Axons rely on guidance cues to reach remote targets during nervous system development. A well-studied model system for axon guidance is the retinotectal projection. The retina can be divided into halves; the nasal half, next to the nose, and the temporal half. A subset of retinal axons, those from the temporal half, is guided by repulsive cues expressed in a graded fashion in the optic tectum, part of the midbrain. Here we report the cloning and functional characterization of a membrane-associated glycoprotein, which we call RGM (repulsive guidance molecule). This molecule shares no sequence homology with known guidance cues, and its messenger RNA is distributed in a gradient with increasing concentration from the anterior to posterior pole of the embryonic tectum. Recombinant RGM at low nanomolar concentration induces collapse of temporal but not of nasal growth cones and guides temporal retinal axons in vitro, demonstrating its repulsive and axon-specific guiding activity.

We We have previously described a chondroitin sulfate proteoglycan isolated from a phosphate-buffered saline extract of rat brain by immunoaffinity chromatography with the 3F8 monoclonal antibody and which is developmentally regulated with respect to its sulfation, carbohydrate composition and oligosaccharide structure, and immunocytochemical localization (1). A chondroitin/keratan sulfate proteoglycan (designated 3H1) was also isolated from rat brain by using monoclonal antibodies to the keratan sulfate chains (1). 3F8 and neurocan, another chondroitin sulfate proteoglycan of brain, for which the primary structure has been described (2), interact with neurons and the neural cell-adhesion molecules (CAMs), Ng-CAM and N-CAM (3). From radioligandbinding studies it was found that the brain proteoglycans bind with high affinity (Kd "O.5 nM) to Ng-CAM and N-CAM but not to other cell-surface and extracellular matrix proteins such as laminin, fibronectin, several collagens, epidermal growth factor and fibroblast growth factor receptors, or the myelin-associated glycoprotein (4). The 3F8 proteoglycan and neurocan inhibited neurite outgrowth and binding of neurons to Ng-CAM when mixtures of these proteins were adsorbed to polystyrene dishes, and direct binding ofneurons to the proteoglycan core glycoproteins was demonstrated with an assay in which cell-substrate contact was initiated by centrifugation (3,4). Recent studies have also shown that embryonic chicken brain neurons bind the 3H1 proteoglycan and contain cell-surface keratan sulfate chains (M. Flad, R.U.M., and R.K.M., unpublished results). These results indicate that brain proteoglycans can bind to neurons and that Ng-CAM and N-CAM may be heterophilic ligands for neurocan and the 3F8 and 3H1 proteoglycans.The primary structure of the 3F8 proteoglycan has now been determined by cDNA cloning, and we have identified this major chondroitin sulfate proteoglycan of brain as a possible mRNA splice variant of a receptor-type protein tyrosine phosphatase (PTP) named PTPC (5) and RPTFP. (6) (RPTPC/P). The tyrosine phosphatases act in concert with tyrosine kinases to regulate the phosphorylation state of proteins, and their structural features and potential physiological roles in signal transduction and cell-cycle regulation have recently been reviewed (7-9). With probes based on conserved sequences in their phosphatase domains, over 30 PTPs have now been cloned, but a major question concerning this family of key regulatory enzymes is the identification of their ligands and substrates. Our fidings indicate that neural CAMs and the exellular matrix protein tenascin may serve as ligands for RPIPC/P. Amino acid sequencing of the 3H1 chondroitin/keratan sulfate proteoglycan of brain demonstrated that it is a glycosylation variant of the 3F8 proteoglycan. We have named the 3F8/3H1 proteoglycan phosphacant and suggest that it may modulate cell interactions and other developmental processes in nervous tissue.

Abstract. Phosphacan is a chondroitin sulfate proteoglycan produced by glial cells in the central nervous system, and represents the extracellular domain of a receptor-type protein tyrosine phosphatase (RPTPg'/B). We previously demonstrated that soluble phosphacan inhibited the aggregation of microbeads coated with N-CAM or Ng-CAM, and have now found that soluble ~25I-phosphacan bound reversibly to these neural cell adhesion molecules, but not to a number of other cell surface and extracellular matrix proteins. The binding was saturable, and Scatchard plots indicated a single high affinity binding site with a Kd of ,~0.1 nM. Binding was reduced by "~15% after chondroitinase treatment, and free chondroitin sulfate was only moderately inhibitory, indicating that the phosphacan core glycoprotein accounts for most of the binding activity.Immunocytochemical studies of embryonic rat spinal cord and early postnatal cerebellum demonstrated that phosphacan, Ng-CAM, and N-CAM have overlapping distributions. When dissociated neurons were incubated on dishes coated with combinations of phosphacan and Ng-CAM, neuronal adhesion and neurite growth were inhibited. 125I-phosphacan bound to neurons, and the binding was inhibited by antibodies against Ng-CAM and N-CAM, suggesting that these CAMs are major receptors for phosphacan on neurons. C6 glioma cells, which express phosphacan, adhered to dishes coated with Ng-CAM, and low concentrations of phosphacan inhibited adhesion to Ng-CAM but not to laminin and fibronectin. Our studies suggest that by binding to neural cell adhesion molecules, and possibly also by competing for ligands of the transmembrane phosphatase, phosphacan may play a major role in modulating neuronal and glial adhesion, neurite growth, and signal transduction during the development of the central nervous system.

We have previously described the cloning of phosphacan, a chondroitin sulfate proteoglycan of nervous tissue which interacts with neurons, glia, neural cell adhesion molecules, and tenascin, and represents the extracellular domain of a receptor-type protein tyrosine phosphatase. We now report the complete cDNA and deduced amino acid sequences of the rat transmembrane phosphatase, and demonstrate that the phosphatase and the extracellular proteoglycan have different 3'-untranslated regions. Northern analysis showed three probable splice variants, comprising the extracellular proteoglycan (phosphacan) and long and short forms of the transmembrane phosphatase. PCR studies of rat genomic DNA indicated that there are no introns at the putative 5' and 3' splice sites or in the 2.6 kb segment which is deleted in the short transmembrane protein. Using variant-specific riboprobes corresponding to sequences in the 3'-untranslated region of phosphacan and in the first or second phosphatase domains of the transmembrane protein, in situ hybridization histochemistry of embryonic rat brain and spinal cord and early postnatal cerebellum demonstrated identical localizations of phosphacan and phosphatase mRNAs.