneous expression of NF-L and NF-M after 48 h of culture.

Fig. 4.

Simultaneous expression of NF-L and NF-M at 48 h in defined serum-free medium (Basic Brazeau Medium). Two different preparations of neural crest cells were processed for double-immunostaining experiments (respectively A, B, Cand D, E, F). A & D: phase contrast micrographs. B & E: immunostaining with anti NF-L (B) or anti NF-M (E) and FITC-conjugated second antibody. C & F: immunostaining with the polyclonal anti-Vim and TRITC-conjugated second antibody. Note the faint NF-L staining at 48 h (B) as compared with NF-M staining (E). Calibration bar: 30 μm.

Fig. 4.

Simultaneous expression of NF-L and NF-M at 48 h in defined serum-free medium (Basic Brazeau Medium). Two different preparations of neural crest cells were processed for double-immunostaining experiments (respectively A, B, Cand D, E, F). A & D: phase contrast micrographs. B & E: immunostaining with anti NF-L (B) or anti NF-M (E) and FITC-conjugated second antibody. C & F: immunostaining with the polyclonal anti-Vim and TRITC-conjugated second antibody. Note the faint NF-L staining at 48 h (B) as compared with NF-M staining (E). Calibration bar: 30 μm.

Early expression of voltage-dependent sodium channels

The ionic channel repertoire of these cells was analyzed using patch-clamp techniques (Hamill et al. 1981), both in BBM with FCS added where they remain undifferentiated and in serum-free BBM where they differentiate. In conditions of nondifferentiation, these cells had resting membrane potentials of -50·0±9·8mV (mean±s.D.; n = 15). When a pulse of depolarizing current was applied, no action potential could be recorded. Only outward currents were found under voltage-clamp conditions (Fig. 5). Whole-cell outward currents were attributed to potassium channels according to the values of their reversal potential and their blockade by 10 HIM-TEA, a blocker of voltage-dependent K+ channels. We also examined the electrophysiological properties of cells presenting a neuronal morphology (Fig. 6A) and cultured 24 h in serum-free BBM: 18 out of 23 were excitable. We selected this early stage of differentiation because cells with short neurites are necessary for a good space clamp. Their resting potential (Vr) was similar to that found for nondifferentiated neural crest cells (Vr = -55·7 ± 11·5mV; mean±s.D.; n = 23). In whole-cell voltageclamp, these cells displayed inward currents in addition to the outward currents similar to those found in undifferentiated cells. The inward currents were identified as Na+ currents on the basis of their blockade by 10−6M-tetrodotoxin, a highly selective blocker of voltage-dependent Na+ channels. The peak value of Na+ current was 94 ± 57 pA (m ±S.D.;n = 5). This value can then be compared with the peak Na+ current value found in embryonic neurones such as dorsal root ganglion (DRG) cells, which are derivatives of neural crest. The Na+ peak current in DRG cells isolated from mouse embryos at day 12 was 939 ± 122pA (m ±S.D.;n = 12), indicating a tenfold increase in Na+ current in about two days of mouse embryonic development (Valmier & Simonneau, unpublished results).

Fig. 5.

Ionic currents recorded in a nondifferentiated crest cell. (A) Phase-contrast photomicrograph of uncommitted neural crest cells cultured for 24 h in BBM supplemented with 15 % FCS and similar to those selected for electrophysiological recording. The bar represents 50 μm. (B) Whole-cell voltage-clamp currents corresponding to depolarizing voltage pulses from -90mV holding potential to the indicated values. The current traces were leak-subtracted. These outward currents were blocked by 10 mm-TEA (not shown). Pipette medium and bath solution as described in Methods.

Fig. 5.

Ionic currents recorded in a nondifferentiated crest cell. (A) Phase-contrast photomicrograph of uncommitted neural crest cells cultured for 24 h in BBM supplemented with 15 % FCS and similar to those selected for electrophysiological recording. The bar represents 50 μm. (B) Whole-cell voltage-clamp currents corresponding to depolarizing voltage pulses from -90mV holding potential to the indicated values. The current traces were leak-subtracted. These outward currents were blocked by 10 mm-TEA (not shown). Pipette medium and bath solution as described in Methods.

Fig. 6.

Early expression of voltage-dependent Na+ channels in cells displaying neuronal morphology after 24h of culture in defined medium. (A) Phase-contrast photomicrograph of cells presenting neuronal morphology similar to those selected for patch-clamp recordings; the bar represents 50 urn. Whole-cell clamp recordings: a current clamp is shown in B; the depolarizing and hyperpolarizing current pulses were of 50 pA amplitude. Whole-cell current obtained for the same cell, under voltage-clamp from Vh = —70 mV to -10 mV before (C) and after (D) 10∼* 6M-TTX bath application. Pipette medium and bath solution were as described in Methods.

Fig. 6.

Early expression of voltage-dependent Na+ channels in cells displaying neuronal morphology after 24h of culture in defined medium. (A) Phase-contrast photomicrograph of cells presenting neuronal morphology similar to those selected for patch-clamp recordings; the bar represents 50 urn. Whole-cell clamp recordings: a current clamp is shown in B; the depolarizing and hyperpolarizing current pulses were of 50 pA amplitude. Whole-cell current obtained for the same cell, under voltage-clamp from Vh = —70 mV to -10 mV before (C) and after (D) 10∼* 6M-TTX bath application. Pipette medium and bath solution were as described in Methods.

Quantification of neurofilament positive cells in different experimental conditions

The total number of cells by explant in serum-free defined culture medium was 415 ± 68, 1030 ± 118 and 1298 ±279 (mean ± S.D.), respectively, at 24, 48 and 72 h of culture (calculated for 10 different expiants observed at these three times of culture).

We estimated the number of cells expressing a neuronal phenotype on the basis of NF-M expression. All the cells migrating out of the explant expressed vimentin. We used the ratio of NF-M+ cells: Vim+ cells as an index of neuronal differentiation. The number of NF-M+ cells increases from less than 5% at 24 h to approximately 20 % at 48 h and then reaches a plateau between 48 and 72 h (Fig. 7). We were not able to maintain in culture the differentiated cells after 72 h in a defined medium. Such a short survival of neuronal derivatives in serum-free medium was described for avian neural crest preparations (Ziller et al. 1983).

Fig. 7.

Expression of neurofilaments as a function of time in culture. At each time tested, the cultures were rinsed, fixed and incubated with antibodies for double immunostaining of vimentin (Vim) and 160X103Mr neurofilament protein (NF-M) as indicated in Methods. Vim+ and NF-M+ cells were counted. The results are expressed as a percent of Vim+ cells which are also NF-M+, for BBM (○) and for BBM added with 15% FCS (▴). These data correspond to the mean ± S.D of eight experiments. For each experimental point, at least 1000 cells were examined.

Fig. 7.

Expression of neurofilaments as a function of time in culture. At each time tested, the cultures were rinsed, fixed and incubated with antibodies for double immunostaining of vimentin (Vim) and 160X103Mr neurofilament protein (NF-M) as indicated in Methods. Vim+ and NF-M+ cells were counted. The results are expressed as a percent of Vim+ cells which are also NF-M+, for BBM (○) and for BBM added with 15% FCS (▴). These data correspond to the mean ± S.D of eight experiments. For each experimental point, at least 1000 cells were examined.

Influence of the neural lube on neuronal differentiation

Recent in vivo experiments, in the avian embryo, demonstrated that the developing spinal cord provides some survival and/or differentiation factors (Kalcheim & Le Douarin, 1986).

Using this mammalian in vitro model, it was possible to quantitatively analyze the influence of the neural tube on the number of cells expressing neurofilament proteins. The removal of the neural tube at 24 h of culture did not modify significantly the number of NF-M+ cells stained at 48h of culture. However, the number of these NF-M+ cells decreased from 19 % to approximately 5 % between 48 and 72 h of culture (Fig. 8). These results can be interpreted on the basis of a factor released by the neural tube, not involved in the neuronal differentiation but required for the survival of neural committed precursors.

Fig. 8.

Effect of removal of neural tube on the number of cells expressing NF-M. Neural tube was removed after 24 h of culture, as described in Methods. Cultures were immunostained either at 48 or 72 h. Data were calculated as described in Fig. 7. Each histogram corresponds to the mean ± S.D. of six experiments.

Fig. 8.

Effect of removal of neural tube on the number of cells expressing NF-M. Neural tube was removed after 24 h of culture, as described in Methods. Cultures were immunostained either at 48 or 72 h. Data were calculated as described in Fig. 7. Each histogram corresponds to the mean ± S.D. of six experiments.

This in vitro system may constitute a suitable model to characterize and isolate such a new neuronal growth factor produced by the embryonic spinal cord.

Effect of growth factors on neuronal differentiation

If both EGF and FGF were omitted in the serum-free BBM, no migration of neural crest cells was obtained. Addition of basic FGF at a concentration of 1μgrnl-1 inhibited the neuronal differentiation, on the basis of the absence of NF-M+ immunostaining. This result can be explained by an inhibitory effect on neuronal differentiation which may correspond either: (i) to the death of the precommitted neurons or (ii) to a mitotic effect on uncommitted progenitors, mimicking the addition of serum which inhibits neuronal differentiation (see Fig. 7 black triangles).

As NGF is known to be a growth factor for a variety of neural crest cell derivatives (Levi-Montalcini, 1987), we tested the effect of NGF on the number of cells expressing NF-M in this mammalian neural crest system. Fig. 9 shows that the addition of NGF either at 24 or 48 h after explantation was without any effect on the number of cells expressing neurofilament proteins. Similar negative results were obtained by Ziller et al. (1983) for avian neural crest cells. They are in agreement with the lack of NGF receptors in neural crest cells at this stage of differentiation, as NGF receptors can be detected only after 3 or 4 days of culture of multipotent neural crest cells (Greiner et al. 1986).

Fig. 9.

Lack of effect of NGF on early neuronal differentiation. NGF was added to the culture medium at 24 h (single dashed histograms) or at 48 h (double dashed histogram) of culture. The cultured cells were immunostained either at 48 or 72h. Data were calculated as described in Fig. 7. Each histogram corresponds to the mean ± S.D. of six experiments.

Fig. 9.

Lack of effect of NGF on early neuronal differentiation. NGF was added to the culture medium at 24 h (single dashed histograms) or at 48 h (double dashed histogram) of culture. The cultured cells were immunostained either at 48 or 72h. Data were calculated as described in Fig. 7. Each histogram corresponds to the mean ± S.D. of six experiments.

Estimation of the number of neurogenic precursors

Neural crest cells were incubated with 3H-TdR 24 h after explantation for a pulse of 1 day, fixed and exposed with emulsion at 48h of culture. Approximately 90 % of the flat polygonal cells were labeled. By counting cells with a neuronal morphology that have incorporated 3H-TdR, we estimated that only 10 % of neurons were labeled.

This result indicates that, in the serum-free culture conditions we used, most of the neurogenic precursors give rise to differentiated neurons without division. Taking into account (i) that the number of cells presenting a neuronal phenotype is roughly equivalent to the number of neurogenic precursors and (ii) 200 cells can be identified as NF-M+ at 48 h (Fig. 7) out of 1000 cells (total number of migrating cells estimated for each explant), one can suggest that 200 neurogenic precursors (out of 500 neural crest cells) were present at 24 h.

This remarkably high frequency (40 %) of neurogenic precursors obtained after 24 h in these serum-free culture conditions may permit identification of the phenotype of this subpopulation of progenitors.

The main aim of the present study was to obtain an in vitro model adapted for the analysis of the early steps of mammalian neuronal differentiation from progenitor cells under strictly defined conditions. With this model, it was possible to identify unequivocally neuronal derivatives from mouse neural crest cells using the characterization of neurofilaments and voltage-dependent sodium channel proteins as early as 24 h in chemically defined medium.

The first advantage of this in vitro model is offered by the choice of the mouse as an experimental animal. If avian embryo permitted considerable advances in the knowledge of neural crest cell migration and differentiation (Le Douarin, 1982, 1986), the mouse embryo offers distinct advantages. First of all, a variety of mutant strains presents phenotypes that correspond to neurological defects linked with early steps of neuronal development (Hogan et al. 1986). Second, the ability to obtain transgenic mice by introducing foreign genes into the germ line allows the genetic manipulation of mouse embryogenesis (Jaenisch, 1988). Such an approach has permitted the study of some aspects of neuronal differentiation using transgenic mouse lines expressing various amounts of human neurofilament genes (Julien et al. 1987), or expressing a HTLV-I tat gene as a model for neurofibromatosis (Hinrichs et al. 1987). Furthermore, different strategies using transgenic mice are now available to suppress a given cell lineage (Palmiter et al. 1987; Borelli et al. 1988). Third, mouse neural crest cells can be reintroduced into the postimplantation embryos and participate in the normal embryonic development (Jaenisch, 1985).

The second advantage of such an in vitro model is linked with the possibility of obtaining neuronal derivatives as early as 24 h of culture in a chemically defined medium. Attempts to obtain a neuronal differentiation in the presence of nondefined medium have been unsuccessful. With the experimental culture conditions used by Ito & Takeuchi (1984), which included extracts of chick embryos and fetal calf serum, catecholaminepositive cells could be obtained only after two weeks of culture. But, the use of a chemically defined medium for in vitro culture of neural crest cells allowed the analysis of the phenotype of neurons corresponding to embryonic day 10 which is a stage difficult to approach in vivo. At this stage, truncal neural crest derivatives migrate and differentiate in the embryo but they cannot easily be identified or isolated in vivo as no markers are presently available in the mammals for these early steps of differentiation. Using the culture conditions described in this article, we were able to show, at a single cell level and as early as one day in culture, that cells presenting an early neuronal phenotype coexpressed two types of filaments, namely vimentin and neurofilaments. Data previously obtained using mouse embryo tissue sections (Cochard & Paulin, 1984) were only suggestive of such a coexpression in a single cell. In the absence of specific markers for mammalian neural crest, one can question the neural crest origin of differentiated neurons we obtained in culture. To examine this question, we took advantage of the expression of a type III intermediate filaments which has been demonstrated to be specific for peripheral neurons and in subpopulations of central nervous system having peripheral projections like the cranial nerve ganglia (Portier et al. 1984; Parysek & Goldman, 1987; Leonard et al. 1988; Escurat et al. 1988). By immunohistochemistry, we found a coexpression of neurofilament proteins and of class III intermediate filament when the neural crest cells were cocultured with target cells (like inactivated 3T3 cells, somite cells, embryonic cardiac cells) after four days in the presence of 10 % of fetal calf serum (Boisseau & Simonneau; in preparation). Class III intermediate filaments were present in all neurofilament-positive cells, making it likely that all the differentiated neuronal cells originate from mouse neural crest progenitors. It will be instructive to use this in vitro model to identify and/or to purify factors released by these target cells and involved in early events of neuronal differentiation.

This approach also led to a gain of new information on the sequence of expression of ionic channels during these early events in mammalian neuronal differentiation. Undifferentiated mouse neural crest cells do not present action potentials and expressed only voltage-dependent potassium channels. To our knowledge, this is the first demonstration that mammalian progenitor cells are not excitable. Furthermore, differentiated cells start to express voltage-dependent sodium channels as early as 24h of culture in defined medium. We confirmed this early expression of sodium channels before that of calcium channels by analysing isolated dorsal root ganglion cells which can be microdissected in Ell mouse embryos. For cells examined 2h after microdissection, we found an expression of Na+ channels similar to that found in neurons obtained in vitro from crest cells (Valmier & Simonneau, unpublished results). It is important to note that sodium channels are expressed before calcium channels in this system as different sequences of ionic channel development have been reported from other nonmammalian preparations (Spitzer, 1979; Mori-Okamoto et al. 1983) and from mouse teratocarcinoma cell lines (Simonneau et al. 1985a,b). The early neuronal phenotype reported here in the mouse resembles that previously described for quail mesencephalic neural crest cells (Ziller et al. 1983; Ziller, 1986; Bader et al. 1983, 1985).

In conjunction with the molecular genetic techniques which are so far applicable only to the mouse embryo, the in vitro mouse neuronal differentiation model we described here may provide a crucial tool for the analysis of neuronal development at a level that was heretofore restricted to lower vertebrates.

We thank Chantal Poujeol for her expert technical assistance and Drs Mireille Fauquet and Catherine Ziller for invaluable advice on microdissection techniques and culture in defined medium, Drs Nicole Baumann, Bernard Bizzini, Anne Marie Hill, Jacqueline Gabrion, Marie-Madeleine Portier and Denise Paulin, for helpful suggestions on immunohistochemistry and for kindly supplying antibodies. We are grateful to Dr Seana O’Regan for her advice on the manuscript. S.B. is supported by a Ministère de la Recherche et la Technologie fellowship. The work was supported by Ministère de la Recherche et la Technologie (n° 88.C.0565), Direction des Recherches Etudes & Techniques (n° 88/083), Association Française contre les Myopathies and Fondation pour la Recherche Médicale grants to M.S.

Bader
,
C. R.
,
Bertrand
,
D.
&
Dupin
,
E.
(
1985
).
Voltage-dependent potassium currents in developing neurons from quail mesencephalic neural crest
.
J. Physiol
.
366
,
129
151
.
Bader
,
C. R.
,
Bertrand
,
D.
,
Dupin
,
E.
&
Kato
,
A. C.
(
1983
).
Development of electrical membrane properties in cultured avian neural crest
.
Nature, Lond
.
305
,
808
810
.
Bignami
,
A.
,
Raju
,
T.
&
Dahl
,
D.
(
1982
).
Localization of vimentin, the non-specific intermediate filament protein in embryonal glia and in early differentiating neurons. In vivo and in vitro immunofluorescence study of the rat embryo with vimentin and neurofilament antisera
.
Devi Biol
91
,
286
295
.
Borelu
,
E.
,
Heyman
,
R.
,
Hsi
,
M.
&
Evans
,
R. M.
(
1988
).
Targeting of an inducible toxic phenotype in animal cells
.
Proc, natn. Acad. Sci. U.S.A
.
85
,
7572
7581
.
Bottenstein
,
J. E.
&
Sato
,
G. H.
(
1979
).
Growth of rat neuroblastoma cell line in serum-free supplemented medium
.
Proc. natn. Acad. Sci. U.S.A
.
76
,
514
517
.
Cochard
,
P.
&
Paulin
,
D.
(
1984
).
Initial expression of neurofilaments and vimentin in the central nervous system of the mouse embryo in vivo
.
J. Neurosci
.
4
,
2080
2094
.
Darmon
,
M.
,
Bottenstein
,
J.
&
Sato
,
G.
(
1981
).
Neural differentiation following culture of embryonal carcinoma cells in serum-free defined medium
.
Devi Biol
.
85
,
463
473
.
Escurat
,
M.
,
Gumpel
,
M.
,
Lachapelle
,
F.
,
Gros
,
F.
&
Portier
,
M. M.
(
1988
).
Expression comparée de deux protéines de filaments intermédiaires, la périphénne et la protéine de neurofilament 68 KD, au cours du développement embryonnaire du rat
.
C. r. hebd Séanc. Acad. Sci
.
306
,
447
456
.
Girdlestone
,
J.
&
Weston
,
J. A.
(
1985
).
Identification of early neuronal subpopulations in avian neural crest cells culture
.
Devl Biol
.
109
,
274
287
.
Greiner
,
C.
,
Lloyd
,
A.
&
Guroff
,
G.
(
1986
).
Ontogeny of the nerve growth factor receptor on neural crest cells
.
Dev. Brain Res
.
26
,
145
150
.
Hamill
,
I. P.
,
Marty
,
A.
,
Neher
,
E.
,
Sakmann
,
B.
&
Sigworth
,
G.
(
1981
).
Improved patch-clamp techniques for high resolution current recording from cell and cell-free membrane patches
.
Pfliigers Arch
.
391
,
85
100
.
Hinrichs
,
S. H.
,
Nerenberg
,
M.
,
Reynolds
,
R. K.
,
Khoury
,
G.
&
Jay
,
G.
(
1987
).
A transgenic mouse model for human neurofibromatosis
.
Science
237
,
1340
1343
.
Hogan
,
B.
,
Costantini
,
F.
&
Lacy
,
E.
(
1986
).
Manipulating the Mouse Embryo. Cold Spring Harbor Laboratory, Cold Spring Harbor, USA
.
Holtfreter
,
J.
(
1968
).
Mesenchyme and epithelia in inductive and morphogenetic processes
. In
Epithelial-Mesenchyme Interactions
(ed.
Fleischmajer
,
P.
&
Billingham
,
R.B.
) pp.
1
30
. (
Williams & Wilkins, Baltimore
).
Ito
,
K.
&
Takeuchi
,
T.
(
1984
).
The differentiation in vitro of the neural crest cells of the mouse embryo
.
J. Embryol. exp. Morph
.
84
,
49
62
.
Jaenisch
,
R.
(
1985
).
Mammalian neural crest cells participate in normal embryonic development on microinjection into post implantation mouse embryos
.
Nature, Lond
.
318
,
181
183
.
Jaenisch
,
R.
(
1988
).
Transgenic animals
.
Science
240
,
1468
1474
.
Julien
,
J. P.
,
Tretkakoff
,
L
,
Beaudet
,
L.
&
Peterson
,
A.
(
1987
).
Expression and assembly of a human neurofilament protein in transgenic mice provide a novel neuronal marking system
.
Genes & Dev
.
1
,
1085
1095
.
Kalcheim
,
C.
&
Le Douarin
,
N. M.
(
1986
).
Requirement of a neural tube signal(s) for the differentiation of neural crest cells into dorsal root ganglia
.
Devi Biol
.
116
,
451
466
.
Le Douarin
,
N.
(
1982
).
The Neural Crest
.
Cambridge University Press
.
Cambridge
.
Le Douarin
,
N.
(
1986
).
Cell line segregation during peripheral nervous system ontogeny
.
Science
331
,
1515
1521
.
Leonard
,
D. G. B.
,
Gorham
,
J. D.
,
Cole
,
P.
,
Greene
,
L. A.
&
Ziff
,
E. B.
(
1988
).
A nerve-growth factor-regulated messenger RNA encodes a new intermediate filament protein
.
J. Cell Biol
.
106
,
181
194
.
Levi-Montalcini
,
R.
(
1987
).
The Nerve Growth Factor: thirty-five years later
.
EMBO J
.
6
,
1145
1154
.
Metcalf
,
D.
&
Moore
,
M. A. S.
(
1971
).
Haematopoietic Cells. North-Holland, Amsterdam
.
Mirsky
,
R.
,
Wendon
,
L. B. M.
,
Black
,
P.
,
Stolkin
,
C.
&
Bray
,
D.
(
1978
).
Tetanus toxin: a cell surface marker for neurons in culture
.
Brain Research
148
,
251
259
.
Mori-Okamoto
,
J.
,
Ashida
,
H.
,
Maru
,
E.
&
Tatsuno
,
J.
(
1983
).
The development of action potentials in cultures of explanted cortical neurons from chick embryos
.
Devi Biol
.
97
,
408
416
.
Palmiter
,
R. D.
,
Behringer
,
R. R.
,
Quaif
,
C. J.
,
Maxwell
,
F.
,
Maxwell
,
I. H.
&
Brinster
,
R. L.
(
1987
).
Cell lineage ablation in transgenic mice by cell-specific expression of a toxic gene
.
Cell
50
,
435
443
.
Parysek
,
L. M.
&
Goldman
,
R. D.
(
1987
).
Characterization of intermediate filaments in PC 12 cells
.
J. Neurosci
.
7
,
781
791
.
Pfeiffer
,
S. E.
,
Jakob
,
H.
,
Mikoshiba
,
P.
,
Dubois
,
P.
,
Guenet
,
J. L.
,
Nicolas
,
J. F.
,
Gaillard
,
J.
,
Chevance
,
G.
&
Jacob
,
F.
(
1981
).
Differentiation of teratocarcinoma line: Preferential development of cholinergic neurons
.
J. Cell Biol
.
88
,
57
66
.
Portier
,
M. M.
,
Bracket
,
P.
,
Croizat
,
B.
&
Gros
,
F.
(
1984
).
Regulation of peripherin in mouse neuroblastoma and rat PC 12 pheochromocytoma cell lines
.
Dev. Neurosci
.
6
,
215
226
.
Raff
,
M. C.
,
Miller
,
R. H.
&
Noble
,
M.
(
1983a
).
A glial progenitor cell that develops in vitro into an astrocyte or an oligodendrocyte depending on culture medium
.
Nature, Lond
.
303
,
300
306
.
Raff
,
M.
,
Abney
,
E. R.
,
Cohen
,
J.
,
Lindsay
,
R.
&
Noble
,
M.
(
1983b
).
Two types of astrocytes in culture of developing rat white matter: differences in morphology, surface gangliosides and growth characteristics
.
J. Neurosci
.
3
,
1289
1300
.
Rossant
,
J.
&
Pederson
,
R.
(
1986
).
Experimental Approaches to Mammalian Embryonic Development. Cambridge Univ. Press, New York
.
Rovasio
,
R. A.
,
Delouvee
,
A.
,
Yamada
,
K. M.
,
Timple
,
R.
&
Thiery
,
J. P.
(
1983
).
Neural crest cell migration: requirement for exogenous fibronectin and high cell density
.
J. Cell Biol
.
96
,
462
473
.
Simonneau
,
M.
,
Distasi
,
C.
,
Tauc
,
L.
&
Poujeol
,
C.
(
1985a
).
Sequence of development of ionic channels during mouse neuronal differentiation
.
J. Physiol. (Paris)
80
,
312
320
.
Simonneau
,
M.
,
Edde
,
B.
,
Nicolas
,
J. F.
&
Jakob
,
H.
(
1985b
).
Single channel currents in mouse embryonic multipotential cells
.
Cell Diff
.
17
,
21
28
.
Spitzer
,
N. C.
(
1979
).
Ion channels in development
.
A. Rev. Neurosci
.
2
,
363
397
.
Temple
,
S.
&
Raff
,
M. C.
(
1986
).
Clonal analysis of oligodendrocyte development in culture: evidence for a developmental clock that counts cell division
.
Cell
44
, TT3-T19.
Vogel
,
K. S.
&
Weston
,
J. A.
(
1988
).
A subpopulation of cultured avian neural crest cells has transient neurogenic potential
.
Neuron
1
,
569
577
.
Weston
,
J. A.
,
Vogel
,
K. S.
&
Marusich
,
M. F.
(
1988
).
Identification of neural crest cell populations in early embryonic development
. In
From Message to Mind, Sinauer Associates, Sunderland, Mass. USA. pp. 224-237
.
Ziller
,
C.
(
1986
).
Development of neuronal properties in neural crest cells cultured in vitro. Currents Topics in Developmental Biology
20
,
177
193
.
Ziller
,
C.
,
Dupin
,
E.
,
Brazeau
,
P.
,
Paulin
,
D.
&
Le Douarin
,
N.
(
1983
).
Early segregation of a neuronal precursor cell line in the neural crest as revealed by culture in a chemically defined medium
.
Cell
32
,
627
638
.