Address correspondence and reprint requests to Melitta Schachner, Zentrum für Molekulare Neurobiologie, Martinistr. 52, D-20246 Hamburg, Germany. E-mail: firstname.lastname@example.org
Recognition molecules that carry carbohydrate structures regulate cell interactions during development and play important roles in synaptic plasticity and regeneration in the adult. Glycans appear to be involved in these interactions. We have searched for binding proteins for oligomannosidic structures using the L3 antibody directed against high mannose-type glycans in an anti-idiotypic approach. A selected monoclonal anti-idiotype antibody was used for affinity chromatography and identified basigin as a binding protein from mouse brain detergent lysates. Basigin was found to bind to high mannose-carrying cell recognition molecules, such as myelin-associated glycoprotein, L1, the β2-subunit of Na+/K+-ATPase and an oligomannosidic neoglycolipid. Furthermore, basigin was involved in outgrowth of astrocytic processes in vitro. A striking homology between the first immunoglobulin (Ig)-like domain of basigin and the fourth Ig-like domain of NCAM, previously shown to bind to oligomannosidic glycans, and the lectin domain of the mannose receptor confirms that basigin is an oligomannose binding lectin. To our knowledge this is the first report that anti-idiotypic antibodies can be used to identify binding partners for carbohydrates.
Increasing evidence indicates that cell surface carbohydrates and carbohydrate-specific binding proteins, also called lectins, mediate recognition between cells (for review see: Varki 1993; Weis and Drickamer 1996; Muramatsu 2000; Feizi 2001; Furukawa et al. 2001; Helenius and Aebi 2001). Despite the overwhelming evidence that carbohydrates and their respective receptors are involved in cell recognition, direct evidence that oligosaccharides play important functional roles in the nervous system has remained relatively sparse. Indirect support for such a function has been derived from the temporally and spatially restricted pattern of expression of cell surface carbohydrates and their putative receptors (for reviews see: Schachner and Martini 1995; Rutishauser 1998; Lander and Selleck 2000). A functional involvement of carbohydrate structures in cell recognition has been demonstrated for the spatially and temporally regulated attachment of polysialic acid to neural cell adhesion molecule NCAM (for review see: Rutishauser 1998). Further examples for carbohydrate structures in the nervous system are the HNK-1, L3, L4 and L5 epitopes which are found on several neural recognition molecules. The L5 epitope, which is identical to the Lewisx-carbohydrate (Streit et al. 1996), is present on several neural recognition molecules and is involved in extension of astrocytic processes (Streit et al. 1993) and neural induction (Roberts et al. 1991). The HNK-1 carbohydrate structure containing an unusual 3′-sulfated glucuronic acid (Chou et al. 1986) occurs on the protein backbone of many neuronal proteins (for review see: Schachner and Martini 1995; references therein) and mediates cell–cell and cell–substrate interactions (Hall et al. 1997). The high mannose-type carbohydrate epitopes L3 and L4 (Schmitz et al. 1993) present on the β2-subunit of the Na+/K+-ATPase, also called adhesion molecule on glia (AMOG), and on the neural recognition molecules L1, MAG and P0 mediate adhesion among neural cells (Kücherer et al. 1987; Fahrig et al. 1990). L1 and NCAM were found to bind to each other in cis-position via oligomannosidic glycans, with L1 expressing the glycans, and NCAM being the receptor (Horstkorte et al. 1993). Heterophilic cis interaction regulates the ability of L1 to bind L1 in trans-position. This process has been called ‘assisted homophilic binding’ (Kadmon et al. 1990). Additional support for the role of glycans and cell recognition is provided by a recent publication showing a sialic acid-dependent binding of CD24 to the L1 molecule, which functions as a siglec (sialic acid immunoglobulin lectin) in CD24 induced effects on neurite outgrowth (Kleene et al. 2001). As the expression of carbohydrate structures is developmentally regulated independently of the protein backbone, subpopulations of glycoproteins are generated that function as recognition molecules at a particular developmental stage (for review see: Varki 1993). The possibility that many yet unknown glycoproteins carrying these carbohydrates may themselves be recognition molecules raises an interest in identifying and characterizing the receptors for these oligosaccharides.
In search for receptor molecules of high mannose-type L3 oligosaccharides, we have generated anti-idiotype antibodies directed against the L3 monoclonal antibody. Two anti-idiotypic antibodies were characterized and found to bind to a brain-derived glycoprotein identified as basigin, a cell surface glycoprotein of the immunoglobulin superfamily. Basigin was shown to bind to oligomannose carrying adhesion molecules and to mediate outgrowth of astrocytic processes in vitro.
Generation and purification of monoclonal anti-idiotype antibodies
Monoclonal L3 antibody, which is an IgM rat antibody, was precipitated with 50% ammonium sulfate from supernatants of hybridoma cultures. The precipitate was dissolved in phosphate buffered saline (PBS), pH 7.3 and subjected to dialysis. Female Lou × Sprague–Dawley F1 hybrid rats (4- to 6-weeks old) were immunized with 600 μg of ammonium sulfate purified monoclonal L3 antibody. After the first subcutaneous injection in complete Freund's adjuvant (Difco, Detroit, OH, USA), booster injections were given intraperitoneally 2, 5, 32 and 37 weeks after the first injection. Four days after the last injection, spleen cells from an animal with a high titre of anti-L3 antibodies were fused with the mouse myeloma clone Ag8.653 according to the method of de St. Groth and Scheidegger (1980). Supernatants of cultured hybridoma cells were screened in 24-well plates (Nunc, Wiesbaden, Germany) by ELISA (enzyme-linked immunosorbant assay). Positive hybridoma cells were subcloned by limiting dilution and maintained in Iscove's modified medium (Gibco/BRL, Gaithersburg, MD, USA) supplemented with 10% fetal calf serum. In one fusion, the monoclonal antibodies A2B1 and A2D2 were obtained. These IgG antibodies were purified from ammonium sulfate precipitates of hybridoma culture supernatants using a protein G column (Amersham Pharmacia, Uppsala, Sweden). The monoclonal antibodies were eluted from the protein G column with 100 mm glycine/HCl buffer, pH 2.7. Eluted proteins were immediately neutralized with 1 m Tris/HCl, pH 8.3, and subjected to dialysis against PBS.
Preparation of crude membrane fractions
Brains from adult mice were homogenized in hypotonic buffer (1 mm NaHCO3, 0.2 mm MgCl2, 0.2 mm CaCl2, 1 mm spermidine, pH 7.9) containing protease inhibitors (Roche, Mannheim, Germany). All steps were carried out at 4°C. After a first centrifugation step (1500 g, 20 min) supernatants were centrifuged for 90 min at 30 000 g. The pellet containing the crude membrane fraction was resuspended in solubilization buffer containing 20 mm Tris/HCl, pH 7.2, 1 mm EDTA, 1 mm EGTA, 150 mm NaCl and 0.5% Triton X-100 and incubated for 1 h. Non-solubilized material was removed by centrifugation at 100 000 g for 1 h.
Purification of basigin
Monoclonal anti-idiotype antibody A2D2 purified by affinity chromatography on a protein G column was coupled to cyanobromide-activated Sepharose (Amersham Pharmacia) at a concentration of 6.4 mg/mL gel matrix. Detergent solubilized crude membrane proteins were passed sequentially over Sepharose 4B columns coupled with rat IgG and monoclonal antibodies to L1, NCAM, AMOG and HNK-1. The last flow through was added to the monoclonal anti-idiotype A2D2 containing column. The column was then washed with 30 column volumes of solubilization buffer, subsequently treated with 15 volumes of solubilization buffer containing 300 mm NaCl and finally washed with 15 volumes of a buffer containing 20 mm Tris/HCl, pH 8.2, 1 mm EDTA, 1 mm EGTA, 150 mm NaCl and 0.1% Na-deoxycholate. Bound material was eluted with a buffer containing 100 mm diethylamine, pH 11.4, 1 mm EDTA, 1 mm EGTA, 150 mm NaCl and 0.5% Na-deoxycholate. The eluate was immediately neutralized by addition of 1 m Tris/HCl, pH 6.8, and subjectd to dialysis against PBS.
Production of basigin-GST fusion protein
The pGEX-KG vector (kindly provided by Dr Guan, West Lafayette, Indiana, USA; Guan and Dixon, 1991) was digested with XhoI and blunted with Klenow enzyme. The 740 bp EcoNI-EcoNI fragment of the mouse basigin cDNA (kindly provided by Dr Altruda, Turin, Italy; Altruda et al. 1989) was treated with Klenow enzyme and ligated into the vector, yielding a pGEX-basigin clone expressing basigin from amino acid 1–241 in frame with the glutathione-S-transferase (GST). Expression and purification of the basigin-GST fusion protein was performed as described elsewhere (Guan and Dixon 1991). For western blot analysis, protein from 1.5 mL bacteria culture was used.
SDS–PAGE and western blot analysis
Sodium dodecyl sulfate – polyacrylamide gel electrophoresis (SDS–PAGE) analysis was performed with 7 or 10% acrylamide slab gels. Silver staining for visualization of proteins and glycoproteins was carried out according to Oakley et al. (1980). Western blot analysis of proteins isolated by SDS–PAGE was performed according to Towbin et al. (1979).
Enzymatic deglycosylation of glycoproteins
To remove N-glycosidically linked carbohydrates, 30 μg basigin was treated with 6 U peptide-N-glycosidase F (Roche) in 20 mm K+-phosphate, pH 7.3, 10 mm EDTA, 0.5% Triton X-100, 0.1% SDS, 0.1%n-octylglucoside and 1%β-mercaptoethanol for 24 h at room temperature (24°C).
Monoclonal antibodies (1 mg of the IgG fraction dissolved in 1 mL 100 mm NaHCO3, 150 mm NaCl, pH 8.4) were allowed to react with 75 μL of a 1-mg/mL dimethylsulfoxide stock solution of biotin N-hydroxysuccinimid ester (Fluka, Buchs, Switzerland) for 4 h at room temperature and dialyzed against 50 mm Tris/HCl, pH 8.0, 150 mm NaCl.
Enzyme-linked immunosorbant assay (ELISA)
To test whether the anti-idiotypic antibodies competed with the binding of L3 antibody to the oligomannosidic neoglycolipid, the ELISA was carried out as follows: 96-well microtitre polyvinylchloride plates (Falcon, Brumath, France) were coated with 50 μL oligomannosidic neoglycolipid solution (0.5 μg/mL ethanol) or ethanol alone as control. After evaporation of ethanol at room temperature, the wells were saturated with 2% fatty acid free bovine serum albumin (BSA) in 100 mm NaHCO3, pH 8.0, for 1 h. Biotinylated monoclonal L3 antibody was pre-incubated for 45 min at a final concentration of 4.5 μg/mL with different concentrations of the various monoclonal antibodies. One hundred microlitres of the pre-incubated solutions were then added to the oligomannosidic neoglycolipid-coated wells and incubated for 45 min. After three washes with PBS, 0.5 μg/mL horseradish peroxidase conjugated streptavidin was added and allowed to react for 1 h. After four washes 1 μg/mL 2,2′-azino-bis-(3-ethyl)-benzthiazoline-6-sulfonic acid (ABTS) dissolved in 100 mm sodium acetate, pH 4.2, 0.01% H2O2 was used for development. The absorption of the reaction product was measured at 405 nm.
To analyze the binding of different molecules to immunoaffinity purified basigin, another version of ELISA was used. Fifty microlitres of 100 mm NaHCO3, pH 8, containing no (control) or 0.2 μg/mL basigin was applied to plastic plates (Maxisorb, Nunc, Wiesbaden Germany) and incubated overnight at 4°C. The plates were washed with 100 mm NaHCO3 and incubated with 2% BSA in PBS for 1 h. The L3 epitope-positive or -negative molecules were then added at a concentration of 5 μg/mL for proteins and 0.05 μg/mL for glycolipids in PBS containing 0.2% BSA and incubated for 2 h. The wells were washed two times with PBS containing 0.05% Tween. The primary antibodies were incubated in the wells for 1 h. After two washes, secondary antibodies conjugated with horseradish peroxidase were added and incubated for 1 h. The wells were washed four times and developed with ABTS/H2O2. Control values were subtracted from the values obtained for the tested molecules.
Immunoaffinity purified basigin was separated by SDS–PAGE and adsorbed to a polyvinylidene difluoride (PVDF) membrane. After staining the membrane with Coomassie Brilliant Blue, visible protein bands were excised and the N-terminal end was sequenced with an Applied Biosystems 475 A sequencer (Foster City, CA, USA). Another part of the material on the membrane was cleaved on the membrane with endoproteinase Lys-C (sequencing grade, Roche). The resulting peptides were treated with iodoacetamide, extracted with 6 m guanidine hydrochloride and separated by reversed phase-HPLC.
For the identification of sequence similarities, the protein databases swiss-prot and trembl were scanned with sequence patterns using the software ScanProsite which is available in the database PROSITE under Tools for PROSITE at the Expasy Molecular Biology Server.
Microexplant outgrowth assay
Cultures of cerebellar microexplants were prepared according to Fischer et al. (1986). Briefly, glass coverslips were treated with poly l-lysine (50 μg/mL, 90 min, 37°C), washed four times with distilled water and incubated overnight at 4°C in basal medium Eagle (BME) without or with 0.7–2.0 μg basigin. The coverslips were washed with BME and cerebellar explants were plated. After 5 hours the different monoclonal antibodies were added. After another 41 h, microexplants were fixed with 4% formaldehyde (4°C, 30 min) and stained with polyclonal antibodies to GFAP. The length of the immunofluorescent GFAP-positive processes emanating radially from the tissue explant core was measured by using a graticule on a Zeiss Axiovert microscope. In each experiment, two coverslips were used for each experimental value and 50 GFAP-positive processes were measured per coverslip.
As anti-idiotype antibodies are believed to react with receptors for the epitopes that the primary antibodies recognize (Kohler et al. 1989), we used this approach to identify binding proteins of oligomannosidic glycans from detergent lysates of adult mouse brain homogenates. The anti-idiotypic antibodies directed against the rat L3 monoclonal antibody that recognize high mannose-type glycostructures were generated by isogenic immunization of rats.
Production and characterization of L3 anti-idiotype monoclonal antibodies
For the production of monoclonal antibodies directed against the idiotype of the monoclonal L3 antibody, the purified L3-IgM fraction was injected isogenically into female F1 hybrid rats of a Lou × Sprague–Dawley cross. Supernatants of resultant hybridoma cells were screened by ELISA using monoclonal L3 antibody as coated substrate (data not shown). From one fusion, two independent hybridoma clones, designated A2B1 and A2D2, were obtained which secreted antibodies binding to L3 antibody.
For further characterization of the binding specificity, the monoclonal anti-idiotypic IgG antibodies A2B1 and A2D2 were tested for their ability to interfere with the binding of biotinylated L3 antibody to an oligomannosidic neoglycolipid in an ELISA. Reduction in binding of the monoclonal L3 antibody was taken as competition for binding to the oligomannosidic carbohydrate epitope. As controls, the previously described monoclonal IgG antibody to the β2 subunit of Na+/K+-ATPase, also called AMOG (Antonicek et al. 1987), and an anti-idiotype IgM antibody to the HNK-1 carbohydrate were used. A2B1 and A2D2 IgG antibodies inhibited the binding of the L3 antibody to the substrate-coated oligomannosidic neoglycolipid in a concentration-dependent manner, thus satisfying the operational criterion of an anti-idiotype antibody (Fig. 1). As expected, neither the AMOG IgG antibody nor the HNK-1 IgM antibody influenced the binding of the monoclonal L3 antibody to the oligomannosidic neoglycolipid. The specificity of the anti-idiotype antibodies A2D2 and A2B1 was further confirmend by the finding that they did not interfere with the binding of the HNK-1 antibody to the HNK-1 neoglycolipid (data not shown).
Purification and characterization of an anti-idiotype immunoreactive protein
Triton X-100 extracts of a crude membrane fraction of adult mouse brain were used for the purification of anti-idiotype immunoreactive protein. Anti-idiotype antibody A2D2 was coupled to beads and used for immunoaffinity chromatography. The eluted protein was analyzed by SDS–PAGE under reducing conditions. A diffuse band ranging from 43 to 45 kDa could be detected by silver staining (Fig. 2a, lane 1). Western blot analysis with the antibodies A2B1 and A2D2 showed a diffuse immunoreactive band at approximately 43 kDa (Fig. 2a, lanes 2 and 3). Upon treatment with peptide-N-glycosidase F (PNGase F) and subsequent SDS–PAGE analysis a sharp band at 35 kDa, a weaker band at 32 kDa and a diffusely migrating band in the range of 38–43 kDa were observed (Fig. 2a, lane 4) showing that the immunoaffinity-purified protein is N-glycosylated.
Identification of the affinity purified glycoprotein as basigin
Sequence analysis of the blotted affinity purified glycoprotein resulted in the following sequences: an N-terminal AAGTIQTSVQEVNXKXQL and an internal sequence YVVVSTPEK. These peptides showed identity to basigin (Fig. 3). Polyclonal antibodies to a bacterially expressed recombinant fusion protein of basigin (Miyauchi et al. 1990) reacted with anti-idiotype immunoaffinity purified protein as seen by western blot analysis (Fig. 2a, lane 5). In comparison to the signal obtained with the anti-idiotype basigin antibodies, the signal intensity obtained with the polyclonal basigin antibody is lower, probably due to different antibody affinities. Furthermore, the recombinant basigin-GST fusion protein containing the extracellular part of basigin was recognized by monoclonal antibodies A2B1 and A2D2 (Fig. 2b, lanes 1 and 3) and the polyclonal basigin antibody (data not shown), whereas the GST tag was not recognized (Fig. 2b, lanes 2 and 4).
A2B1 and A2D2 antibodies recognized a second band, which showed a lower molecular weight than the basigin-GST fusion protein (Fig. 2b, lanes 1 and 3) and most likely represents a proteolytic fragment of the fusion protein containing part of the basigin. These experiments confirm that the affinity-purified glycoprotein is basigin.
Binding of basigin to oligomannosidic glycans
We then investigated whether basigin binds to oligomannosidic glycans and used an ELISA with substrate-coated basigin to test the binding of the oligomannosidic neoglycolipid and different high mannose-type positive glycoproteins from brain tissue, such as AMOG, L1 and MAG. All molecules previously identified to carry oligomannosidic carbohydrates bound specifically to basigin while the negative controls CD9 and HNK-1 neoglycolipid did not (Fig. 4).
Identification of a putative oligomannose-specific carbohydrate recognition domain in the basigin sequence
Further indication that basigin is a mannose binding protein was obtained from sequence analysis. It has previously been reported that the fourth immunoglobulin (Ig)-like domain of NCAM binds to oligomannosides and shows some sequence similarity to carbohydrate recognition domains (CRD) of the macrophage mannose receptor (Horstkorte et al. 1993). Using the consensus sequence CXXXXXX[V,I,L]X[S,T][V,I]XXXX[E,S] as a pattern to scan protein databases, several molecules were found to contain this sequence. Among these proteins was basigin from rat. A direct comparison of NCAM and basigin from rat showed that both proteins contain the pattern CXAXGXX[V,I,L][P,T] S[V,I]XXXXS. This pattern was present in NCAM from bovine, human, mouse and chicken. In basigin from mouse the conserved [V,I,L] was not found. NCAM from Xenopus and newt showed a substitution of serine by alanine at position 16 of the pattern, while basigin from rabbit showed a substitution of cysteine by serine and basigin from human showed this substitution and an additional exchange of serine to threonine in position 10 of the pattern. Consideration of all differences found in the individual sequences resulted in a more refined pattern [C,S]XAXGXXXX[S,T][V,I]XXXX [E,S,A].
Using this pattern to scan protein databases, several proteins carrying this sequence pattern either in C2-type Ig-like domains or in CRDs were found (Fig. 5). The pattern is carried by L1 from different species in the fifth Ig-like domain, by the Down's syndrome cell adhesion molecule (DSCAM) from human and mouse in the eighth Ig-like domain, by BIG-2 from rat in the second Ig-like domain, by perlecan and telencephalin from mouse and human in the fourth and fifth Ig-like domain, by human CCK-4/PTK-7 (colon carcinoma kinase 4/tyrosine protein kinase like protein 7) in the sixth Ig-like domain, by amalgam and lachesin from Drosophila in the first Ig-like domain and by roundabout 1 and roundabout 2 from Drosophila in the fourth Ig-like domain (Fig. 5). The pattern was also present in the Ig-like domain of the hypothetical K09C8.5 and C27B7.7 proteins from C. elegansand the CG8619, CG14009, CG11320, CG14828, CG5481, robo and lac proteins from Drosophila (data not shown). The macrophage mannose receptor carries the motif as part of CRD-4 and also in brevican from mouse and human the pattern is part of the single CRD (Fig. 5). Furthermore, this pattern was also found in the Q17452, Q17459 and P91429 proteins from C. elegans which are similar to C-type lectins (data not shown). Interestingly, the first of the two Ig-like domain of basigin showed an additional similarity at its C-terminus to the Ig-like domains present in the above mentioned proteins of the Ig-superfamily.
In summary, most of these Ig-like domains show the pattern [C,S]XAXGXXX[PTA][S,T][V,I,L]XXXX[E,S,A]X(n)[V,I,L]XXXDX1,2GYXCXXX[P,N,S] with n ranging from 19 to 40 residues. NCAM and basigin show the highest degree of similarity over the whole range of the pattern. This similarity includes the variable alternatively spliced exon (VASE-exon) in NCAM. The striking similarity between the first Ig-like domain of basigin (76 amino acids) and the fourth Ig-like domain of NCAM containing the VASE exon (80 amino acids) comprises 13 identical amino acids and 47 conserved amino acids with similar physico-chemical properties (tiny, small, hydophobic, polar or aliphatic; according to Taylor 1986). The observation that all identified proteins showed similarity to C-type lectins suggests that these proteins bind mannose.
Functional properties of basigin
Basigin was analyzed in its functional properties in the brain by studying cell interactions in culture. Explants of cerebellum from 7-day-old mice were maintained for 41 h on coverslips coated with poly l-lysine alone or together with basigin immunoaffinity-purified from mouse brains. Substrate-coated basigin promoted the outgrowth of astrocytic processes by approximately 20% in comparison with the poly l-lysine control, while neurite outgrowth was not influenced by basigin (data not shown). In the presence of the monoclonal basigin IgG antibodies A2B1 and A2D2 the outgrowth of astrocytic processes from explants was inhibited significantly on substrate-coated basigin but not on poly l-lysine (Fig. 6). A2B1 inhibited glial process outgrowth on basigin by approximately 60% and A2D2 by approximately 40%, whereas the control IgG antibody against AMOG showed no inhibition of astrocytic outgrowth, either on subtrate-coated basigin or on the poly l-lysine control (Fig. 6).
The specifity of basigin in promoting astrocyte outgrowth is supported by the observation that several adhesion molecules of the Ig superfamily, such as L1 and NCAM (Frei et al. 1992) do not affect astrocyte outgrowth nor do the corresponding antibodies inhibit astrocyte outgrowth (Fischer et al. 1986).
In this study we report that it is possible to isolate a carbohydrate receptor molecule from brain homogenates by generating anti-idiotypic antibodies that mimic the carbohydrate structure. These anti-idiotype antibodies were isolated by isogenic immunization and interfered with the binding of the oligomannosidic-specific antibody L3 to oligomannosidic glycans, thus indicating mimicry of the carbohydrate structure. An anti-idiotype antibody was used by affinity chromatography to isolate a glycoprotein that was identified as basigin. It is important to mention in this context that NCAM, which binds to mannose-type glycans, was not eluted from the affinity column, as it had been removed from the detergent lysate of brain homogenates with an anti-NCAM column before affinity chromatography with the anti-idiotype antibody A2D2. The binding of basigin to the anti-idiotype antibody suggested that it binds to oligomannosidic glycans.
The notion that basigin is a binding protein or receptor molecule for the high mannose-type L3 carbohydrate was supported by results obtained from binding studies and from sequence similarities. Oligomannoside carrying glycoproteins and neoglycolipids bound to basigin, whereas oligomannoside-deficient glycoproteins and neoglycolipids did not. A striking homology of the first immunoglobulin-like domain of basigin with the fourth Ig-like domain of NCAM was observed by sequence comparison. This NCAM domain was described to mediate the binding to high mannose-type sugars (Horstkorte et al. 1993). A significant similarity to the sequence motif in the first Ig-like domain of basigin was found in other recognition molecules of the immunoglobulin superfamily, such as L1, the close family member BIG−2, Down syndrome cell adhesion molecule DSCAM, colon carcinoma kinase 4/tyrosine protein kinase-like protein 7 (CCK-4/PTK-7), the integrin binding telencephalin, the extracellular matrix proteoglycan perlecan, and adhesion molecules from Drosophila, such as amalgam, lachesin and roundabout 1 and 2. In particular, basigin showed sequence homology to mannose receptors from human and mouse in their CRD, strongly suggesting that basigin is a mannose-specific lectin.
It is interesting that L1 not only contains the putative oligomannose binding domain, but also oligomannosidic carbohydrates. Thus, it is conceivable that homophilic interactions between L1 molecules are mediated, not only by the protein backbone, but also by carbohydrates, such as oligomannosides. Whether other molecules with oligomannosidic lectin binding domains also express oligomannosides and may therefore interact with each other, remains a topic for future investigations. In any case, carriers of oligomannosidic carbohydrates are potential ligands for the oligomannosidic lectin family described in this study. These heterophilic interactions may be differentially regulated during development and by different neural cell types depending on the expression of oligomannosidic glycans.
In view of the broad expression of basigin, it is not surprising that basigin is involved in various functions during development and in the adult. The pivotal role of basigin in early ontogenetic development has been revealed by a series of studies of basigin deficient mice. Most basigin null mutant embryos are lost around the time of implantation (Igakura et al. 1998). Interestingly, a small percentage of mutant survives with deficits in various organs. Spermatogenesis and survival of spermatids are severly impaired during the first meiotic division in these mice, resulting in male sterility (Maekawa et al. 1998). Adult mutant mice show deficits in spatial learning and memory (Naruhashi et al. 1997) and exhibit a decreased sensitivity to irritating odours (Igakura et al. 1998). In vitro studies have shown that basigin plays an important role in neuron–glia interactions during development of the chicken retina (Fadool and Linser 1993). It is noteworthy that basigin interacts with the β2-subunit of Na+/K+-ATPase, originally called AMOG which may serve as a ligand for basigin-dependent astrocytic outgrowth. AMOG is a highly glycosylated cell surface glycoprotein with about 80% oligomannosidic glycans (Schmitz et al. 1993). Expression of AMOG is detectable at the same time in late embryonic development as expression of the HT7 antigen, which correlates with the appearance of the blood–brain barrier (Risau et al. 1986). It is therefore tempting to speculate that the expression of oligomannoside-carrying AMOG by astrocytes may be involved in the formation of the blood–brain barrier which is known to develop in differentiating endothelial cells depending on astrocytes (for review see: Goldstein 1988). The ability of basigin to increase outgrowth of astrocytic processes in vitro may thus be related to the formation and/or maintenance of the blood–brain barrier.
In summary, it is conceivable that oligomannosidic carbohydrate and oligomannosidic interacting recognition molecules are important mediators of cis- and trans-interactions between cell surface glycoproteins in the nervous system. Whether these inter-relationships are of particular functional significance will be the topic of future investigations.
The authors are grateful to Bashar Saad for monoclonal anti-idiotype HNK-1 antibody, Brigitte Schmitz for oligomannosidic neoglycolipid and HNK-1 neoglycolipid, Fiorella Altruda (Universita di Torino, Italy) for the gp42 cDNA clone, Teruo Miyauchi and Takashi Muramatsu (Kagoshima University, Japan) for polyclonal antibodies to basigin, Kun Liang Guan (West Lafayette, USA) for the pGEX-KG vector, and Thomas Vorherr (Roche, Basel) for protein sequencing. This project was supported by the Deutsche Forschungsgemeinschaft (Scha 185/27, 1–2).