Analysis of high PSA N-CAM expression during mammalian spinal cord and peripheral nervous system development

Informative interactions between cell surfaces are thought to be essential to the processes of cell recognition, sorting and migration during development and for subsequent stabilisation of tissues in the adult (Dodd and Jessell, 1988). A variety of surface proteins involved in cell contact has been identified in the nervous system. The best characterized are a group of high molecular weight glycoproteins known collectively as the neural cell adhesion molecules (N-CAMs) (Edelman, 1985).

These N-CAMs have been implicated in a wide range of morphogenic events including cell segregation and axonal bundling (Rutishauser and Jessell, 1988; Thanos et al. 1984), the migration of neurons along glial pathways (Silver and Rutishauser, 1984), and in several aspect of myogenesis including muscle innervation and synaptogenesis (Covault et al. 1986; Landmesser et al. 1990; Rieger et al. 1988; Tosney et al. 1988).

During these processes, the adhesive properties and functions of N-CAMs are apparently modulated by the differential expression of variant N-CAM polypeptide isoforms whose diversity is generated both by alternative splicing and differential polyadenylation site selection within mRNA products coded by a single complex gene (Santoni et al. 1989; Thompson et al. In addition, N-CAMs are subject to a variety of post-translational modifications including glycosylation, phosphorylation and sulfation (Finne et al. 1983; Gennarini et al. 1984; Key and Acheson, 1990; Rutishauser et al. 1988).

N-CAM varies in its content of a 2 – 8 polysialic acid (PSA) from a maximum of about 30 % (wt/wt) to less than 10 %. It has been proposed that changes in sialic acid content are involved in the function of N-CAMs in neuronal development (Finne et al. 1983).

For avian nervous tissue development, it was demonstrated that high PSA N-CAM (i) can induce fasciculation in DRG neurons (Rutishauser et al. 1988), (ii) can be involved in the formation of connexions between the retinal ganglion axons and their targets (Schlosshauer et al. 1984) and (iii) can act as a regulator of intramuscular nerve branching during embryonic development (Landmesser et al. 1990).

However, for the mammahan embryo, no experimental data were available for either (i) the spatial and temporal expression of high PSA N-CAM forms during nervous tissue development or (ii) the effect of in vitro perturbation of high PSA N-CAM levels.

In the present paper, we used a monoclonal antibody specific for high PSA N-CAM (Rougon et al. 1986) to study the expression of high PSA N-CAM, using combination of immunostaining of mouse embryo sections, immunodetection on microquantities of nervous tissues and experimental perturbation of high PSA N-CAMs in in vitro models.

Five specific questions are addressed: (i) are sialylation and synthesis of N-CAM isoforms co-regulated, (ii) are high PSA N-CAM molecules homogeneously or heterogeneously distributed on individual embryonic neurons, (iii) is this high PSA N-CAM expression linked with gangliogenesis in peripheral nervous system development, (iv) is this expression linked with a neuronal phenotype and (v) are high PSA N-CAM molecules involved in nervous tissues morphogenesis when neurons grow on a fibronectin substratum?

The anti-PSA monoclonal antibody was prepared by immunizing mice with viable meningococcus group B bacteria (strain P355) which shares the carbohydrate (NeuAc alpha 2 – 8) _n with high PSA N-CAM. This anti-Men B antibody is a mouse IgM and its specificity had previously been described (Rougon et al 1986). It was used at a dilution of 1100.

The rabbit antiserum directed against total N-CAM was a site-directed antibody recognizing the seven N3/4-terminal residues of N-CAM whose sequence is shared by every isoform (Rougon and Marshak, 1986) and was used at a dilution of 1:1000

The 160× 10³M_r neurofilament protein was detected using a mouse IgGi monoclonal antibody (Boehringer Mannheim) which specifically recognizes NFM and was used at a 1:20 dilution.

The anti-vimentm antisera was a gift from A M. Hill (Université d’Orsay). It was used at a dilution of 1.500.

Spinal cords were isolated from E9 to E17 mouse embryos and from neonates, using fine tungsten needles The microdissection of dorsal root ganglia was performed as described in Simonneau et al. (1987) and in Valmier et al. (1989) 

Developing mouse tissues were homogenized in 5 volumes of lysis buffer containing 2 % NP40, protease inhibitors and deoxyneurammic acid (1 m.?) as a neuraminidase inhibitor (Rougon et al 1982). The homogenate was left for 10 mm at 4 °C and then centrifuged at 140 000 g for 60 min.

The supernatant was made at 10 mg proteins ml⁻¹ and boiled for 3 min with an equal volume of 2×Laemmh sample buffer containing mercaptoethanol and submitted to electrophoresis (Laemmh, 1970)

Molecular weight markers were myosin (200000), β- galactosidase (116000), bovine serum albumin (66800) and ovalbumine (45000).

Immunoblotting was performed essentially as previously described (Rougon et al 1982, 1986; Rougon and Marshak, 1986). Nitrocellulose sheets were soaked in saturation buffer (3% low-fat milk in PBS) for 2h at 37 °C and incubated overnight at 4°C with dilutions of antibodies (1:1000 for rabbit antiserum, 1:1000 for anti-Men B ascite).

After three washes of 15min in PBS, the sheets were saturated again and incubated for 30 min in 10⁶ctsmin⁻¹iodinated ¹²⁵I-protein A (rabbit antiserum) for 3h or in 10⁶ctsmin⁻¹ iodinated ¹²⁵I-rat IgM (Institut Pasteur, France) The dried nitrocellulose sheets were autoradiographed on Fuji films at –70°C in the presence of sensitizing screens for 20 h.

Truncal regions from mouse embryos and neonates were isolated, embedded and oriented in Tissue-Tek (Miles) and rapidly frozen in cold isopentane. Sections of 10μm were attached to glass shdes which were previously coated with poly-L-lysine (Sigma, 260 000; 0.1 mg ml⁻¹).

After incubation with 3% BSA in PBS for 10 min at room temperature, sections were washed with PBS before being incubated with dilutions of antibodies for 1 h at 37 °C Sections were washed three times with PBS, incubated with second appropriate fluorescent antibodies for 30 min at 37 °C, washed three times again with PBS and mounted with Fluoprep (Institut Merieux).

Dorsal root ganglia were dissected out and cultured as previously described in Simonneau et al. (1987) and in Valmier et al. (1989). Spinal cords were similarly dissected out from E12 mouse embryos and dissociated in PBS, using a Pasteur pipette. Cell suspensions were plated on fibronectin-coated coverslip in 10% FCS supplemented culture medium. Neural crest cells from E9 mouse embryos were cultured on fibronectin-coated glass covershps using the technique previously described in Boisseau and Simonneau (1989).

Immunodetection of cell surface molecules was performed on living cells covershps were incubated either with anti-Men B monoclonal antibody or with the rabbit antisera against NH₂-terminal residues of N-CAM, diluted in the culture medium for 40min at 37°C, in a humidified atmosphere, washed three times in wash medium (DME+10% FCS) and incubated with the appropriate second fluorescent antibody under the same conditions described above After three washes in wash medium, cells were fixed and permeabihzed with methanol at —20 °C for 10 min to detect intermediate filaments. Coverslips were incubated in 3 % BSA in PBS for 10 mm at room temperature, washed in PBS and incubated either with anti-l?O×lO³!!/,. neurofilament antibody or antivimentin antiserum for 1 h at room temperature, in a humidified atmosphere After three washes in PBS, cells were incubated with appropriate second fluorescent antibody for 30min, washed again three times in PBS and mounted with Fluoprep.

Sections and cell cultures were observed for immunofluorescence on a Nikon Diaphot TMD microscope The anti-Men B monoclonal antibody was detected using a fluorescem-conjugated goat anti-mouse IgM antibody (1:100, Cappel). The monoclonal anti-160× 10³M_r neurofilament protein anti-body was detected using a rhodamine-conjugated goat antimouse IgG antibody (1:100; Nordic) The anti-vimentin and site-directed antibodies were revealed with a rhodamine- or a fluorescein-conjugated goat anti-rabbit antibody (1:100; Nor-dic).

For high PSA N-CAM perturbation experiments, spinal ganglia were obtained from E14 mouse embryos as described in Simonneau et al (1987) and were cultured without cell dissociation on glass covershps coated either with human fibronectin (lOμgml⁻¹) or with a collagen gel substratum. The medium was supplemented with 100 ng ml⁻¹ 7S nerve growth factor (Collaborative Research). Neural crest cells were cultured as previously described (Boisseau and Simon-neau, 1989) or co-cultured with embryonic cardiac cells microdissected from E14 mouse embryos (Boisseau et al. unpublished data). The endoneuraminidase (endoN) was prepared using a technique modified from Finne and Makela (1985) (activity=3.5×10’¹⁰pfuml^l) The enzyme was added at the culture medium at the concentration of 7.10⁷pfuml⁻¹.

For spinal ganglia, extent of outgrowth was expressed in terms of distance of radial outgrowth from the ganglion and fasciculation was estimated as described in Rutishauser et al. (1985). For neurons derived from neural crest cells, fasciculation was estimated by measuring three classes of neurite diameters.

Experiments were designed to test if, during neural development, transitions between the adult form of N-CAM in neuronal progenitors, the high PSA form of N-CAM in immature neurons and adult form of N-CAM in mature neurons can be detected, as proposed by Thiery et al. (1990).

We microdissected developing spinal cords at different embryonic ages to compare the expression of high PSA N-CAM with that of total N-CAM (Fig. 1).

High PSA N-CAM was first detectable in embryonic spinal cord at day 11 as shown in Fig. 1. The diffusible staining increased in intensity to reach a maximum at day 14, and then slowly decreased. The site-directed antibody, which recognizes all the isoforms of N-CAM (Rougon and Marshak, 1986), revealed a 140×10³M_r chain as early as E9 (Fig. 1). From Ell to neonate, both 140 and 180× 10³M_r isoforms were expressed. This expression increased in intensity until E14 and then decreased. The 120×10³ polypeptide isoform was only detectable on neonate samples.

It is interesting that the level of high PSA N-CAM in embryonic stages appeared to parallel that of total N-CAMs.

To determine whether high PSA N-CAM expression and synthesis of N-CAM isoforms were co-regulated in developing spinal cord, we calculated the relative ratio of high PSA N-CAMs versus total N-CAMs, by excising and counting the radioactivity of the bands revealed by the anti-Men B and site-directed antibodies, respectively. This is shown in Fig. 2. From E12 to E14, the calculated ratio was almost constant, but a sharp decrease in N-CAM polypeptides was observed after E14 whereas the level of high PSA N-CAM remained high. Between E17 and birth, N-CAM polypeptide level was maintained but polysialylation was highly repressed. These data demonstrate that, during development, the post-translational modification, polysialylation, is independently regulated from the expression of the polypeptide moieties of N-CAMs.

This absence of co-regulation strongly suggests that high level of high PSA N-CAM isoforms are especially required at some stages for the development of specific regions and could be involved in particular morphogenetic functions.

Schlosshauer et al. (1984) reported that high PSA N-CAMs are differentially distributed in the developing chick visual system. They found that high PSA N-CAM is present at the surface of growing axons, at the difference of the perikaryons and proximal axons, which expressed low PSA form of N-CAM.

To test if such a distribution is a common feature of all embryonic neurons expressing N-CAM, we studied the distribution of high PSA N-CAM in spinal cord and peripheral nervous system neurons using immunostained sections of truncal region of mouse embryos. Motoneurons can be visualized at E10.5 (Fig. ?A) on the basis of the l?O×lfP?fr neuro filament protein immunodetection in the ventral area of the neural tube, as reported by Cochard and Paulin (1984). These neurons are known first to proliferate and then to migrate anterolaterally along vimentin-rich glial fibers (Bignami et al. 1982) to form lateral motor columns. Fig. ?B shows that those motoneurons that started to extend their ventral roots lateroventrally into the mesenchyme, highly express high PSA N-CAM both at the level of their perikaryons and axons. At E12, when the motor axons had reached the myotome and established synaptic contact (Tosney et al. 1988), they still express the high PSA N-CAM isoform homogeneously on their cell bodies and all along their axons (Fig. ?C,D).

The second type of neurons studied was the commissural neurons. They can easily be detected at E10.5 when some 160× 10³M_r neurofilament positive cells were first observable on the lateral border of the neural tube (Fig. ?A) (Dodd et al. 1988; Holley, 1982; Holley et al. 1982). However, no expression of high PSA N-CAM was detectable in the region of the cell body (Fig. 3B). By contrast, at E12, when commissural fibers extended ventrally in the direction of the floor plate (Fig. 3C), a high level of high PSA N-CAM can be detected both at the level of cell bodies (Fig. 3D) and in the region of the floor plate where commissural fibers decussate (Fig. 4D). It is interesting to note that epithelial cells of the floor plate express high PSA N-CAM (Fig. 4B).

The third type of cells studied was the dorsal root ganglia cells. High PSA N-CAM was first detectable at E12, both on the cell body and on the axons located in the dorsal root and in the white columns (Fig. ?C,D).

A homogeneous distribution of high PSA N-CAMs was confirmed on short-time cultures of both spinal cord neurons and DRG cells isolated from E12 mouse embryos. Using western blot analysis on microdissected tissues, we demonstrated that high PSA N-CAM was clearly detectable at E12 in spinal cord (Fig. 1). For the same developmental stage, very low high PSA N-CAM level relative to total N-CAM polypeptide was detect-able on western blot (Fig. 6). However, in both cases, we found that in short-term cultures, high PSA N-CAM was homogeneously expressed on DRG neurons (Fig. 5A,B) and spinal cord neurons (Fig. 5C,D) both at the level of their cell bodies and of their neurites.

An immunoblot analysis was performed using dorsal root ganglia dissected from developing mouse embryos (Fig–6).

High PSA N-CAM was weakly expressed in spinal ganglia whatever the stage of development (from E12 to birth) when compared to the profile observed for developing spinal cord (compare Fig. 6 with Fig. 1). In striking contrast, total N-CAMs were highly expressed as soon as E12, the earliest stage at which it was possible to dissect out the DRGs, and throughout embryonic development.

From mouse embryo sections, no expression of high PSA N-CAM can be detected in the DRG primordium at E10.5 (Fig. ?B), but a clear expression of high PSA N-CAM was detectable at E12 (Fig. 3D). As spinal ganglia are known to be heterogeneous and to present a variety of different types of progenitors (Le Douarin, 1986), one could expect to find cells with different levels of high PSA N-CAM. This is not the case, as no heterogeneity in the staining was observed inside the ganglion both on transverse or sagittal sections of spinal ganglia. Similarly, all DRG neurons microdissected from E12 mouse embryos and cultured for 24 h were stained by antibody specific to high PSA N-CAM (Fig. 5B).

These results demonstrate that high PSA N-CAM can be expressed in neurons after their aggregation into a ganglion.

Peripheral nervous system offers a rather unique situation where common progenitors are responsible for the formation of distinct ganglia, namely the dorsal root ganglia, the sympathetic ganglia and the parasym-pathetic ganglia (Le Douarin, 1982). We took advantage of this situation to analyze the possible relation of high PSA N-CAM expression with gangliogenesis.

Using sections of embryos, we compare high PSA N-CAM immunostaining in (i) dorsal root ganglia (Fig. 3), (ii) sympathetic ganglia and (iii) aortic plexuses (Fig. 7) at E12 when these three structures are known to be fully organized (Cochard et al. 1978). Data on DRG immunostaining have been discussed above. Fig. 7A shows a staining of sympathetic ganglia with anti-160×10^3M_r neurofilament antibody. Along the ventral surface of aorta, sympathetic neurons forming the prirnordium of aortic plexuses were stained intensively by the same antibody as previously reported by Cochard and Paulin (see Fig. 12 in Cochard and Paulin, 1984). High PSA N-CAM is differentially expressed in these two derivatives since the sympathetic ganglia are faintly but clearly stained by anti-Men B monoclonal antibody and the sympathetic neurons of the aortic plexuses do not show any detectable expression of high PSA N-CAM.

To study a possible relationship between high PSA N-CAM expression and neuronal phenotype, which is difficult to examine in vivo, we used an in vitro model.

By using a chemically defined medium and human fibronectin as a substratum, we already described an in vitro model where mouse neural crest cells can be differentiated into neurons as early as 24 h in culture (Boisseau and Simonneau, 1989). In such a system, we observed that (i) only cells with a neuronal phenotype obtained in defined medium were stained by anti-Men B monoclonal antibody or by polyclonal antibody against total N-CAM (Fig. 8D,E,F), (ii) the staining was strictly superimposable, suggesting that no cells express adult isoforms in isolation.

When neural crest cells were cultured in FCS-supplemented medium, a condition where no neuronal differentiation occurs (Boisseau and Simonneau, 1989), they expressed vimentin but no detectable forms of N-CAMs (Fig. 8A,B,C).

The role of high PSA N-CAM molecules in peripheral nerve tissue morphogenesis were demonstrated by Rutishauser e? al. (1985). These authors found that specific cleavage of high PSA molecules by endoneur-aminidase increased fasciculation and decreased neurite growth of chick DRG neurons, provided these neurons were cultured on a collagen substratum.

It is well documented that fibronectin is present in vivo in the pathways of neural crest cell migration and in the regions where the neural crest cells aggregate to form peripheral ganglia (Duband et al. 1986; Thiery et al. 1990). We directly studied a possible role of high PSA N-CAM molecules on neurite growth and fasciculation of neural crest cells and their derivatives when they were cultured on a fibronectin substratum.

We found an increase in fasciculation and a decrease in neurite growth for mouse DRG ganglia explanted from E14 and cultured in the presence of endoneuraminidase, provided that the cells were plated on a collagen substratum (Table 1 and Fig. 9A). This result is comparable to that described by Rutishauser et al. (1985). However, no significant difference was found for fasciculation and neurite growth when DRG ganglia were grown on a fibronectin substratum (Table 1 and Fig. 9C,D). These results suggest that high PSA N-CAM molecules are not prevalent among other cell adhesion molecules when these molecules interact with a fibronectin substratum. However, in contrast to collagen substratum, on a fibronectin substratum, neurites extend on flattened cells that have emigrated from the explant, thus adding new types of adhesions (neurite-cell interactions).

The role of high PSA N-CAM molecules was further studied using the in vitro neuronal differentiation model of mouse neural crest cells on a human fibronectin substratum. When neural crest cells are co-cuitured with embryonic cardiac cells in a defined medium (Boisseau et al. in preparation) and on a fibronectin substratum, the neuronal derivatives are organized in clusters, with neurites presenting fasciculation (Fig. 10A,B and Fig. 12A). Using double immunostaining of 160×10³M_r neurofilament and of high PSA N-CAM, we found no significant difference in their neuritic growth or fasciculation when endoneuraminidase was applied to these co-cultures (Fig. 10C,D) as compared to control culture. We quantitatively studied clusters present in these co-cultures and found no differences in cluster size after endoneuraminidase perturbation experiment (Fig. 11). Furthermore, a quantitative analysis of neurite diameters in these cocultures showed that fasciculation was not modified by endoneuraminidase application (Fig. 12).

The use of a specific antibody against high PSA N-CAM (Rougon et al. 1986) combined with immunoblots applied to microquantities of tissues and with experimental perturbations of PSA using endoneuraminidase on in vitro models gave several new insights into the role of these high PSA N-CAM molecules during mammalian neuronal development.

An elegant model of regulation of neural crest migration involving both the level of fibronectin in the extracellular matrix and the expression of N-CAM was proposed by Rovasio et al. (1983). Schematically, neural crest cells start to migrate when the level of expression of N-CAM decreases. The phase of aggregation leading to the formation of the ganglion rudiments of the peripheral nervous system is characterized by the re-expression of N-CAM at the surface of neural crest derivatives.

Our study gave additional information as we have monitored the relative expression of N-CAM isoforms. In particular, we showed that the relative ratio of high PSA N-CAM to total N-CAM vanes over the peripheral nervous system development period. As an example, different ratios of high N-CAM to total N-CAM were found at Ell and at E12. This suggests distinct regulations for the expression of low PSA forms of N-CAM and for the sialylation-desialylat?n processes involved in high PSA N-CAM. In the spinal cord, from E14 to E17, total N-CAM decreased more rapidly than high PSA N-CAM and this could reflect precise regulation of sialydases and sialytransferases. However, their regulation remains largely unknown during prenatal development (Breen and Regan, 1988).

The absence of co-regulation between sialylation and synthesis of N-CAMs that we demonstrated here suggests more subtle changes during neural crest cell differentiation than those proposed earlier.

The three types of neurons that we studied, motoneurons, dorsal root ganglia cells and commissural neurons, present a homogeneous distribution of high PSA N-CAMs. Using western blot analysis, we found that DRG cells and spinal cord neurons expressed different levels of high PSA N-CAM at E12. Very little high PSA relative to total N-CAM polypeptide was found for DRG cells (Fig. 6). However, a similar homogeneous distribution of high PSA N-CAM was found in shortterm cultures of spinal cord neurons and DRG cells, on both their cell bodies and axons. These results suggest (i) that a homogeneous distribution of high PSA N-CAM can be a common feature of earlier stages of neuronal development and (ii) that a heterogeneous distribution would appear at later stages of neuronal development, as demonstrated in the developing chick visual system (Schlosshauer et al. 1984).

Neuronal derivatives obtained in vitro from neural crest cells when they are co-cultured with peripheral cells (Boisseau et al. in preparation) may permit us to test this hypothesis directly.

Using immunohistochemistry on embryo sections, we studied three distinct derivatives of neural crest cells. High PSA N-CAM expression was detectable (i) in spinal ganglia, (ii) in sympathetic ganglia but (iii) was not detectable in the aortic plexuses. For DRGs, these results are in agreement with those of Duband et al. (1985) which detected total N-CAM in dorsal root ganglia only when the neuronal differentiation had occurred.

These results showing distinct levels of high PSA N-CAM in different types of ganglia may be explained by complex interactions with extracellular matrix. For DRGs, the aggregation may be a consequence of the arrest of the migrating neural crest cells, not rigorously correlated with the expression of adhesive molecules (Duband et al. 1985). Other candidates may be responsible for ganghogenesis, for instance’ (1) disappearance of pathways of migration due to site-restricted expression of some adhesion proteins like cytotactin (Crossin et al. 1986) and/or (ii) interactions with specific molecules of extracellular matrix.

It is well established that for some substrata, high PSA N-CAM is involved in fasciculation and in neuritic growth (Rutishauser et al. 1985,1988; Landmesser et al. Using perturbation experiments, we demonstrated here that high PSA N-CAM do not appear to be involved in fasciculation or neuritic growth for neuronal derivatives of neural crest cells that grow on fibronectin, which is known to be present in neural crest migration pathways (Le Douarin, 1982; Duband et al. 1986; Thiery et al. 1990). These results support the importance of the hierarchies of adhesion molecules as proposed by Dodd and Jessell (1988). Such hierarchies can be directly studied using these in vitro neural crest models by experimental perturbations of integrins and N-CAM molecules.

The authors wish to thank Drs Cochard, Duband and Levi for theu invaluable advice on immunohistochemistry. This work was supported by grants from MRT (no. 88.C 0565 to M S.), ARC (no. 6895 to G.R.), Association Française contre les Myopathies (to G.R. and M.S.) and DRET (no. 90/197 to M.S.). We thank Dr A. M. Hill (Université d’Orsay) for the gift of polyclonal antibody against vimentin and Dr Seana O’Regan for critical reading of manuscript. S B. was supported by MRT and Association des Amis de la Science and Ligue contre le Cancer fellowships