We have shown previously that the predominant N-CAM isoform in skeletal muscle myotubes contains as a result of alternative splicing a novel domain (MSD1) in its extracellular region. Here we show that this region represents a site for O-linked carbohydrate attachment. The lipid tailed N-CAM in myotubes was found to bind peanut lectin while the transmembrane isoform from myoblasts lacking MSD1 did not. In addition, N-CAM from a variety of neural sources failed to bind the lectin. Analysis of 3T3 fibroblasts transfected with various N-CAM cDNAs, showed that peanut lectin binding was correlated specifically with the expression of the MSD1 region. The oligosaccharides isolated from a purified preparation of myotube N-CAM were shown to contain an O-linked oligosaccharide whose core structure was a sialylated version of Galβ1→3GalNac which is the structure recognized specifically by peanut lectin. These data provide the first evidence for the expression of O-linked carbohydrate on any N-CAM isoform and more specifically target this oligosaccharide to the MSD1 region of myotube N-CAM.

Neural cell adhesion molecule (N-CAM) is a cell surface sialoglycoprotein that is involved in specific homophilic adhesive interactions in a variety of cells such as neurones, glia and skeletal muscle (Edelman, 1985, 1986; Rutishauser & Goridis, 1986; Nybroe et al. 1988). The gene encoding N-CAM is present as a single copy and diversity in the observed primary structure of N-CAM can be accounted for by specific patterns of alternative RNA splicing and polyadenylation site selection (Cunningham et al. 1987; Owens et al. 1987; Goridis & Wille, 1988; Walsh, 1988). N-CAM is also subject to a number of post-translational modifications involving phosphorylation, sulfation and glycosylation that can probably modify homophilic recognition events. The glycosylation patterns of N-CAM isoforms have been extensively analysed (Cunningham et al. 1983; Crossin et al. 1984). Recent primary amino acid sequence analyses suggest there are a number of potential sites for asparagine-linked (N-linked) carbohydrate attachment in N-CAM from chicken brain (Cunningham et al. 1987), mouse brain (Barthels et al. 1987) and human muscle (Barton et al. 1988), although it is not known how many of these sites are occupied. Furthermore no O-linked carbohydrate has yet been identified in any N-CAM isoforms found in brain (Lyles et al. 1984; Nybroe et al. 1988). A number of tissue- and isoform-specific glycosylation modifications have been found in N-CAM. First, the monoclonal antibodies L2 and HNK-1 recognize a carbohydrate epitope that is expressed on subpopulations of N-CAM isoforms of 180 ×103Mr (N-CAM-180) and 140×103Mr (N-CAM-140), but not of 120×103Mr (N-CAM-120) (Kruse et al. 1984). Furthermore this epitope is not found in muscle N-CAM (Sanes et al. 1986). Certain N-CAM isoforms also contain polysialic acid in an α-2,8-linkage and this modification appears to be developmentally regulated and variable between specific brain regions (Chuong & Edelman, 1984; Finne et al. 1983). Polysialic acid may be of considerable functional importance since its removal increases the rate of N-CAM homophilic interactions (Hoffman & Edelman, 1983; Rutishauser et al. 1988).

We have recently determined the primary amino acid sequence of N-CAM isoforms from skeletal muscle (Dickson et al. 1987; Barton et al. 1988) and have shown that myoblasts express predominantly a transmembrane N-CAM isoform of about 145×103Mr (N-CAM-145), while myotubes express nontransmembrane N-CAM forms of 125 and 155 ×103Mr (N-CAM-125, N-CAM-155). N-CAM-125 and N-CAM-155 are of interest as they have a glycosylphosphatidylinositol (GPI) tail and can be released from the plasma membrane by treatment with phosphatidylinositol-specific phopholipase C (PLC) (Moore et al. 1987). Furthermore, sequence analysis has identified an additional domain (MSD1) which is introduced into the extracellular coding region of myotube N-CAM mRNA by a process of alternative splicing and whose expression is tissue-specific and developmentally regulated (Dickson et al. 1987).

In the present report, we show that the GPI-tailed N-CAM isoforms in skeletal myotubes express a specific carbohydrate-binding site recognized by peanut lectin, which is not expressed by the transmembrane N-CAM isoform found in skeletal muscle myoblasts or by N-CAM from brain and a variety of neural cell lines. Analysis of the carbohydrate of myotube N-CAM revealed both N- and O-linked oligosaccharides, the latter of which was found to contain a sialylated version of galactose β l→3 N-acetylgalactosamine, the binding site for peanut lectin. Furthermore, analysis of cells transfected with different N-CAM cDNAs showed that peanut lectin binding is correlated with expression of the MSD1 region thus accounting for the tissue and isoform-specific nature of the O-linked carbohydrate attachment.

Purification of N-CAM from skeletal muscle cells

N-CAM from skeletal muscle was isolated on a column of anti-N-CAM Ig. C2 muscle cells were grown to the myotube stage (Moore et al. 1987) and then extracted with 1% NP40. The soluble fraction was passed over a preimmune rabbit Ig Sepharose 4B precolumn followed by an anti-N-CAM Ig Sepharose 4B column. Bound N-CAM was eluted with 50 mm-diethylamine pH 11·5 in 1% NP40 and was immediately neutralized with glycine. The purity of the N-CAM preparation was assessed by polyacrylamide gel electrophoresis.

Analysis of N-CAM oligosaccharide

Purified N-CAM (salt-free) was lyophilized, dried cryogenically, and subjected to hydrazinolysis as described previously (Ashford et al. 1987). The liberated oligosaccharides were re-N-acetylated, and radioactive labelled by reduction with NaB3H4(70·4 Ci mmole−1, New England Nuclear), all as described previously (Ashford et al. 1987). The resulting pool of radioactively labelled oligosaccharide alditols was then analysed by a combination of high voltage paper electrophoresis (at 80 V cm−1) in either pyridine/acetic acid/water (3:l:387v/v, pH5·4) to separate neutral and acid structures, or 15mm-sodium tetraborate, pH9·5, to fractionate mono- and disaccharide alditols. Bio-Gel P·4 (–400 mesh) gelpermeation chromatography (of desialylated structures), and sequential exoglycosidase digestion were performed as described previously (Parekh et al. 1987).

Cell culture and metabolic labelling

The C2 skeletal muscle cell line (Yaffe & Saxel, 1977) was cultured at the myoblast and myotube stage of growth as described previously (Moore et al. 1987). The N2-A neuroblastoma cell line, the C6 glioma cell line and the SWA Schwann cell line (Tenekoon et al. 1987) were grown on culture dishes in Dulbecco’s modified Eagle’s medium containing 10% fetal calf serum except that the SWA cells were supplemented with 0·1mm-ZnCl2. The SWA cell line was a generous gift from Dr G. Tenekoon, Department of Neurology, Johns Hopkins University, Baltimore MD. Metabolic labelling with [35S]methionine and pulse-chase experiments on C2 myotube cell cultures were carried out exactly as described previously (Moore et al. 1987).

Antibodies, lectins and Western blotting

The main antibodies used in the present study were rabbit anti-mouse muscle N-CAM (Moore et al. 1987) and rabbit anti-human brain N-CAM (Dickson et al. 1987). Peanut lectin labelled with horseradish peroxidase was obtained from Sigma Chemical Company, Poole, Dorset, England. Western blotting procedures were as described previously (Moore et al. 1987; Barton et al. 1988). N-CAM immunoprecipitates or purified N-CAM preparation released from cells by phospholipase C treatment were fractionated by SDS–PAGE and, after transfer to nitrocellulose, were reacted with rabbit antimouse muscle or rabbit anti-human brain N-CAM. Sheep anti-rabbit peroxidase was used as a detecting antibody. Peanut lectin labelled with horseradish peroxidase was used to detect peanut lectin binding sites on specific N-CAM forms. The Biorad HRB colour development reagent in the presence of 0·02% H2O2 was used to visualize specific binding of antibodies or lectin.

Immunoprecipitation of N-CAM polypeptides and lectin affinity chromatography

Immunoprecipitation of N-CAM polypeptides from either metabolically labelled C2 muscle cells or unlabelled muscle cells and a variety of cell lines or tissues was carried out as described previously, utilizing rabbit anti-mouse muscle N-CAM antibody (Moore et al. 1987). After immunoprecipitation, samples were resuspended in 25mm-acetate buffer, pH5·0 with 0·1i.u. neuraminidase (Sigma Type. X) incubated at 37 °C for one hour, boiled in SDS sample buffer and analysed by SDS–PAGE as described above.

Affinity chromatography using peanut lectin–agarose (Sigma Chemical Company, Poole, Dorset, England) was used to identify peanut-lectin binding N-CAM forms from metabolically labelled C2 myotube cultures. C2 myotube cells were labelled with [35S]methionine as described above, extracted with 1% NP40, followed by treatment of the cell extract with neuraminidase as described previously (Moore et al. 1987). The cell extract was then mixed with peanut lectin–agarose beads in a batch method for 2h at 4 °C. The beads were washed with NP40 buffer and glycoconjugates that were specifically bound to the column eluted with 0–1 M-D-galactose. Eluted material was then subjected to immunoprecipitation with anti-N-CAM antibody as described above along with samples of the initial NP40 extract and the material that failed to bind to the peanut lectin–agarose column. The three immunoprecipitates were fractionated by SDS–PAGE and specific N-CAM bands identified by fluorography.

Preparation of N-CAM released by phospholipase C from C2 myotubes

Cell cultures of C2 myotubes were prepared as described above. Cells were treated with phospholipase C (Sigma, type III) at a concentration of 10i.u.m1−1 in DMEM for 1h at 37°C. The supernatant was collected, filtered through a 0·22 μm filter and N-CAM immunoreactive material immunoprecipitated as above. .

Construction of full-length N-CAM cDNAs

The isolation and characterization of a full-length N-CAM cDNA (CHB1) encoding a GPI-linked, muscle-specific isoform (Barton et al. 1988), and partial clones encoding GPI-linked (λ9·5) or the membrane proximal and cytoplasmic domains of a transmembrane isoform (λ4·4) (Dickson et al.1987) of N-CAM have been previously described. In order to make a full-length construct encoding a GPI-linked N-CAM lacking MSD1 and transmembrane derivatives, use was made of an additional cDNA clone, λ5, which is identical to the 5’ EcoRI fragment of CHB1, but features a unique XmnI restriction site within its 5’ untranslated sequence. The EcoRI to HindIII fragment of λ9·5 (Dickson et al. 1987) (bases 1 to 967) was ligated into a similarly digested pUC 12 plasmid, generating p9·5E-H. Sequence incorporating MSD1 was removed from this clone by digestion with KpnI and EcoRI (bases 1 to 550) and replaced with a similar fragment from clone λ4·4 (bases 1 to 442, Dickson et al. 1987) which lacks the MSD1 sequence (bases 1 to 442) generating p9·5K-H/4·4E-K. Full-length constructs encoding lipid-linked N-CAMs with and without the MSD1 sequence were generated by ligating in the whole EcoRI insert of λ5 into the unique EcoRI sites of p9·5E-H and p9·5K-H/4·4E-K generating constructs B and A, respectively. Recombinants of the appropriate orientation were selected by standard techniques.

Transmembrane N-CAMs with and without the MSD1 sequence (constructs E and D, respectively) were prepared from the lipid-linked N-CAMs by ligating the large KpnI fragment from λ4·4 (bases 443 to 1644) into the unique KpnI sites of construct B and construct A. In addition to N-CAM sequence the KpnI fragment of λ4·4 contains a unique SstI restriction site (from the polylinker sequence) at the 3’ end which was utilized in subsequent subcloning steps.

Full-length sequences for the four constructs were excised by double digestion with XmnI and HindIII for lipid-linked isoforms and with XmnI and SstI for transmembrane N-CAMs The inserts were end-repaired with Klenow enzyme and SI nuclease for lipid-linked and transmembrane isoforms, respectively, and subcloned into the HincII site of pGEM. Recombinants were selected with 5’ and 3’ ends adjacent to the HindIII and BamHI sites of the pGEM polylinker, respectively, to enable directional cloning of the full-length-inserts to be performed into the unique HindIII and BamHI sites of p4.4.4 (pH β Apr-l-neo) for DNA tranfection studies. DNA transfections were performed by the standard calcium phosphate procedure as described previously (Gower et al.1988).

Skeletal muscle myotube N-CAM binds peanut lectin

It was shown previously that the MSD1 region found in the extracellular domain of myotube N-CAM is rich in serine and threonine residues and shows some sequence homology with the hinge region of immunoglobulin (Ig) (Dickson et al. 1987, Barton et al. 1988; Walsh, 1988), this region being a site of extensive O-linked carbohydrate attachment (Baenziger & Kornfeld, 1974). To determine whether similar post-translational modifications could be found on N-CAM, we first analysed peanut lectin binding (Lotan et al. 1975) to N-CAMs since this lectin has previously been utilized to identify O-linked carbohydrate in other membrane molecules such as the low-density lipoprotein receptor (Russell et al. 1984). N-CAM was immunoprecipitated from C2 muscle cultures and, after neuraminidase treatment, was subjected to Western blot analysis, using either anti-N-CAM (Fig. 1a–d) or peroxidase labelled peanut lectin (Fig. 1e–h). As was shown previously (Moore et al. 1987) the profile of N-CAM expression changes as C2 myoblasts (Fig. la) fuse to form myotubes (Fig. 1b). Peanut lectin binding was predominantly associated with the N-CAM isoform of 125×103Mrexpressed specifically in myotube cultures (Fig. 1f) and which can be released from the plasma membrane by treatment with PLC (Fig. 1c,g). Weak staining was sometimes found associated with a 155×103band. In contrast, no reactivity was found with the N-CAM bands expressed by myoblasts. Furthermore, fully sialylated N-CAM from myotubes did not bind peanut lectin (Fig. 1h) consistent with results reported for other membrane proteins including the low-density lipoprotein receptor (Russell et al. 1984) and a thymocyte sialoglycoprotein (Brown & Williams, 1982). The observed peanut lectin binding to myotubes was abolished by prior incubation with 0·1 M-D-galactose confirming the specificity of the lectin binding. Similar results to those obtained with peanut lectin were found also with jacalin lectin (data not shown) which binds specifically to galactose β1→3 N-acetylgalactosamine (Hagiwara et al. 1988).

Fig. 1

Western blots of immunopurified N-CAMs from C2 muscle cultures. Tracks (a) and (e) contain desialo-N-CAMs from myoblasts, tracks (b) and (f) contain desialo-N-CAMs from myotubes, while tracks (c) and (g) contain desialo–N-CAM released from C2 myotubes by phosphatidylinositol specific phospholipase C. Tracks (d) and (h) contain fully sialylated N-CAMs from C2 myotube cultures. N-CAMs were analysed for anti-N-CAM (a–d) or peanut lectin (e–h) binding. Molecular mass markers are ×10−3.

Fig. 1

Western blots of immunopurified N-CAMs from C2 muscle cultures. Tracks (a) and (e) contain desialo-N-CAMs from myoblasts, tracks (b) and (f) contain desialo-N-CAMs from myotubes, while tracks (c) and (g) contain desialo–N-CAM released from C2 myotubes by phosphatidylinositol specific phospholipase C. Tracks (d) and (h) contain fully sialylated N-CAMs from C2 myotube cultures. N-CAMs were analysed for anti-N-CAM (a–d) or peanut lectin (e–h) binding. Molecular mass markers are ×10−3.

It has recently been shown that, in cultures of C2 myotubes, the 120 and 145×103Mr N-CAM isoforms are processing intermediates of the 125 and 155×103Mrisoforms, respectively, and that only the latter fully mature isoforms are apparently released by phospholipase C treatment (Moore et al. 1987). To determine if the processing intermediates bind peanut lectin, metabolically labelled desialo N-CAMs immunopurified from C2 myotube cultures were fractionated on columns of peanut lectin agarose. The unbound material, and that retarded and eluted with galactose, were immunoprecipitated and analysed by SDS–PAGE. Of the four N-CAM bands found in C2 cells (Fig. 2, track a), only the mature products of 125 and 155×103Mrwere retarded by peanut lectin (Fig. 2, track b) whereas the bands at 120 and 145×103Mr passed through the column (Fig. 2, track c).

Fig. 2

Autoradiograph of metabolically-labelled immunoprecipitated C2 myotube N-CAMs. Track (a) contains total immunoprecipitable desialo-N-CAMs; track (b) shows N-CAMs which bound to peanut lectin–agarose and were eluted with 0·1 M-D-galactose; track (c) shows N-CAMs which did not bind to peanut lectin–agarose. Molecular mass markers are ×10−3.

Fig. 2

Autoradiograph of metabolically-labelled immunoprecipitated C2 myotube N-CAMs. Track (a) contains total immunoprecipitable desialo-N-CAMs; track (b) shows N-CAMs which bound to peanut lectin–agarose and were eluted with 0·1 M-D-galactose; track (c) shows N-CAMs which did not bind to peanut lectin–agarose. Molecular mass markers are ×10−3.

In order to address the question of whether the peanut lectin binding was specific to lipid-linked N-CAM forms in myotubes only, a variety of cell lines that express N-CAM (Pollerberg et al. 1986; He et al. 1986; Gennarini et al. 1986; G. Dickson, unpublished observations), and a sample of mouse brain were immunoprecipitated and analysed for lectin binding. Fig. 3 shows the pattern of N-CAM expression and peanut lectin binding in C2 myotubes (1), the neuroblastoma line N2-A (2), the C6 glioma cell line (3), the Schwann cell line SWA (4) and adult mouse brain (5,6). None of the desialo-N-CAM forms present in the N2-A, C6, SWA cells or adult mouse brain bound peanut lectin, nor did sialylated mouse brain N-CAM (Fig. 3b, track 5). These results indicate that peanut lectin binding is not a general feature of lipid-linked N-CAM forms, but appears to be a tissue-specific post-translational modification.

Fig. 3

Western blots of immunopurified desialo-N-CAMs analysed for anti-N-CAM binding (a) or peanut lectin binding (b). N-CAMs were isolated from C2 cells (1), N2-A cells (2), Schwannoma cells (3), C6 glioma cells (4) or mouse brain (5, 6). All N-CAMs were neuraminidase-treated except for track 5. Molecular mass markers are in ×10−3.

Fig. 3

Western blots of immunopurified desialo-N-CAMs analysed for anti-N-CAM binding (a) or peanut lectin binding (b). N-CAMs were isolated from C2 cells (1), N2-A cells (2), Schwannoma cells (3), C6 glioma cells (4) or mouse brain (5, 6). All N-CAMs were neuraminidase-treated except for track 5. Molecular mass markers are in ×10−3.

Analysis of oligosaccharides from myotube

N-CAM The ability of myotube N-CAM to bind peanut lectin and jacalin strongly suggested that myotube N-CAM carries O-linked oligosaccharides. To analyse directly the oligosaccharide pool for a compatible lectin-binding site, N-CAM was purified from C2 cell myotubes by affinity chromatography on an anti-N-CAM Ig-Sepha-rose 4B column. The preparation was analysed by SDS–PAGE and was shown to consist predominantly of N-CAM-125 and not to contain contaminating protein.

Oligosaccharides were released from myotube N-CAM by hydrazinolysis. By this procedure, N-inked oligosaccharides are recovered quantitatively, whereas the recovery of O-linked oligosaccharides is not always quantitative (Ashford et al. 1987). Following re-N-acetylation and reduction with NaB3H4, the reduced 3H-labelled oligosaccharides were subjected to high-voltage paper electrophoresis (pH 5·4), which fractionates on the basis of their charge to mass ratio (Fig. 4). Treatment of the oligosaccharide pool with neuraminidase prior to high voltage paper electrophoresis clearly changed the electrophoretic profile (Fig. 4) clearly indicating that the majority of oligosaccharides associated with skeletal muscle N-CAM are acidic by covalent association with nonreducing terminal sialic acid. The entire pool of oligosaccharides obtained after treatment with neuraminidase was subjected to Bio-Gel P-4 (–400 mesh) gel permeation chromatography, which separates oligosaccharides on the basis of their effective hydrodynamic volume (Fig. 5). Since insufficient material was available for analysis by GC-MS or n.m.r., a sample of fraction X, eluting as a single peak of elution volume 3·5 glucose units (gu), was treated with E. coli β-galactosidase and the products further separated by P-4 chromatography (Fig. 5). The entire fraction was converted to a product (Y) eluting at 2·5 gu, indicating the removal by the β-galactosidase of one nonreducing terminal β-linked galactose residue. However, fraction X was totally resistant to jack bean β-galactosidase. The sensitivity of O-linked oligosaccharide with nonreducing Galβ1→3 residues to E. coli β-galactosidase but not jack bean β-galactosidase has been reported previously (Conzelman & Kornfeld, 1984). Product Y was then mixed with a known amount of fraction X, and a known amount of [3H]-2-deoxyribitol (to serve as an internal standard during subsequent analysis), and the mixture fractionated by high-voltage paper electrophoresis (Fig. 5). Peaks were identified as deriving from [3H]-2-deoxyribitol, fraction X, or product Y by recovery from paper by elution with water and comparison of the radioactivity in each to that known to have been applied followed by analysis of their elution volume by Bio-Gel P-4 (–400 mesh) gel-permeation chromatography. Clearly, product Y comigrated with N-acetyl-galactosaminitol, and fraction X with Galβ1→3Gal-NAOL. These results are consistent with fraction X being an O-linked disaccharide of structure Galβ1→3GalNAOL.The presence of this disaccharide on oligosaccharides derived from myotube N-CAM accounts for the ability of this N-CAM isoform to bind peanut lectin and jacalin. A sialylated form(s) of this disaccharide is probably predominant on the polypeptide, since the majority of oligosaccharides were originally sialylated (Fig. 4) and no reactivity was found with peanut lectin prior to neuraminidase treatment. Detailed structural analysis of the remaining oligosaccharides is in progress, but the recovery of only [3H]-N-acetylgluco-saminitol (and not [3H]-N-acetylgalactosaminitol) following acid hydrolysis of a sample of the entire pool of oligosaccharides of elution volume above 8 gu (Fig. 5), indicated that essentially all of these are originally N-linked to polypeptide. The oligosaccharide(s) eluting with a peak position of 6·4 gu has not yet been analysed, and may therefore be either N- or O-linked to peptide.

Fig. 4

High voltage radioelectrophoretogram of radiolabelled and reduced skeletal muscle N-CAM-derived oligosaccharides prior to (a) and post (b) treatment with neuraminidase. Arrows indicate the positions to which lactitol (L) and 3 (6)’ sialyllactitol (SL) migrated.

Fig. 4

High voltage radioelectrophoretogram of radiolabelled and reduced skeletal muscle N-CAM-derived oligosaccharides prior to (a) and post (b) treatment with neuraminidase. Arrows indicate the positions to which lactitol (L) and 3 (6)’ sialyllactitol (SL) migrated.

Fig. 5

(a) Bio-Gel P-4 (–400 mesh) gel permeation chromatogram of the desialylated oligosaccharides derived from skeletal muscle N-CAM. Numerical superscripts refer to the elution position of glucose oligomers in glucose units (gu). V is the void position, (b) Bio-Gel P-4 (–400 mesh) gel permeation chromatogram of the product generated after incubation of fraction X (a) with jack bean j3-galactosidase. High voltage radioelectrophoretogram of a mixture of [3H]-2-deoxyribitol, fraction X (a), and product Y (b). Indicated are the positions to which known standard monodisaccharide alditols migrated. (2-DR, 2-deoxyribitol; GalNAcOL, N-acetylgalactosaminitol; GICNACOL, N-acetylglucosaminitol; XylOL, xylitol; GICOL, glucitol). (d) High voltage radioelectrophoretogram of a double acid hydrolysate (1M-HC1, 100 °C, 2 h) of an aliquot of the reduced radioactive oligosaccharides eluting above 8gu in (a). The migration positions of standard mono and disaccharide alditols are shown in (c). The single peak of radioactivity migrates to the same position as GICNACOL.

Fig. 5

(a) Bio-Gel P-4 (–400 mesh) gel permeation chromatogram of the desialylated oligosaccharides derived from skeletal muscle N-CAM. Numerical superscripts refer to the elution position of glucose oligomers in glucose units (gu). V is the void position, (b) Bio-Gel P-4 (–400 mesh) gel permeation chromatogram of the product generated after incubation of fraction X (a) with jack bean j3-galactosidase. High voltage radioelectrophoretogram of a mixture of [3H]-2-deoxyribitol, fraction X (a), and product Y (b). Indicated are the positions to which known standard monodisaccharide alditols migrated. (2-DR, 2-deoxyribitol; GalNAcOL, N-acetylgalactosaminitol; GICNACOL, N-acetylglucosaminitol; XylOL, xylitol; GICOL, glucitol). (d) High voltage radioelectrophoretogram of a double acid hydrolysate (1M-HC1, 100 °C, 2 h) of an aliquot of the reduced radioactive oligosaccharides eluting above 8gu in (a). The migration positions of standard mono and disaccharide alditols are shown in (c). The single peak of radioactivity migrates to the same position as GICNACOL.

Peanut lectin binding is associated specifically with the MSD1 region on myotube N-CAM

In order to determine the site of peanut lectin binding on myotube N-CAM, 3T3 fibroblasts were transfected with full-length cDNA constructs (Fig. 6) encoding brain N-CAM-120 (construct A) and myotube N-CAM-125 (construct B). These constructs are identical except for the addition of the MSD1 region in construct B and both N-CAMs are linked to the plasma membrane via a GPI tail. N-CAM expressed by transfected cells was immunoprecipitated and analysed by Western blotting using anti-N-CAM or peanut lectin conjugated with peroxidase. Cells transfected with construct B generated a larger desialo N-CAM than construct A transfectants, compatible with the inclusion of MSD1 (Fig. 7). Furthermore, peanut lectin binding was restricted to this larger isoform (Fig. 7), indicating that the MSD1 region contains the lectin-binding site. To further analyse the specificity of expression of the peanut lectin binding and to formally disassociate its expression from the carbohydrate associated with GPI tails, analyses were performed on two additional transfectants (Fig. 6). These were a transmembrane N-CAM isoform lacking MSD1 (construct D) and an insertional mutant of this construct with the MSD1 region (construct E). Fig. 7 shows that the desialo N-CAM encoded by construct D was smaller than that encoded by construct E which is compatible with the notion that the size difference resulted from the insertion of the MSD1 region. Furthermore, the N-CAM synthesized by construct E exhibited peanut lectin binding whereas construct D did not. The most likely explanation of these data is that O-linked carbohydrate is directly associated with the MSD1 region normally present in myotube N-CAM and that the MSD1 region itself rather than any other structural perturbation subsequent to insertion of the MSD1 region signals the addition of the O-linked oligosaccharide. Thus the specificity of the O-linked glycosylation mechanism depends on the primary structure of the N-CAM molecule which is controlled at the level of alternative splicing of the primary N-CAM transcript.

Fig. 6

(a) Human skeletal muscle cDNA clones used in the construction of physiological and mutant N-CAM isoforms. The clones are aligned relative to the full-length clone, CHB1, encoding a GPI-linked N-CAM-125 (Barton et al. 1988). Only the essential restriction endonuclease cleavage sites used in the subcloning steps are indicated: E-EcoRI, H-HindIII, K-KpnI, S-SstI, X-XmnI N-CAM encoding sequences are represented as bars, common to all isoforms (open), GPI-linked isoform-specific (vertical lines), transmembrane-isoform-specific (horizontal lines) and MSD1 (solid). Triangles below the restriction maps indicate the positions of alternative splicing within the COOH-terminal domain involving both MSD1 and alternative forms of membrane attachment, (b) A diagrammatic representation of the four constructs is shown with the NH2 to COOH orientation left to right and with the membrane region hatched. The essential features depicted include the five C2 immunoglobulin homology domains (Williams, 1987), MSD1 (Dickson et al. 1987), the six potential N-linked carbohydrate attachment sites (arrowed) and alternative membrane association either via a GPI-linkage (A, B) or membrane-spanning (D, E regions). Also shown are the putative phosphorylation site(s) within the cytoplasmic domain of transmembrane N-CAMs.

Fig. 6

(a) Human skeletal muscle cDNA clones used in the construction of physiological and mutant N-CAM isoforms. The clones are aligned relative to the full-length clone, CHB1, encoding a GPI-linked N-CAM-125 (Barton et al. 1988). Only the essential restriction endonuclease cleavage sites used in the subcloning steps are indicated: E-EcoRI, H-HindIII, K-KpnI, S-SstI, X-XmnI N-CAM encoding sequences are represented as bars, common to all isoforms (open), GPI-linked isoform-specific (vertical lines), transmembrane-isoform-specific (horizontal lines) and MSD1 (solid). Triangles below the restriction maps indicate the positions of alternative splicing within the COOH-terminal domain involving both MSD1 and alternative forms of membrane attachment, (b) A diagrammatic representation of the four constructs is shown with the NH2 to COOH orientation left to right and with the membrane region hatched. The essential features depicted include the five C2 immunoglobulin homology domains (Williams, 1987), MSD1 (Dickson et al. 1987), the six potential N-linked carbohydrate attachment sites (arrowed) and alternative membrane association either via a GPI-linkage (A, B) or membrane-spanning (D, E regions). Also shown are the putative phosphorylation site(s) within the cytoplasmic domain of transmembrane N-CAMs.

Fig. 7

Western blot of immunopurified N-CAM from 3T3 cells transfected with individual N-CAM isoforms. Tracks (a) and (e) contain desialo-N-CAM from a GPI-linked N-CAM-120 brain isoform. Tracks (b) and (f) contain desialo-N-CAM from the GPI-linked N-CAM-125 muscle isoform containing the MSD1 region. Tracks (c) and (g) contain desialo-N-CAM from a transmembrane N-CAM isoform. Tracks (d) and (h) contain N-CAM from an insertional mutant of the transmembrane isoform that contains the MSD1 region. N-CAMs were analysed for reactivity with human-specific anti-N-CAM (tracks a–d) or peanut lectin (tracks d–h). (A single non-N-CAM immunoreactive band of 170×103Mr which binds peanut lectin, appears in all immunoprecipitates.) Molecular mass markers are ×10−3.

Fig. 7

Western blot of immunopurified N-CAM from 3T3 cells transfected with individual N-CAM isoforms. Tracks (a) and (e) contain desialo-N-CAM from a GPI-linked N-CAM-120 brain isoform. Tracks (b) and (f) contain desialo-N-CAM from the GPI-linked N-CAM-125 muscle isoform containing the MSD1 region. Tracks (c) and (g) contain desialo-N-CAM from a transmembrane N-CAM isoform. Tracks (d) and (h) contain N-CAM from an insertional mutant of the transmembrane isoform that contains the MSD1 region. N-CAMs were analysed for reactivity with human-specific anti-N-CAM (tracks a–d) or peanut lectin (tracks d–h). (A single non-N-CAM immunoreactive band of 170×103Mr which binds peanut lectin, appears in all immunoprecipitates.) Molecular mass markers are ×10−3.

A number of isoform-specific and developmentally regulated glycosylation patterns which profoundly influence biochemical and cellular functions have been described for brain N-CAM species (Nybroe et al. 1988). In skeletal muscle, however, despite the complex regulation of core N-CAM polypeptide isoforms which is closely coordinated with developmental status (Moore et al. 1987; Walsh, 1988) similar variations in post-translational modification have not been identified (Sanes et al. 1986). The striking observation in the present study is the detection on certain skeletal muscle N-CAM forms of an O-linked carbohydrate structure(s) which forms a binding site(s) for peanut lectin. Peanut lectin binding is a property of GPI-linked N-CAMs present in differentiated myotubes and is associated with the operation of a muscle-specific alternative splicing pathway which introduces a serine- and threonine-rich membrane-proximal putative-hinge region (MSD1) into the extracellular domain of these isoforms. Thus in skeletal myogenesis, isoform switching from transmembrane to GPI-tailed N-CAM species is associated not only with tissue-specific alteration in the primary structure, but also in the carbohydrate composition of mature N-CAM glycoproteins. Analysis of different human N-CAM isoforms in 3T3 cells stably transfected with different N-CAM cDNAs showed clearly that the expression of the MSD1 region is correlated with the expression of peanut lectin binding in N-CAM. The other main region on the N-CAM structure that is a possible site of O-glycosylation is the GPI-tail which also contains carbohydrate. However, no evidence was found to support this notion since the GPI-tailed N-CAM-120 form did not bind peanut lectin. In addition, we showed that an insertional mutant of a transmembrane N-CAM isoform containing the MSD1 region was also able to bind peanut lectin whereas the parental isoform did not. It is clear that the MSD1 region must contain the recognition signals for O-linked carbohydrate addition rather than the cellular environment since the modification can be carried out by both 3T3 fibroblasts and myotubes. The simplest explanation of the data is that the O-linked oligosaccharide is attached to the MSD1 region of the N-CAM polypeptide backbone and is not associated with a cell- or tissue-specific glycosylation of regions of the N-CAM molecule that are common to all isoforms. We cannot, however, formally exclude the possibility that glycosylation of a related site on N-CAM occurs as a consequence of insertion of the MSD1 region. Recent studies, however, indicate that the MSD1 region is indeed glycosylated since antibodies generated to the deduced amino acid sequence of this region only bind to myotubes or MSD1 expressing transfected cells after removal of carbohydrate (Thompson et al. 1989).

Different types of O-linked oligosaccharides contain N-acetylgalactosamine linked to serine or threonine residues but there is considerable scope for diversity in oligosaccharide structure thereafter (Kornfeld & Kornfeld, 1980). Skeletal muscle N-CAM expresses the most common form of O-linked carbohydrate which is a standard mucin type with a sialylated Gal β1 →3 N-acetylgalactosamine backbone. Interestingly this structure is present on the mature form of the LDL receptor (Cummings et al. 1983) and Ig (Baenziger & Kornfeld, 1974) where it is associated with a polypeptide domain containing a prevalence of threonine and serine residues homologous to the MSD1 region in muscle N-CAM.

While carbohydrate structure is known to clearly influence N-CAM function at the biochemical and cellular levels (Edelman, 1986; Rutishauser et al. 1988), the function of the O-linked oligosaccharides in myotube N-CAM shown here, and indeed overall expression of the MSD1 region, remain as yet undefined. With the LDL receptor, increased turnover has been noted on O-glycosylation-defective mutant cells suggesting a possible role for this modification in stabilizing membrane molecules and hence increasing surface expression (Kozarsky et al. 1988). This may be of considerable importance in the case of N-CAM since small variations in the surface density of N-CAM profoundly affect the strength of homophilic binding and the rate of cell–cell adhesion. With similar effect, the MSD1 region could also influence local N-CAM concentration within the plasma membrane by regulating side-to-side interactions to produce microdomains or molecular gradients with consequent effects on local cell binding capacity. Alternatively, the MSD1 region and its associated carbohydrate might directly influence or expand the ligand properties of N-CAM, either by directly modifying homophilic binding or by enhancing its capacity to interact with the extracellular matrix or other specific cell types such as neurones and Schwann cells.

Cell transfection approaches involving N-CAM gene constructs encoding MSD1 and also appropriate mutated constructs will allow hypotheses concerning the. role of this region to be directly addressed using in vitro models of cell–cell and cell–substratum adhesion and in addition more complex systems designed to measure cellular interactions such as neurite outgrowth, myoblast fusion and Schwann cell migration and proliferation.

We thank Dr Tony Magee, NIMR, Mill Hill for suggesting a number of experiments. This work was supported by the Muscular Dystrophy Group of Great Britain and the Wellcome Trust. F. S. Walsh is a Wellcome Trust Senior Lecturer. The Oxford Oligosaccharide Group (R.B.P., R.A.D. and T.W.R.) is supported by Monsanto Company, USA.

Ashford
,
D.
,
Dwek
,
R. A.
,
Welply
,
J. K.
,
Amatayakul
,
S.
,
Homans
,
S. W.
,
Lis
,
H.
,
Taylor
,
G. N.
,
Sharon
,
N.
&
Rademacher
,
T. W.
(
1987
).
The beta 1-2-D-xylose and alpha 1-3-L-fucose substitued N-linked oligosaccharides from Erythrina crystagalli lectin. Isolation, characterisation and comparison with other legume lectins
.
Eur. J. Biochem
166
,
311
320
.
Baenziger
,
J.
&
Kornfeld
,
S.
(
1974
).
Structure of the carbohydrate units of IgAI immunoglobulin. II. Structure of the O-glycosidically linked oligosaccharide chains
.
J. biol. Chem
.
249
,
7270
7281
.
Barthels
,
D.
,
Santoni
,
M-J.
,
Wille
,
W.
,
Ruddert
,
C.
,
Chaix
,
JC.
,
Hirsch
,
M-R.
,
Fontecilla-Camps
,
J. C.
&
Goridis
,
C.
(
1987
).
Isolation and nucleotide sequence of mouse N-CAM cDNA that codes for a Mr 79 000 polypeptide without a membrane-spanning region
.
EM BO J
6
,
907
914
.
Barton
,
C. H.
,
Dickson
,
G.
,
Gower
,
H. J.
,
Rowett
,
L. H.
,
Putt
,
W.
,
Elsom
,
V. L.
,
Moore
,
S. E.
,
Goridis
,
C.
&
Walsh
,
F. S.
(
1988
).
Complete sequence and in vitro expression of tissuespecific phosphatidylinositol-linked N-CAM isoform in skeletal muscle
.
Development
104
,
165
173
.
Brown
,
W. R. A.
&
Williams
,
A. F.
(
1982
).
Lymphocyte cell surface glycoproteins which bind to soybean and peanut lectins
.
Immunology
46
,
713
726
.
Chuong
,
C.M.
&
Edelman
,
G. M.
(
1984
).
Alterations in neural cell adhesion molecules during development of different regions of the nervous system
.
J. Neurosci.
4
,
2354
2368
.
Conzelmann
,
A.
&
Kornfeld
,
S.
(
1984
).
Beta-linked N-acetylgalactosamine residues present at the non-reducing termini of O-linked oligosaccharides of a cloned murine cytotoxic T lymphocyte line are absent in a Vicia villosa lectin-resistant mutant cell line
.
J. biol. Chem
.
259
,
12536
12542
.
Crossin
,
K. L.
,
Edelman
,
G. M.
&
Cunningham
,
B. A.
(
1984
).
Mapping of three carbohydrate attachment sites in embryonic and adult forms of the neural cell adhesion molecule
.
J. Cell Biol
.
99
,
1848
1855
.
Cummings
,
R. D.
,
Kornfeld
,
S.
,
Schneider
,
W. J.
,
Hobgood
,
K. K.
,
Tolleshaug
,
H.
,
Brown
,
M. S.
&
Goldstein
,
J. L.
(
1983
).
Biosynthesis of N- and O-linked oligosaccharides of the low density lipoprotein receptor
.
J. Biol Chem
.
258
,
15
261-15 273.
Cunningham
,
B. A.
,
Hemperly
,
J. J.
,
Murray
,
B. A.
,
Prediger
,
E. A.
,
Brackenbury
,
R.
&
Edelman
,
G. M.
(
1987
).
Neural cell adhesion molecule: structure, immunoglobulin-like domains, cell surface modulation and alternative RNA splicing
.
Science
236
,
799
806
.
Cunningham
,
B. A.
,
Hoffman
,
S.
,
Rutishauser
,
U.
,
Hemperly
,
J. J.
&
Edelman
,
G. M.
(
1983
).
Molecular topography of the neural cell adhesion molecule N-CAM: surface orientation and location of sialic acid-rich and binding regions
.
Proc. natn. Acad. Sci. U.S.A
.
80
,
3116
3120
.
Dickson
,
G.
,
Gower
,
H. J.
,
Barton
,
C. H.
,
Prentice
,
H. M.
,
Elsom
,
V. L.
,
Moore
,
S. E.
,
Cox
,
R. D.
,
Quinn
,
C. A.
,
Putt
,
W.
&
Walsh
,
F. S.
(
1987
).
Human muscle neural cell adhesion molecule (N-CAM): identification of a muscle specific sequence in the extracellular domain
.
Cell
50
,
1119
1130
.
Edelman
,
G. M.
(
1985
).
Cell adhesion and the molecular processes of morphogenesis
.
A. Rev. Biochem
.
54
,
135
169
.
Edelman
,
G. M.
(
1986
).
Cell adhesion molecules in the regulation of animal form and tissue pattern
.
A. Rev. Cell Biol
.
2
,
81
116
.
Finne
,
J.
,
Finne
,
V.
,
Deagostini-Bazin
,
H.
&
Goridis
,
C.
(
1983
).
Occurrence of alpha-2-8 linked polysialosyl units in a neural cell adhesion molecule
.
Biochem. biophys. Res. Commun
.
112
,
482
487
.
Gennarini
,
G.
,
Hirch
,
M-R.
,
He
,
H-T.
,
Hirn
,
M.
,
Finne
,
J.
&
Goridis
,
C.
(
1986
).
Differential expression of mouse neural cell adhesion molecule mRNA species during brain development and in neural cell lines
.
J. Neurosci
6
,
1983
1990
.
Goridis
,
C.
&
Wille
,
W.
(
1988
).
The three size classes of mouse N-CAM proteins arise from a single gene by a combination of alternative splicing and use of different polydenylation sites
.
Neurochem Internat
.
12
,
269
272
.
Gower
,
H. J.
,
Barton
,
C. H.
,
Elsom
,
V. L.
,
Thompson
,
J.
,
Moore
,
S. E.
,
Dickson
,
G.
&
Walsh
,
F. S.
(
1988
).
Alternative splicing generates a secreted form of N-CAM in muscle and brain
.
Cell
55
,
955
964
.
Hagiwara
,
K.
,
Collet-Cassart
,
D.
,
Kobayashi
,
K.
&
Vaerman
,
J-P.
(
1988
).
Jacalin: isolation, characterisation and influence of various factors on its interaction with human IgAI, as assessed by precipitation and latex agglutination
.
Molec. Immunol
.
25
,
69
83
.
He
,
H-T.
,
Barbet
,
J.
,
Chaix
,
J.C.
&
Goridis
,
C.
(
1986
).
Phosphatidylinositol is involved in the membrane attachment of N-CAM-120, the smallest component of the neural cell adhesion molecule
.
EMBO J
5
,
2489
2494
.
Hoffman
,
S.
&
Edelman
,
G. M.
(
1983
).
Kinetics of homophilic binding by embryonic and adult forms of the neural cell adhesion molecule
.
Proc natn. Acad. Set. U.S.A
.
80
,
5762
5766
.
Kornfeld
,
R.
&
Kornfeld
,
S.
(
1980
).
Structure of glycoproteins and their oligosaccharide units in the Biochemistry of Glycoproteins and proteoglycans, ed
W. J.
Lennarz
.
Plenum
.
New York
, pp.
1
34
.
Kozarsky
,
K.
,
Kingsley
,
D.
&
Krieger
,
M.
(
1988
).
Use of a mutant cell line to study the kinetics and function of O-linked glycosylation of low density lipoprotein receptors
.
Proc. natn. Acad. Sci. U.S.A
.
85
,
4335
4339
.
Kruse
,
J.
,
Mailhammer
,
R.
,
Wernecke
,
H.
,
Faissner
,
A.
,
Sommer
,
L
,
Goridis
,
C.
&
Schachner
,
M.
(
1984
).
Neural cell adhesion molecules and myelin-associated glycoprotein share a common carbohydrate moiety recognised by monoclonal antibodies L2 and HNK-1
.
Nature
311
,
153
155
.
Lotan
,
R.
,
Shutelsky
,
E.
,
Danon
,
D.
&
Sharon
,
N.
(
1975
).
The purification, composition and specificity of the anti-T lectin from peanut (Arachis hypogaea)
.
J. biol. Chem
.
250
,
8518
8523
.
Lyles
,
J. M.
,
Linnemann
,
D.
&
Bock
,
E.
(
1984
).
Biosynthesis of the D2-cell adhesion molecule. Posttranslational modifications, intracellular transport and developmental changes
.
J. Cell Biol
.
99
,
2082
2091
.
Moore
,
S. E.
,
Thompson
,
J.
,
Kirkness
,
V.
,
Dickson
,
J. G.
&
Walsh
,
F. S.
(
1987
).
Skeletal muscle neural cell adhesion molecule: changes in protein and mRNA species during myogenesis of muscle cell lines
.
J. Cell Biol
.
105
,
1377
1386
.
Nybroe
,
O.
,
Linnemann
,
D.
&
Bock
,
E.
(
1988
).
N-CAM biosynthesis in brain
.
Neurochem Internat
.
12
,
251
262
.
Owens
,
G. C.
,
Edelman
,
G. M.
&
Cunningham
,
B. A.
(
1987
).
Organisation of the neural cell adhesion molecule (N-CAM) gene: alternative exon usage as the basis for different membrane-associated domains
.
Proc. natn. Acad. Sci. U.S.A
.
84
,
294
298
.
Parekh
,
R. B.
,
Tse
,
A. G. D.
,
Dwek
,
R. A.
,
Williams
,
A. F.
&
Rademacher
,
T. W.
(
1987
).
Tissue-specific N-glycosylation sitespecific oligosaccharide patterns and lentil lectin recognition of rat Thy-1
.
EMBO. J
.
6
,
1233
1244
.
Pollerberg
,
E. G.
,
Sadoul
,
R.
,
Goridis
,
C.
&
Schachner
,
M.
(
1985
).
Selective expression of the 180Kd component of the neural cell adhesion molecule N-CAM during development
.
J. Cell Biol
.
101
,
1921
1929
.
Russell
,
D. W.
,
Schneider
,
W. J.
,
Kamamoto
,
T.
,
Luskey
,
K. L.
,
Brown
,
M. S.
&
Goldstein
,
J. L.
(
1984
).
Domain map of the LDL receptor: sequence homology with the epidermal growth factor precursor
.
Cell
37
,
577
585
.
Rutishauser
,
U.
,
Acheson
,
A.
,
Hall
,
A. K.
,
Mann
,
D. M.
&
Sunshine
,
J.
(
1988
).
The neural cell adhesion molecule (N-CAM) as a regulation of cell-cell interactions
.
Science
240
,
53
57
.
Rutishauser
,
U.
&
Goridis
,
C.
(
1986
).
N-CAM: the molecule and its genetics
.
Trends Genet
2
,
72
76
.
Sanes
,
J. R.
,
Schachner
,
M.
&
Covault
,
J.
(
1986
).
Expression of several adhesive macromolecules (N-CAM, LI, JI, Uvomorulin, Laminin, Fibronectin, and a Heparan Sulfate Proteoglycan) in embryonic, adult and denervated adult skeletal muscle
.
J. Cell Biol
.
102
,
420
431
.
Tenekoon
,
G.
,
Yoshino
,
J.
,
Peden
,
K. W. L.
,
Bigbee
,
J.
,
Rutkowski
,
J. L.
,
Kishimoto
,
Y.
,
De Vries
,
G. H.
&
McKhann
,
G. M.
(
1987
).
Transfection of neonatal rat Schwann cells with SV-40 large T antigen gene under control of the metallothionein promoters
.
J. Cell Biol
.
105
,
2315
2325
.
Thompson
,
J.
,
Dickson
,
G.
,
Moore
,
S. E.
,
Gower
,
H. J.
,
Putt
,
W.
,
Kenimer
,
J. G.
,
Barton
,
C. H.
&
Walsh
,
F. S.
(
1989
).
Alternative splicing of the neural cell adhesion molecule (N-CAM) gene generates variant extracellular domain structure in skeletal muscle and brain
.
Genes and Development (in press)
.
Walsh
,
F. S.
(
1988
).
The N-CAM gene is a complex transcriptional unit
.
Neurochem Internat
.
12
,
263
267
.
Williams
,
A. F.
(
1987
).
A year in the life of the immunoglobulin superfamily
.
Immunol Today
8
,
298
303
.
Yaffe
,
D.
&
Saxel
,
O.
(
1977
).
Serial passaging and differentiation of myogenic cells isolated from dystrophic mouse muscle
.
Nature, Lond
.
270
,
725
727
.