We have previously shown that the neural adhesion molecules L1 and NCAM interact with each other to form a complex which binds more avidly to L1 than L1 to L1 alone (Kadmon, G., A. Kowitz, P. Altevogt, and M. Schachner. 1990a. J. Cell Biol. 110:193-208). This cis-association between L1 and NCAM is carbohydrate-dependent (Kadmon, G., A. Kowitz, P. Altevogt, and M. Schachner. 1990b. J. Cell Biol. 110:209-218). In the present study, we report that L1 and NCAM bind to each other via oligomannosidic carbohydrates expressed by L1, but not by NCAM, as shown in several experiments: (a) complex formation between L1 and NCAM is inhibited by a mAb to oligomannosidic carbohydrates and by the oligosaccharides themselves; (b) NCAM binds to oligomannosidic carbohydrates; (c) within the L1/NCAM complex, the oligomannosidic carbohydrates are hidden from accessibility to a mAb against oligomannosidic carbohydrates; (d) the recombinant protein fragment of NCAM containing the immunoglobulin-like domains and not the fragment containing the fibronectin type III homologous repeats binds to oligomannosidic glycans. Furthermore, the fourth immunoglobulin-like domain of NCAM shows sequence homology with carbohydrate recognition domains of animal C-type lectins and, surprisingly, also with plant lectins. A peptide comprising part of the C-type lectin consensus sequence in the fourth immunoglobulin-like domain of NCAM interferes with the association between L1 and NCAM. The functional importance of oligomannosidic glycans at the cell surface was shown for neurite outgrowth in vitro. When neurons from early postnatal mouse cerebellum were maintained on laminin or poly-L-lysine, neurite outgrowth was inhibited by oligomannosidic glycans, by glycopeptides, glycoproteins, or neoglycolipids containing oligomannosidic glycans, but not by nonrelated oligosaccharides or oligosaccharide derivates. Neurite outgrowth was also inhibited by the peptide comprising part of the C- type lectin consensus sequence in the fourth immunoglobulin-like domain of NCAM. The combined results suggest that carbohydrate-mediated cis- associations between adhesion molecules at the cell surface modulate their functional properties.

Based on the observation that in adult mice the carbohydrate epitope L2/HNK-1 is detectable on Schwann cells in ventral spinal roots, but only scarcely in dorsal roots (Martini et al., Dev. Biol., 129, 330 - 338, 1988), the possibility was investigated that the carbohydrate is involved in the outgrowth of regenerating motor neuron axons on peripheral nerve substrates expressing the epitope. To monitor whether the L2 carbohydrate remains present during the time periods in which regenerating axons penetrate the denervated distal nerve stumps, the expression of L2 in motor and sensory branches of the femoral nerve was investigated in normal animals and after a crush lesion. During the first two postoperative weeks, L2 immunoreactivity remained high in the myelinating Schwann cells of the motor branch, whereas L2 immunoreactivity was virtually absent in the sensory branch. In a first experimental approach, cryosections of ventral and dorsal spinal roots and of motor and sensory nerves of adult rats and mice were used as substrates for neurite outgrowth. Neurites of motor neurons from chicken embryos were approximately 35% longer after 30 h of maintenance on ventral roots than on dorsal roots. Neurites from sensory neurons had the same length on dorsal as on ventral motors and were as long as neurites from motor neurons grown on dorsal roots. L2 antibodies reduced neurite outgrowth of motor neurons on ventral roots but not on dorsal roots. Neurite outgrowth of sensory neurons on both roots was not altered by the antibodies. Neurite outgrowth of motor neurons on a mixture of the extracellular matrix glycoprotein laminin and the L2 carbohydrate-carrying glycolipid was significantly higher than on the laminin substrate mixture with GD1b ganglioside or sulphatide. L2 antibodies reduced neurite outgrowth of motor neurons by 50% on the L2 glycolipid, but not on GD1b or sulphatide. These observations indicate that the L2 carbohydrate promotes neurite outgrowth of motor neurons in vitro and may thus contribute to the preferential reinnervation of motor nerves by regenerating motor axons in vivo.

The beta-amyloid precursor protein (APP) has been implicated in the etiology of Alzheimer's disease (Kang et al.: Nature 325:733-736, 1987; Selkoe: Science 248:1058-1060, 1990; Selkoe: In Cowan et al. (eds): "Annual Review of Neuroscience." Palo Alto, CA: Annual Reviews, Inc., pp 489-519, 1994) and numerous studies have shown that beta-amyloid is involved in amyloid plaque formation (Rumble et al.: N Engl J Med 320:1446-1452, 1989; Sisodia et al.: Science 248: 492-495, 1990). Evidence is presented that APP is modified with N-acetylglucosamine linked to cytoplasmic serine or threonine residues (O-GlcNAc). This is the first report of a plasma membrane protein modified with this carbohydrate. It has been postulated that this modification, which is ubiquitous in all organisms studied to date except bacteria (Haltiwanger et al.: Biochem Soc Trans 20:264-269, 1992; Dong et al.: J Biol Chem 268:16679-16687, 1993; Elliot et al.: J Neurosci 13:2424-2429, 1993; Kelly et al.: J Biol Chem 268:10416-10424, 1993), may function as an alternative to phosphorylation (Dong et al., 1993) and is involved in the multimerization of proteins (Haltiwanger et al., 1992; Dong et al., 1993). O-GlcNAc occurs at "PEST" sequences (Rogers et al.: Science 234:364-368, 1986) and it has been suggested that this modification within such a sequence leads to increased proteolytic stability of the molecule (Dong et al., 1993).(ABSTRACT TRUNCATED AT 250 WORDS)

We describe here a novel monoclonal antibody (mab H6) which recognizes CD9, an integral cell surface constituent previously described in cells of the hematopoietic lineage and involved in the aggregation of platelets. Mab H6 was raised against membranes of immature mouse astrocytes and reacted with a protein of 25-27 kD in detergent extracts of adult mouse brain membranes. Sequence analysis of the N-terminal amino acids revealed an identity of 96% with CD9 from mouse kidney. CD9 was localized in the central and peripheral mouse nervous systems: in the spinal cord of 11-day-old mouse embryos, CD9 was strongly expressed in the floor and roof plates. In the adult mouse sciatic nerve, myelin sheaths were highly CD9-immunoreactive. Mab H6 reacted with the cell surfaces of both glial cells and neurons in culture and inhibited migration of neuronal cell bodies, neurite fasciculation and outgrowth of astrocytic processes from cerebellar microexplants. Neurite outgrowth from isolated small cerebellar neurons was increased in the presence of mab H6 on substrate-coated laminin, but not on substrate-coated poly-L-lysine. Addition of mab H6 elicited an increase in intracellular Ca2+ concentration in these cells on substrate-coated laminin. Immunoprecipitates of CD9 from cultured mouse neuroblastoma N2A cells contained the alpha 6/beta 1 integrin. Moreover, preparations of CD9 immunoaffinity-purified from adult mouse brain using a mab H6 column contained the neural adhesion molecule L1, but not other neural adhesion molecules. CD9 bound to L1, but not to NCAM or MAG. Both the alpha 6/beta 1 integrin and L1 could be induced to coredistribute with CD9 on the surface of cultured neuroblastoma N2A cells. The combined observations suggest that CD9 can associate with L1 and alpha 6/beta 1 integrin to influence neural cell interactions in vitro.

The novel intracellular carbohydrate O-linked N-acetylglucosamine (O-GlcNAc) is present on proteins ranging from those of viruses to those of humans and include cytosolic, nuclear and plasma-membrane proteins. In this report we have examined the effect of manipulation of phosphorylation on the levels of O-GlcNAc in cerebellar neurons from early postnatal mice. Our results indicate a reciprocal response of O-GlcNAc levels to phosphorylation. Activation of protein kinase A or C, for example, results in reduced levels of O-GlcNAc specifically in the fraction of cytoskeletal and cytoskeleton-associated proteins, while inhibition of the same kinases results in increased levels of O-GlcNAc. These data are in keeping with a reciprocal action of O-GlcNAc with respect to phosphorylation and suggest that this modification may have a role in signal transduction.