Expression of tenascin in response to neural induction in amphibian embryos

In the amphibian embryo, it has been possible to study the sequence of interactions and morphogenetic movements that lead to the organization of the basic body plan (Slack, 1983). The first interaction occurs between animal hemisphere and vegetal hemisphere. It leads to the commitment of mesodermal cells in the equatorial region of the blastula (Nieuwkoop, 1969, 1973). This initial mesodermal rudiment is believed to contain a ‘dorsal’ and a ‘ventral’ zone which interact during the process of dorsalization thus determining the further patterning of mesoderm (Smith & Slack, 1983; Dale el al. 1985; Dale & Slack, 1987). Thereafter, the three germ layers are set up through the morphogenetic movements of gastrulation (Vogt, 1929; Holtfreter, 1943, 1944; Johnson, 1970; Nakatsuji et al. 1982; Keller, 1986; Shi et al. 1987). In gastrulating embryos, the chordomesoderm contacts the prospective neurectoderm and induces it to form neural tissue (Spemann & Mangold, 1924; Gimlich & Cooke, 1983; Duprate et al. 1984; Jacobson, 1984).

Tenascin is an extracellular matrix (ECM) protein which was first described as myotendinous antigen (Chiquet & Fambrough, 1984a; Chiquet-Ehrismann et al. 1986). It is an oligomer composed of polypeptides of relative molecular masses 220, 200 and 190X10 ³ (Chiquet & Fambrough, 1984b). In electron microscopy and by rotary shadowing, tenascin bears six arms attached to a central domain (Vaughan et al. 1987) in an identical manner to hexabrachions (Erickson & Iglesias, 1984; Erickson & Taylor, 1988). Both in its structure and tissue distribution, tenascin is very similar to cytotactin (Grumet et al. 1985), glycoprotein JI (Kruse et al. 1985) and glioma mesenchymal extracellular matrix protein (GMEM) (Bourdon et al. 1985).

During development, tenascin is expressed spatially and temporally in close relation to important developmental events. In the mammalian fetus, tenascin is observed in condensing mesenchyme inducing the differentiation of epithelia during morphogenesis of kidney (Aufderheide et al. 1987), tooth (Thesleff et al. 1987), vibrissa and mammary gland (Chiquet-Ehrismann et al. 1986). It is present at sites of bone and cartilage differentiation, and promotes chondrogenesis in vitro (Mackie et al. 1987). Like cytotactin in the chick embryo (Crossin et al. 1986), tenascin is found bordering the neural crest cell migration pathways during embryogenesis of rat, quail and Xenopus (Mackie et al. 1988).

The function of tenascin is not completely understood. In vitro, tenascin causes fibroblasts to round up on fibronectin substrates (Chiquet-Ehrismann et al. 1988). Cytotactin and tenascin could be involved in the modulation of the migratory behaviour of neural crest cell in vitro (Tan et al. 1987; Mackie et al. 1988).

The present study concerns the identification and expression of tenascin in Pleurodeles waltl embryos. On the basis of biochemical analyses, we provide evidence that amphibian polypeptides reacting with antisera to chick fibroblast tenascin are closely related to those already described as tenascin subunits in other vertebrates. Using such cross-reacting antibodies, we show that tenascin is distributed spatially and temporally in a way similar to bird and mammal embryos. Finally, we took advantage of the amphibian embryo to investigate the influence of early inductive events in the development of the pattern of tenascin. We principally focused on the formation of the neural tube and show that in vitro and in vivo, expression of tenascin in ectoderm is a response to neural induction.

Spawnings of Pleurodeles waltl were obtained naturally at the laboratory. Embryos were manually dejellied and kept at room temperature (RT) in Steinberg’s solution (SS) (Steinberg, 1957). They were staged according to Gallien & Durocher (1957).

Production of the rabbit antiserum against chick fibroblast tenascin has been previously reported (Chiquet & Fambrough, 1984n). Preparation of polyclonal rabbit anti-Ambystoma mexicanum plasma fibronectin IgG has been described elsewhere (Boucaut & Darribère, 1983). Mouse anti-Ambystoma mexicanum plasma fibronectin antisera used in double immunofluorescence labelling experiments were obtained as follows. Four mice received each a primary injection of 30/ μg of fibronectin and were boosted twice with 15 μg of immunogen. Sera were collected 7 days after the last boost. They were found to react specifically with fibronectin on immunoblots.

Extraction of proteins from adult Pleurodeles waltl tissues was performed according to Crossin et al. (1986). Organs were taken from animals anaesthetized with MS 222 (Sandoz), immediately placed in 5 vol. of ImM-EDTA, 30 mM-diethylamine, 2 mM-phenylmethylsulphonyl fluoride (PMSF), 5, μg ml ^-1 aprotinin, 0 ·5 μg ml’ ¹ leupeptin, pH 11·1 and homogenized on ice in a dounce homogenizer with a glass pestle. After a 1 h extraction on ice, homogenates were dialysed overnight at 4°C against distilled water. Extracts were centrifuged at 5000 g for 10 min at 4 °C and supernatants lyophilized. Proteins were fractionated on 7·5% (denaturing conditions) or 6% (nondenaturing conditions) continuous SDS-polyacrylamide gels (Laemmli, 1970). The proteins myosin (205×10 ³ Af_r), fi galactosidase (116×10 ³M_T), phosphorylase B (97×lO ³Af_r), bovine serum albumin (66×lO ³Af_r) and ovalbumin (45 ×lO ³Af_r) were used as molecular mass markers. Fractionated proteins were transferred onto nitrocellulose sheets according to Towbin et al. (1979). Immunodetection was carried out with anti-chick tenascin antiserum (1/500) and [ ¹²⁵I]protein A (Amersham) (2 μiCiml’ ¹).

Embryos at the early blastula stage (stage 5) were microinjected with 20 nl of nondiluted [ ³⁵S]methionine (Amersham) (12mCiml ^-1). Tailbud stage embryos (stage 26) were finely chopped with sharpened scissors and deposited in agar-coated wells containing 100 μl of a 0·2mCiml ^-1 solution of [ ³⁵S]methionine in SS. In both cases, incubation was performed for 4h. At the end of the incubation time, the stage-5 embryos had reached the midblastula stage (stage 6). Homogenization was carried out in solubilization buffer (0·lM-NaCl, 1% (w/v) NP40, 2HIM-PMSF, 2HIM-EGTA, 5gml ^-1 aprotinin, 0-Sggml’ ¹ leupeptin, 50 nw-Tris-HCl, pH 7·4). Extraction of proteins was done during 1 h on ice. Prior to immunoprecipitation, the yolk was spun down at 10 000g for 5 min and was discarded.

Tenascin was immunoprecipitated from metabolically labelled extracts using the procedure described previously (Riou et al. 1987). Protein A-Sepharose C14B beads (Pharmacia) were incubated with anti-tenascin antiserum at a ratio of 5 mg beads for 101 antiserum. Immunoprecipitates were analysed on a 7·5 % continuous SDS-polyacrylamide gel. Gels were processed for fluorography using Amplify (Amersham).

Embryos were frozen in isopentane cooled in liquid nitrogen and fixed in methanol at —80°C for 3 days as described in Levi et al. (1987). They were embedded in PEG 400 distearate (Dreyer et al. 1983) or rehydrated and then impregnated with 30 % sucrose in SS for cryostat sectioning. Adult tissues were dissected as described above for the Western blotting experiments. They were fixed with 4% paraformaldehyde and impregnated with 30% sucrose in SS. PEG sections were cut at 10 pm. 15–20 μm-thick frozen sections were produced with a Bright 5030 cryostat.

Before immunodetection, PEG sections were treated with 95 % ethanol in order to remove the embedding wax and then rehydrated in SS. After blocking nonspecific binding sites with 5% bovine serum albumin (BSA) in phosphate-buffered saline (PBS; 0·15M-NaCl, 10mM-sodium phosphate, pH7·l), sections were incubated with anti-tenascin antiserum (1/1000 dilution) or with rabbit anti-fibronectin IgG (2μgml ^-1). All dilutions of antibodies were done in PBS with 2·5% BSA. Thereafter, sections were rinsed in PBS and reacted with fluorescein isothiocyanate (FTTC)-conjugated sheep anti-rabbit IgG antibodies (Biosys, 10/tgml ^-1) for Ih at RT. After two final washes in PBS, sections were mounted in Mowiol 4–88 (Hoechst).

Double immunofluorescence labelling experiments were carried out on PEG sections made from tailbud-stage embryos (stage 26). After the blocking step, sections were successively incubated in anti-tenascin antiserum (1/1000 dilution), FTTC-conjugated sheep anti-rabbit IgG antibodies (Biosys, 10 μgml ^-1), mouse anti-fibronectin antiserum (1/400 dilution) and tetramethylrhodamine isothiocyanate conjugated rabbit anti-mouse IgG antibodies (Biosys, 10 μgmP ¹). Each incubation was performed for 1 h at RT and was followed by two rinses in PBS. Sections were mounted in Mowiol 4–88.

Observation was done on a Laborlux D Leitz epifluorescence microscope with filters for fluorescein and rhodamine.

All experiments were performed in sterile SS with 50 μgml ^-1 gentamycin (Gibco, BRL). In vivo ventral induction of a secondary embryo was carried out using standard procedures (Spemann & Mangold, 1924). Briefly, the dorsal blastoporal lip was dissected from an embryo at early gastrula stage and transplanted ventrally in a recipient embryo of the same stage (Fig. 1). After healing, the embryo was allowed to develop until tailbud stage (stage 26). In vitro induction of ectoderm into neural tissue was performed with the ‘sandwich’ method of Holtfreter (1933) or with the lectin concanavalin A (Takata et al. 1984). In the first case, the dorsal blastoporal lip which was used as inducer, was removed after 4h of contact with ectoderm in order to avoid the presence of mesodermal derivatives in the explant (Gualandris & Duprat, 1981). In the second case, animal pole explants were placed in a solution of 300 μgml ^-1 concanavalin A for 3h. Thereafter explants were cultured in SS with gentamycin until control embryos reached the stage 26.

Cross-reactivity of anti-chick tenascin antiserum was assayed on Pleurodeles waltl adult tissues by Western blotting and indirect immunofluorescence. Brain, lung and stomach were selected because tenascin or a related molecule, cytotactin, had been found in these organs in the chick (Chiquet & Fambrough, 1984a; Crossin et al. 1986). Conversely, these analyses were carried out on a putatively negative tissue, the liver. Immunoblotting of high pH extracts from brain and liver is shown in Fig. 2. In brain extracts after reduction, anti-tenascin antiserum strongly reacted with a major 220 x 10 ³M_T polypeptide and more weakly with 240, 200, 190 and 94X10 ³ components (Fig. 2A, lane 1). Values of 240, 220, 200 and 190×lO ³ agree well with those reported for subunits of chick, mouse or rat tenascin (Chiquet ×0026; Fambrough, 1984b; Aufderheide et al. 1987; Vaughan et al. 1987). Although a lOO×lO ³ component was described for chick gizzard cytotactin (Crossin et al. 1986), the 94×10 ³ band might correspond to a prominent proteolytic fragment (Chiquet & Fambrough, 1984b). Under nonreducing conditions, reactivity was found in material barely entering the 6% polyacrylamide gel and in two groups of compounds of 350–290 and 210–180×10 ³, respectively (Fig. 2B, lane 1). A similar pattern has been observed after limited trypsin cleavage of purified tenascin and was found to correspond to the oligomeric, dimeric and monomeric forms of the molecule (Taylor et al. 1987; M. Chiquet, unpublished data). This indicates that some proteolysis occurred during the high pH extraction. Since the 94×10 ³ band was not visible on nonreducing gels, this component must be disulphide-linked to the large tenascin complexes. No reactivity could be detected in blots of fiver extracts (Fig. 2A,B, lanes 2). Western blotting experiments performed with lung and stomach extracts provided qualitatively similar results to those with brain (not shown). The only difference was the appearance of a more heavily labelled 94×10 ³ component.

In order to specify the distribution of tenascin in adult tissues, frozen sections were processed for immunofluorescence. In the brain, tenascin staining was observed in the extracellular spaces between neurone cell bodies. It was also widely distributed in the white matter (Fig. 3A). In lung and stomach, reactivity to anti-tenascin antiserum was concentrated in the ECM associated with smooth muscle extracellular spaces between neurone lung, tenascin was present in the muscular trabeculae forming the borders of the alveoli (Fig. 3B). In the stomach, fluorescence occurred in the muscular coat (Fig. 3C), the muscularis mucosae and the smooth muscle cells of the mucous coat (not shown). As expected, liver parenchyme did not exhibit reactivity to anti-tenascin antiserum (Fig. 3D).

Distribution of tenascin in the embryo was investigated by indirect immunofluorescence. Tenascin staining appeared first at the early-tailbud stage (stage 22). It was present at all levels of the embryo in the ECM located between the dorsal epidermis and the neural tube where the neural crests are individualizing (Fig. 4A). Thereafter, from stage 24 onward, tenascin expression extended laterally in a dorsal area including the neural crest cell migration pathways. In the cephalic region, reactivity to anti-tenascin antiserum occurred in the ECM present at the edge of the developing brain and in the neighbouring cephalic mesoderm (not shown). In more caudal regions (Fig. 4B), tenascin was revealed in the ECM bordering the neural tube and in extracellular fibres in the vicinity of the notochord. Tenascin was also observed in the ECM in contact with the somites except on the edge facing the lateral epidermis. Tenascin-positive ECM was particularly developed in the intersomitic area. Some fluorescence also underlined the surface of cells forming the rostrointemal part of the somite. Laterally, tenascin was present in the ECM surrounding the Wolffian duct. At the late-tailbud stage (stage 28), tenascin distribution in the dorsal region of the embryo was similar to that described at stage 24. In addition, some reactivity to anti-tenascin antiserum occurred ventrally. It was concentrated in the ECM present at the periphery of the aortic arches (Fig. 5A) and the endocardial tube (Fig. 5B). At the other levels, tenascin expression remained dorsal (Fig. 5C).

Using double immunofluorescence staining, the distribution of tenascin was compared with that of fibronectin. The experiments were performed at mid-tailbud stage (stage 26). Obviously, the spatial distributions of the molecules differ. Indeed, whereas fibronectin appeared in every tissue of the embryo, tenascin, as was pointed out above, was almost exclusively expressed dorsally. Therefore, attention was focused on the dorsal region of the embryo. In this area, tenascin appeared to be distributed together with fibronectin. All the structures revealed by anti-tenascin antiserum (Fig. 6A) were equally stained by anti-fibronectin antiserum (Fig. 6B). In contrast, numerous fibronectin-positive extracellular fibres were totally devoid of reactivity to anti-tenascin antibodies. It appears then that tenascin-containing ECM is a subset of the fibronectin-containing one.

In order to provide biochemical data concerning the expression of tenascin in the embryo, immunoprecipitation experiments were carried out. [ ³⁵S]methionine-labelled total protein from blastula stage (stage-6) or tailbud stage (stage-26) embryos containing the same quantity of radioactivity were subjected to immunoprecipitation with anti-tenàscin antiserum. As shown in Fig. 7, no tenascin-like protein was detected at blastula stage. On the contrary, at stage 26, a 220X10 ³ polypeptide was specifically precipitated. The appearance of immunoreactivity for tenascin in the ECM in the embryo can therefore be correlated with the synthesis of the 220 ×10 ³ subunit of the molecule.

In vitro and in vivo induction experiments were performed in order to investigate the patterning of tenascin in the embryo. They were focused on the question whether tissues isolated at early gastrula stage (ectoderm, dorsal blastoporal lip) are able to produce tenascin and whether neural induction influences tenascin expression.

Ectoderm and dorsal blastoporal lip explants were excised at early gastrula stage (stage 8a) and were cultured 3 days until control embryos had reached the tailbud stage (stage 26). When isolated and cultured, ectoderm differentiated into epidermis, and dorsal blastoporal lip differentiated into notochord, mesenchyme and striated muscle cells. Serial sections of differentiated epidermal vesicles, when reacted with anti-tenascin antiserum, were totally devoid of fluorescence (Fig. 8A). As a control, presence of fibronectin in ECM, which is independent of neural induction (Duprat & Gualandris, 1984; Boucaut et al. 1985; Grunz et al. 1987), was investigated. As expected, fibronectin staining was highly positive in ECM that is deposited in the epidermal vesicle (Fig. 8B). As shown in Fig. 8C,D, sections of cultured dorsal blas-topbral lip contained extracellular fibres strongly positive for tenascin staining (Fig. 8C) and fibronectin staining (Fig. 8D). On the basis of these results, the question arose to what extent the tenascin in the ECM enclosing the neural tube was produced by induced ectodermal cells or by mesodermal cells such as notochord cells. To answer this question, we performed in vitro neural induction of ectoderm at early gastrula stage. At first, we used dorsal blastoporal lip as inducer (Holtfreter, 1933). Ectoderm and dorsal blastoporal lip were associated for 4h at the end of which the dorsal blastoporal lip was removed (Gualandris & Duprat, 1981). Thereafter, the induced ectoderm was cultured until controls reached the stage 26. At that time, the explant exhibited many neural-tube-like structures (Fig. 8E,F). Unlike noninduced ectoderm, staining with anti-tenascin of sections from induced tissue revealed a strongly reacting ECM at the borders of neural structures (Fig. 8E). To confirm this result, a second set of in vitro neural induction was carried out using the lectin concanavalin A as inducer (Takata et al. 1984; Fig. 8G,H). As expected, strong reactivity to anti-tenascin antiserum was revealed at the limits of neural vesicles induced by the lectin (Fig. 8G). These results are summarized in Table 1.

In a second set of experiments done in vivo, we have induced ventrally a secondary embryo using the classical method of Spemann & Mangold (1924). Here, two kinds of interactions could influence the expression of tenascin. First, the grafted dorsal blastoporal lip induces ventral ectoderm of the recipient embryo into neural tissue (Spemann & Mangold, 1924; Jacobson, 1984). Second, the grafted tissue influences the ventral mesoderm of the host embryo to differentiate in more dorsal structures (Smith & Slack, 1983). Eleven specimens exhibiting a characteristic secondary embryo were serially sectioned and processed for immunofluorescence. In all cases, the expression pattern of tenascin in the secondary embryo was identical to that occurring in the primary one (Fig. 9). Reactivity to anti-tenascin antiserum was observed in the ECM located around the ventrally induced neural tube. This confirms in vivo that tenascin can be expressed by ectoderm as a response to neural induction. Tenascin was also revealed in the secondary embryo in the ECM associated with notochord and somites. Notochord as well as some somite cells derives from the grafted dorsal blastoporal lip. Other cells constituting the somites come from the respecified host ventral mesoderm (Smith & Slack, 1983). Some of the tenascin associated with somites might be produced by the respecified host mesoderm. Nevertheless, this remains unclear because the grafted dorsal blastoporal lip itself produces tenascin.

These experiments show that, once induced, neural cells are able to produce tenascin. This point is consistent with the proposition that tenascin is expressed in ectodermal cells as a response to neural induction.

The present work was undertaken in order to explore the pattern of expression of tenascin during embryogenesis of the amphibian Pleurodeles waltl. The study was performed using anti-chick tenascin antiserum as a probe. This antiserum was found to cross-react with polypeptides from adult Pleurodeles waltl whose molecular weight and tissue distribution agreed well with those reported for tenascin or a related molecule, cytotactin, in the chick. In the Pleurodeles waltl embryo, tanascin was first detected after completion of neurulation. It was present in the ECM concomitantly with fibronectin at sites of directed cell migration (neural crest cell migration pathways; extension of the pronephretic duct) or where mesenchyme condensation occurred (formation of endocardium and aortic arches). At that time, expression of tenascin could be correlated with the synthesis of the 220×10 ³ chain of the molecule. Finally, tenascin was found to be absent from ectoderm differentiating into epidermis. However, it appeared that both in vivo and in vitro, tenascin could be expressed in ectoderm as a response to neural induction.

The molecule recognized in Pleurodeles waltl by the cross-reacting antiserum appeared to be homologous to chick tenascin according to biochemical and immunocytochemical data. It should be noted that, in Pleurodeles waltl brain, tenascin is more widely distributed than cytotactin which was only found in the molecular layer of cerebellum and the forebrain of the chick (Crossin et al. 1986). Otherwise, tenascin was revealed in smooth muscle which belongs to the tissues described to contain tenascin in the chick (Chiquet & Fambrough, 1984a).

The results concerning the distribution of tenascin in Pleurodeles waltl embryos are in good agreement with the data obtained for tenascin or cytotactin in embryos of other vertebrate species (Crossin et al. 1986; Mackie et al. 1988). They confirm the distribution of tenascin observed by Mackie et al. (1988) in the truncal region of Xenopus embryos at stages 25/26 and 32. These stages in Xenopus embryos (Nieuwkoop & Faber, 1967) are equivalent to stages 28 and 30 in Pleurodeles waltl embryos (as a comparison, appearance of tenascin in Pleurodeles waltl at stage 22 would correspond to stage 22/23 in Xenopus development).

The spatial and temporal patterns of tenascin expression in the amphibian embryo are more restricted than those of other ECM constituents, fibronectin and laminin. These molecules are present in the ECM as early as the midblastula stage (Boucaut et al. 1985; Nakatsuji et al. 1985; Darribère et al. 1986), whereas tenascin appears at extracellular sites much later, after neurulation is complete. At that time, fibronectin is present in every tissue of the embryo. Laminin is located in the basement membranes of all the epithelial structures (unpublished results). In contrast, tenascin is distributed essentially in the dorsal region of the embryo at sites of directed cell migration, i.e. along the principal neural crest cell migration pathways, (Chibon, 1967; Vogel & Model, 1977; Lofberg et al. 1980) and at the level of the extending Wolffian duct (Poole & Steinberg, 1981, 1982). This observation suggests that tenascin might play a role in these processes. The function of fibronectin in cell locomotion has been extensively described (for review, see Duband et al. 1987). Tenascin, which we showed to be always distributed together with fibronectin, might be involved in directing neural crest cell and pronephretic duct migrations by modulating the interaction of locomoting cells with the fibronectin substrate. Indeed, tenascin was recently found to inhibit the spreading of chick or rat fibroblasts on fibronectin substrate in a dose-dependent manner. This interference in the fibronectin-to-cell interaction might involve a cell surface receptor for tenascin (Chiquet-Ehrismann et al. 1988). In vitro migration assays indicate that cytotactin affects the migratory behaviour of chick neural crest cell on fibronectin in a similar way (Tan et al. 1987). However, quail neural crest cell have been found to migrate to some extent on a substrate of tenascin in vitro (Mackie et al. 1988). Tenascin was also observed at sites of mesenchyme condensation, in endocardium and aortic arches. This can be related to kidney morphogenesis in the mouse embryo (Aufderheide et al. 1987). In that case, tenascin surrounded newly formed S-shaped tubules. It was proposed that tenascin has a role in epithelial-mesenchymal interactions. Nevertheless, the function of tenascin in this process remains unclear.

The results presented here on Pleurodeles waltl confirm that tenascin expression during embryogenesis is correlated spatially and temporally with important developmental events. However, it is not known how the pattern of expression of tenascin in the embryo is produced. The tissues that deposit tenascin-containing ECM arise as a consequence of morphogenetic movements and inductive processes during early development. In this paper, we provide data concerning the relationship that exists between neural induction and the patterning of tenascin. The results obtained by experimentally inducing neural tissue in vitro showed that tenascin-containing ECM was absent from ectoderm derivatives in the absence of neural induction but always formed after induction. Particularly significant were the inductions with the lectin concanavalin A where ectoderm cannot be contaminated by any mesodermal tissue. A confirmation was obtained in vivo by Spemann experiments in which ventrally induced neural tube expressed tenascin. This indicates that tenascin found at the border of the neural tube is probably produced by this tissue. Although this was not studied in detail, tenascin expression seems to follow a timing after in vitro induction similar to that observed in vivo (unpublished results). This suggests that the appearance of tenascin-containing ECM at the level of the individualizing neural crest might be conditioned by neural induction.

Finally, whereas the location of tenascin in the vicinity of the neural tube seems to be a consequence of neural induction, this process alone cannot explain the dorsal pattern of expression of tenascin in the embryo. As pointed out above, tenascin-containing fibres were abundant in the isolated dorsal blastoporal lip. This suggests that, together with neural induction, mesoderm induction by endoderm (Nieuwkoop, 1969) can influence tenascin expression in embryonic tissues. It is striking that, with the exception of endocardial tube and aortic arches (which differentiate somewhat later), ventral mesoderm is devoid of tenascin at the tailbud stage although dorsal and lateral mesodermal derivatives (notochord, somites, pronephretic duct) all express the protein. Although the influence of dorsalization of ventral mesoderm on the expression of tenascin could not be recorded in our Spemann experiments, it is tempting to speculate that the mechanisms that lead to the organization of dorsal and lateral mesoderm in the embryo (Smith & Slack, 1983; Dale & Slack, 1987; Cooke et al. 1987) are responsible for the expression of tenascin at extraneural sites in the tailbud-stage embryo.

The research was supported by CNRS and FRM. The authors are most grateful to Dr S. Huang for his contribution to in vitro induction experiments and to Dr A. M. Duprat for critical reading of the manuscript. They also would like to thank Dr M. Delarue for help and advice, and C. Montmory, P. Groué and J. Desrosiers for their technical assistance.