A proteoglycan with HNK-1 antigenic determinants is a neuron-associated ligand for cytotactin

Variability in the structural requirements for binding of human monoclonal anti-myelin-associated glycoprotein immunoglobulin M antibodies and HNK-1 to sphingoglycolipid antigens

A high proportion of patients with neuropathy have immunoglobulin M (IgM) paraproteins that react with carbohydrate determinants on the myelin-associated glycoprotein (MAG) and two sphingoglycolipids, 3-sulfoglucuronyl paragloboside (SGPG) and 3-sulfoglucuronyl lactosaminyl paragloboside. In order to characterize the fine specificities of these human antibodies further, the binding of 10 anti-MAG paraproteins to several chemically modified derivatives of SGPG was compared with the binding to intact SGPG by both TLC-overlay and enzyme-linked immunosorbent assay. The following derivatives were tested: the desulfated lipid, glucuronyl paragloboside (GPG); the methyl ester of GPG (MeGPG); the methyl ester of SGPG, 3-sulfomethylglucuronyl paragloboside (SMeGPG); and 3-sulfoglucosyl paragloboside (SGlcPG) produced by reduction of the carboxyl group of the glucuronic acid with sodium borohydride. All 10 IgM paraproteins and the related mouse IgM antibody, HNK-1, reacted most strongly with intact SGPG, but variations in the reactivity with the derivatives revealed striking differences in the structural requirements for binding between the antibodies.(ABSTRACT TRUNCATED AT 250 WORDS)

Localization during development of alternatively spliced forms of cytotactin mRNA by in situ hybridization

Cytotactin, an extracellular glycoprotein found in neural and nonneural tissues, influences a variety of cellular phenomena, particularly cell adhesion and cell migration. Northern and Western blot analysis and in situ hybridization were used to determine localization of alternatively spliced forms of cytotactin in neural and nonneural tissues using a probe (CT) that detected all forms of cytotactin mRNA, and one (VbVc) that detected two of the differentially spliced repeats homologous to the type III repeats of fibronectin. In the brain, the levels of mRNA and protein increased from E8 through E15 and then gradually decreased until they were barely detectable by P3. Among the three cytotactin mRNAs (7.2, 6.6, and 6.4 kb) detected in the brain, the VbVc probe hybridized only to the 7.2-kb message. In isolated cerebella, the 220-kD polypeptide and 7.2-kb mRNA were the only cytotactin species present at hatching, indicating that the 220-kD polypeptide is encoded by the 7.2-kb message that contains the VbVc alternatively spliced insert. In situ hybridization showed cytotactin mRNA in glia and glial precursors in the ventricular zone throughout the central nervous system. In all regions of the nervous system, cytotactin mRNAs were more transient and more localized than the polypeptides. For example, in the radial glia, cytotactin mRNA was observed in the soma whereas the protein was present externally along the glial fibers. In the telencephalon, cytotactin mRNAs were found in a narrow band at the edge of a larger region in which the protein was wide-spread. Hybridization with the VbVc probe generally overlapped that of the CT probe in the spinal cord and cerebellum, consistent with the results of Northern blot analysis. In contrast, in the outermost tectal layers, differential hybridization was observed with the two probes. In nonneural tissues, hybridization with the CT probe, but not the VbVc probe, was detected in chondroblasts, tendinous tissues, and certain mesenchymal cells in the lung. In contrast, hybridization with both probes was observed in smooth muscle and lung epithelium. Both epithelium and mesenchyme expressed cytotactin mRNA in varying combinations: in the choroid plexus, only epithelial cells expressed cytotactin mRNA; in kidney, only mesenchymal cells; and in the lung, both of these cell types contained cytotactin mRNA. These spatiotemporal changes during development suggest that the synthesis of the various alternatively spliced cytotactin mRNAs is responsive to tissue-specific local signals and prompt a search for functional differences in the various molecular forms of the protein.

Purification of hexabrachion (tenascin) from cell culture conditioned medium, and separation from a cell adhesion factor

We describe a protocol for purifying hexabrachion from conditioned medium of cell cultures, using gel filtration chromatography on Sephacryl 500, followed by anion-exchange chromatography on a Mono Q column, followed optionally by a second gel filtration or zone sedimentation on glycerol gradients. The protocol has several advantages over previous procedures based on affinity chromatography on monoclonal antibodies. Perhaps foremost, the protein is never exposed to the denaturing solvents that are required for elution from the antibody column. The Mono Q column also separated hexabrachion from a prominent cell adhesion activity that eluted with the hexabrachion on the first gel filtration, and co-sedimented with hexabrachions on glycerol gradients. The cell adhesion fractions showed several bands between 190 and 400 kDa. A single band at 220 kDa stained prominently with a polyclonal antibody against mouse EHS laminin, and a band at 190 kDa stained with a monoclonal antibody against s-laminin. The purification protocol gave hexabrachion at high concentration and with no detectable contamination by fibronectin or laminin. The highest yield of hexabrachion (1-4 mg from 400 ml of conditioned medium) was from human glioblastoma cell cultures, but the same procedure allowed us to purify and characterize the rat hexabrachion. Protein purified from primary cultures of rat embryo fibroblasts showed approximately equal amounts of three subunit sizes: 280, 230, and 220 kDa. These different subunits, presumably derived from alternative RNA splicing, appeared to be segregated into large and small hexabrachions, which could be separated on glycerol gradients.

Fibroblast-derived J1 adhesion glycoproteins show binding properties to extracellular matrix constituents different from those of central nervous system origin

The J1 extracellular adhesion molecule from mouse brain consists of several immunochemically related glycoproteins of different molecular weights and distinct functional properties. Like the brain J1 glycoproteins, the fibroblast-derived J1 glycoproteins interact with all collagen types tested (collagen G and types I-IV and IX), as measured by binding of 125I-labeled J1 glycoproteins to immobilized collagens. As tested for collagen type I, this binding can be inhibited more effectively by chondroitin sulfate than by heparin. After electrophoretic separation and transfer to nitrocellulose, fibroblast-derived J1 only binds to a limited number of collagen types (collagen types I, VI, and IX and G), whereas brain-derived J1 glycoproteins bind to all collagen types tested (collagen types I-VI and IX and G). These results show that fibroblast-derived J1 glycoproteins, although immunochemically related to J1 glycoproteins from brain, differ from these in their binding specificities to extracellular matrix constituents.

Binding of the J1 adhesion molecules to extracellular matrix constituents

The J1 glycoproteins can be obtained in multiple forms in the soluble fraction of developing and adult mouse brain tissue. They are recovered as two forms of apparent molecular weights of 160,000 and 180,000 (J1-160) from adult mouse brain and as forms of apparent molecular weights of 200,000 and 220,000 (J1-220) from developing brain. J1-160 and J1-220 share common epitopes but are considered as separate entities, with J1-220 being immunochemically closely related if not identical to tenascin. Based on the observation that J1 immunoreactivity appears on basement membrane and interstitial collagens after denervation of the neuromuscular junction in adult rodents, we became interested in investigating the binding properties of J1 glycoproteins to extracellular matrix constituents in vitro. Both J1-160 and J1-220 bound to collagens type I-VI and IX but not to laminin, fibronectin, bovine serum albumin, or gelatin under hypotonic buffer conditions. Under isotonic buffer conditions, J1-220 bound to all collagen types, whereas J1-160 bound only to collagen types V and VI with values that could be examined by Scatchard analysis. Binding of J1-220 to collagens displayed two binding constants (KD) between 1.5 and 4.4 X 10(-9) and 1.8 and 5.5 X 10(-8) M, respectively, under hypotonic buffer conditions and a single KD of 2.1-8.0 X 10(-8) M under isotonic buffer conditions. Binding of J1-160 to collagens had an apparent KD of 1.9-8.0 X 10(-9) M under hypotonic buffer conditions. Under isotonic buffer conditions, binding constants of J1-160 to collagen types V and VI were approximately 2 X 10(-8) M. Binding of J1-220 to collagen type I could be inhibited by J1-220, J1-160, and collagen type VI but not by fibronectin or gelatin. Conversely, binding of J1-160 was inhibited by J1-220, J1-160, and collagen type VI (in order of decreasing efficacy of competition). J1-160 and J1-220 were retained on a heparin-agarose column and eluted in a salt gradient at approximately 0.5 M NaCl. The formation of the J1-heparin complexes was inhibited 100-fold more efficiently by heparin than by chondroitin sulfate. These experiments show that J1 glycoproteins resemble in many respects the extracellular matrix constituents fibronectin, laminin, vitronectin, and von Willebrand factor.

Two Monoclonal Antibodies Recognizing Carbohydrate Epitopes on Neural Adhesion Molecules Interfere with Cell Interactions

We have studied two monoclonal antibodies raised against crude fractions of membrane glycoproteins from adult mouse brain and found them to react with two carbohydrate epitopes expressed on several neural cell adhesion molecules. Other identified and unidentified glycoproteins from different cell types, organs and species were also recognized by these antibodies. Both epitopes are N-glycosidically linked mannosidic or hybrid type oligosaccharides and co-expressed on all the glycoproteins so far tested. In spite of their remarkable similarities, the glycan epitopes are different as shown by ELISA competition assays. In microexplant outgrowth and cell adhesion assays, both antibodies interfere with neural cell adhesion, migration, and neurite outgrowth. These observations, together with previous studies on the L2/HNK-1 glycan (Künemund et al., 1988), indicate that adhesion molecules carry various carbohydrate epitopes mediating different cell interactions in in vitro assays.

Adhesion molecules and animal development

In recent years considerable progress has been made in the identification and characterization of molecules that mediate cell adhesion during animal development. This review attempts to pick out from the vast amount of information in this rapidly expanding field some of the key features of adhesion molecules, to present ideas about their role in development, and to indicate the directions in which the field is now moving.

Cell adhesion to fibronectin and tenascin: quantitative measurements of initial binding and subsequent strengthening response

Cell-substratum adhesion strengths have been quantified using fibroblasts and glioma cells binding to two extracellular matrix proteins, fibronectin and tenascin. A centrifugal force-based adhesion assay was used for the adhesive strength measurements, and the corresponding morphology of the adhesions was visualized by interference reflection microscopy. The initial adhesions as measured at 4 degrees C were on the order of 10(-5)dynes/cell and did not involve the cytoskeleton. Adhesion to fibronectin after 15 min at 37 degrees C were more than an order of magnitude stronger; the strengthening response required cytoskeletal involvement. By contrast to the marked strengthening of adhesion to FN, adhesion to TN was unchanged or weakened after 15 min at 37 degrees C. The absolute strength of adhesion achieved varied according to protein and cell type. When a mixed substratum of fibronectin and tenascin was tested, the presence of tenascin was found to reduce the level of the strengthening of cell adhesion normally observed at 37 degrees C on a substratum of fibronectin alone. Parallel analysis of corresponding interference reflection micrographs showed that differences in the area of cell surface within 10-15 nm of the substratum correlated closely with each of the changes in adhesion observed: after incubation for 15 min on fibronectin at 37 degrees C, glioma cells increased their surface area within close contact to the substrate by integral to 125-fold. Cells on tenascin did not increase their surface area of contact. The increased surface area of contact and the inhibitory activity of cytochalasin b suggest that the adhesive "strengthening" in the 15 min after initial binding brings additional adhesion molecules into the adhesive site and couples the actin cytoskeleton to the adhesion complex.

The mouse neuronal cell surface protein F3: a phosphatidylinositol-anchored member of the immunoglobulin superfamily related to chicken contactin

Several members of the Ig superfamily are expressed on neural cells where they participate in surface interactions between cell bodies and processes. Their Ig domains are more closely related to each other than to Ig variable and constant domains and have been grouped into the C2 set. Here, we report the cloning and characterization of another member of this group, the mouse neuronal cell surface antigen F3. The F3 cDNA sequence contains an open reading frame that could encode a 1,020-amino acid protein consisting of a signal sequence, six Ig-like domains of the C2 type, a long premembrane region containing two segments that exhibit sequence similarity to fibronectin type III repeats and a moderately hydrophobic COOH-terminal sequence. The protein does not contain a typical transmembrane segment but appears to be attached to the membrane by a phosphatidylinositol anchor. Antibodies against the F3 protein recognize a prominent 135-kD protein in mouse brain. In fetal brain cultures, they stain the neuronal cell surface and, in cultures maintained in chemically defined medium, most prominently neurites and neurite bundles. The mouse f3 gene maps to band F of chromosome 15. The gene transcripts detected in the brain by F3 cDNA probes are developmentally regulated, the highest amounts being expressed between 1 and 2 wk after birth. The F3 nucleotide and deduced amino acid sequence show striking similarity to the recently published sequence of the chicken neuronal cell surface protein contactin. However, there are important differences between the two molecules. In contrast to F3, contactin has a transmembrane and a cytoplasmic domain. Whereas contactin is insoluble in nonionic detergent and is tightly associated with the cytoskeleton, about equal amounts of F3 distribute between buffer-soluble, nonionic detergent-soluble, and detergent-insoluble fractions. Among other neural cell surface proteins, F3 most resembles the neuronal cell adhesion protein L1, with 25% amino acid identity between their extracellular domains. Based on its structural similarity with known cell adhesion proteins of nervous tissue and with L1 in particular, we propose that F3 mediates cell surface interactions during nervous system development.

Sulfated proteoglycans synthesized by Neuro 2a neuroblastoma cells: comparison between cells with and without ganglioside-induced neurites

Mouse neuroblastoma Neuro 2a cells are known to extend neurite-like processes in response to gangliosides added to the culture medium. We compared the structural features of proteoglycans (PG) synthesized by conventional Neuro 2a cells with those of neurite-bearing cells. Two different proteoglycans labeled with [35S]sulfate, namely, chondroitin sulfate proteoglycan (CS-PG) and heparan sulfate proteoglycan (HS-PG), were found both in the cell layer and in the culture medium of the conventional cells. CS-PG isolated from the cell layer had a Kav value of 0.38 on Sepharose CL-6B, and had CS side chains with Mr of 27,000. HS-PG in the cell layer was slightly larger (Kav of 0.33) in terms of hydrodynamic size than CS-PG, and the apparent Mr of the heparan sulfate side chains was 10,000. The structural parameters of CS-PG and HS-PG isolated from the medium were almost identical to those of the PGs in the cell layer. In addition to these PGs, single-chain HS, with an average Mr of 2,500, was observed only in the cell layer and this component was the major sulfated component in the cell layers of both control and ganglioside treated cells. The neurite-bearing cells also synthesized both CS-PG and HS-PG which were very similar in hydrodynamic size to those synthesized by the conventional cells, but the size of HS side chains was greater. Radioactivity, as 35S, of each sulfated component from the ganglioside-treated culture seemed to be slightly less than that of the corresponding component from the control culture. These findings indicate that the marked morphological change in Neuro 2a cells, induced by gangliosides is not accompanied by major changes in the synthesis of PGs.

Tenascin/hexabrachion in human skin: biochemical identification and localization by light and electron microscopy

Tenascin/hexabrachion is a large glycoprotein of the extracellular matrix. Previous reports have demonstrated that tenascin is associated with epithelial-mesenchymal interfaces during embryogenesis and is prominent in the matrix of many tumors. However, the distribution of tenascin is more restricted in adult tissues. We have found tenascin to be present in normal human skin in a distribution distinct from other matrix proteins. Immunohistochemical studies showed staining of the papillary dermis immediately beneath the basal lamina. Examination of skin that had been split within the lamina lucida of the basement membrane suggested a localization of tenascin beneath the lamina lucida. In addition, there was finely localized staining within the walls of blood vessels and in the smooth muscle bundles of the arrectori pilorem. Very prominent staining was seen around the cuboidal cells that formed the basal layer of sweat gland ducts. The sweat glands themselves did not stain. The distribution of tenascin in the papillary dermis was studied at high resolution by immunoelectron microscopy. Staining was concentrated in small amorphous patches scattered amongst the collagen fibers beneath the basal lamina. These patches were not associated with cell structures, collagen, or elastic fibers. Tenascin could be partially extracted from the papillary dermis by urea, guanidine hydrochloride, or high pH solution. The extracted protein showed a 320-kD subunit similar to that purified from fibroblast or glioma cell cultures. We have developed a sensitive ELISA assay that can quantitate tenascin at concentrations as low as 5 ng/ml. Tests on extracts of the papillary dermis showed tenascin constituted about 0.02-0.05% of the protein extracted.

Immunoelectron microscopic localization of hyaluronic acid-binding region and link protein epitopes in brain

The 1C6 monoclonal antibody to the hyaluronic acid-binding region weakly stained a 65-kD component in immunoblots of the chondroitin sulfate proteoglycans of brain, and the 8A4 monoclonal antibody, which recognizes two epitopes in the polypeptide portion of link protein, produced strong staining of a 45-kD component present in the brain proteoglycans. These antibodies were utilized to examine the localization of hyaluronic acid-binding region and link protein epitopes in rat cerebellum. Like the chondroitin sulfate proteoglycans themselves and hyaluronic acid, hyaluronic acid-binding region and link protein immunoreactivity changed from a predominantly extracellular to an intracellular (cytoplasmic and intra-axonal) location during the first postnatal month of brain development. The cell types which showed staining of hyaluronic acid-binding region and link protein, such as granule cells and their axons (the parallel fibers), astrocytes, and certain myelinated fibers, were generally the same as those previously found to contain chondroitin sulfate proteoglycans and hyaluronic acid. Prominent staining of some cell nuclei was also observed. In agreement with earlier conclusions concerning the localization of hyaluronic acid and chondroitin sulfate proteoglycans, there was no intracellular staining of Purkinje cells or nerve endings or staining of certain other structures, such as oligodendroglia and synaptic vesicles. The similar localizations and coordinate developmental changes of chondroitin sulfate proteoglycans, hyaluronic acid, hyaluronic acid-binding region, and link protein add further support to previous evidence for the unusual cytoplasmic localization of these proteoglycans in mature brain. Our results also suggest that much of the chondroitin sulfate proteoglycan of brain may exist in the form of aggregates with hyaluronic acid.

An alternatively spliced region of the human hexabrachion contains a repeat of potential N-glycosylation sites

We have cloned and sequenced two cDNA molecules that code for parts of two forms of human hexabrachion. The smaller clone has a sequence that corresponds to the previously published sequence of a cDNA clone coding for a part of chicken hexabrachion [Jones, F. S., Burgoon, M. P., Hoffman, S., Crossin, K. L., Cunningham, B. A. & Edelman, G. M. (1988) Proc. Natl. Acad. Sci. USA 85, 2186-2190]. It has eight consecutive domains similar to the type III homology units from fibronectin, several epidermal growth factor repeats, and a domain similar to the beta and gamma chains of fibrinogen. The larger clone has 5' and 3' ends that are identical to the smaller clone but also has an alternatively spliced 1.9-kilobase internal segment. The unique segment contains remarkable repeats of potential glycosylation sites and an additional seven type III homology units.

Expression of cytotactin in the normal and regenerating neuromuscular system

Cytotactin is an extracellular glycoprotein found in a highly specialized distribution during embryonic development. In the brain, it is synthesized by glia, not neurons. It is involved in neuron-glia adhesion in vitro and affects neuronal migration in the developing cerebellum. In an attempt to extend these observations to the peripheral nervous system, we have examined the distribution and localization of cytotactin in different parts of the normal and regenerating neuromuscular system. In the normal neuromuscular system, cytotactin accumulated at critical sites of cell-cell interactions, specifically at the neuromuscular junction and the myotendinous junction, as well at the node of Ranvier (Rieger, F., J. K. Daniloff, M. Pinçon-Raymond, K. L. Crossin, M. Grumet, and G. M. Edelman. 1986. J. Cell Biol. 103:379-391). At the neuromuscular junction, cytotactin was located in terminal nonmyelinating Schwann cells. Cytotactin was also detected near the insertion points of the muscle fibers to tendinous structures in both the proximal and distal endomysial regions of the myotendinous junctions. This was in striking contrast to staining for the neural cell adhesion molecule, N-CAM, which was accumulated near the extreme ends of the muscle fiber. Peripheral nerve damage resulted in modulation of expression of cytotactin in both nerve and muscle, particularly among the interacting tissues during regeneration and reinnervation. In denervated muscle, cytotactin accumulated in interstitial spaces and near the previous synaptic sites. Cytotactin levels were elevated and remained high along the endoneurial tubes and in the perineurium as long as muscle remained denervated. Reinnervation led to a return to normal levels of cytotactin both in inner surfaces of the nerve fascicles and in the perineurium. In dorsal root ganglia, the processes surrounding ganglionic neurons became intensely stained by anticytotactin antibodies after the nerve was cut, and returned to normal by 30 d after injury. These data suggest that local signals between neurons, glia, and supporting cells may regulate cytotactin expression in the neuromuscular system in a fashion coordinate with other cell adhesion molecules. Moreover, innervation may regulate the relative amount and distribution of cytotactin both in muscle and in Schwann cells.

P-epitope is characteristic for the neural tissue of vertebrates

In a search for antigens immunologically related to chordin, a notochord-specific glycoprotein of sturgeneous fishes, extracts from 55 samples of human and rabbit tissues were tested for inhibition of [125I]chordin binding to rabbit polyclonal antibodies. The strongest inhibition was observed with brain extracts of both species. Human, chicken, rabbit, and newt brain extracts also inhibited chordin binding in liquid phase to monoclonal antibodies (MAbs) against the P-epitope, the most immunogenic epitope of this glycoprotein. Immunohistochemical studies done on human and chicken embryos, newt, sterlet, and sturgeon embryos, larvae, and juveniles revealed a strong immunoreactivity of the brain, spinal cord, and tissue of the peripheral nervous system with an anti-P MAb. Other tissues, with several exceptions, showed a negative reaction in immunohistochemical experiments. The authors found that the P-epitope is ontogenetically expressed in the neural tissue of chicken, newt, and sterlet at the period of cytodifferentiation. Gel chromatography of human, chicken, and newt brain extracts showed that in each case the P-epitope was associated with a polydisperse macromolecular material of similar size. These antigens were designated as neurochordins. Prolonged pronase digestion of human and chicken brain extracts resulted in fragments with M about 3 kDa (presumably glycopeptides), which reacted with anti-P MAbs. These fragments were of the same size as corresponding glycopeptides of the pronase digest of chordin. Thus, in the present study, the P-epitope has been shown to be characteristic for the neural tissue of several vertebrate species; in the brain, it has been found in association with neurochordins, macromolecular antigens that are presumably protein conjugates with carbohydrates.

Functional mapping of cytotactin: proteolytic fragments active in cell-substrate adhesion

Cytotactin is an extracellular matrix glycoprotein with a restricted distribution during development. In electron microscopic images, it appears as a hexabrachion with six arms extending from a central core. Cytotactin binds to other extracellular matrix proteins including a chondroitin sulfate proteoglycan (CTB proteoglycan) and fibronectin. Although cytotactin binds to a variety of cells including fibroblasts and neurons, in some cases it causes cells in culture to round up and it inhibits their migration. To relate these various effects of cytotactin on cell behavior to its binding regions, we have examined its ability to support cell-substrate adhesion and have mapped its cell-binding function onto its structure. In a cell-substrate adhesion assay, fibroblasts bound to cytotactin but remained round. In contrast, they both attached and spread on fibronectin. Neither neurons nor glia bound to cytotactin in this assay. In an assay in which cell-substrate contact was initiated by centrifugation, however, neurons and glia bound well to cytotactin; this binding was blocked by specific anti-cytotactin antibodies. The results suggest that neurons and glia can bind to cytotactin-coated substrates and that these cells, like fibroblasts, possess cell surface ligands for cytotactin. After applying methods of limited proteolysis and fractionation, these assays were used to map the binding functions of cytotactin onto its structure. Fragments produced by limited proteolysis were fractionated into two major pools: one (fraction I) contained disulfide-linked oligomers of a 100-kD fragment and two minor related fragments, and the second (fraction II) contained monomeric 90- and 65-kD fragments. The 90- and 65-kD fragments in fraction II were closely related to each other and were structurally and immunologically distinct from the fragments in fraction I. Only components in fraction I were recognized by mAb M1, which binds to an epitope located in the proximal portion of the arms of the hexabrachion and by a polyclonal antibody prepared against a 75-kD CNBr fragment of intact cytotactin. A mAb (1D8) and a polyclonal antibody prepared against a 35-kD CNBr fragment of cytotactin only recognized components present in fraction II. In cell-binding experiments, fibroblasts, neurons, and glia each adhered to substrates coated with fraction II, but did not adhere to substrates coated with fraction I. Fab fragments of the antibody to the 35-kD CNBr fragment strongly inhibited the binding of cells to cytotactin, supporting the conclusion that fraction II contains a cell-binding region. In addition, Fab fragments of this antibody inhibited the binding of cytotactin to CTB pr

Expression and localization of the fibronectin receptor in the mouse nervous system

Cell surface receptors for extracellular matrix components have recently been characterized as integral membrane complexes with common features in their structural and functional properties. We have investigated the expression of the mammalian fibronectin receptor in the mouse nervous system using immunocytological and immunochemical methods. The fibronectin receptor was detectable on immature oligodendrocytes and immature and mature astrocytes in culture, while central nervous system neurons did not reveal detectable levels of fibronectin receptor at the developmental stages studied. In the peripheral nervous system both glia and neurons were found to express the fibronectin receptor. The receptor complex in both peripheral and central nervous system has an apparent molecular weight of approximately 140 kD under reducing conditions and resolves into two or three distinct protein bands under nonreducing conditions. The fibronectin receptor expresses the L2/HNK-1 epitope that is characteristic of several adhesion molecules, including L1, N-CAM, the myelin-associated glycoprotein, and J1 and thus is another member of the L2/HNK-1 family of adhesion molecules. The L2/HNK-1 carbohydrate epitope is expressed differently and independently of the fibronectin receptor protein backbone in that it is detectable in neonatal brain but not in adult brain. Our observations attribute a functional role to the fibronectin receptor and its L2/HNK-1 carbohydrate epitope during development and maintenance of cell interactions in the central and peripheral nervous systems.

A cDNA clone for cytotactin contains sequences similar to epidermal growth factor-like repeats and segments of fibronectin and fibrinogen

Cytotactin is an extracellular glycoprotein that influences neuron-glia interactions. It has been shown to appear in multiple forms that are differentially expressed in neural and non-neural tissues during vertebrate development. We report here the isolation and characterization of a cytotactin cDNA clone (lambda C801) that encodes 933 amino acids, equivalent to about half of a cytotactin polypeptide. Clone lambda C801 is an authentic cytotactin cDNA: it encodes a polypeptide that reacts with a monoclonal anti-cytotactin antibody and its deduced amino acid sequence is identical for 15 amino acids to the directly determined sequence of a CNBr fragment that reacted with the same antibody. Southern blot analyses with fragments of lambda C801 suggested that there may be only one cytotactin gene, but RNA transfer blots detected multiple mRNAs ranging in size from 6.5 to 8.0 kilobases. An 8.0-kilobase message and a Mr 240,000 cytotactin polypeptide were present in embryonic gizzard but not brain, while a 7.2-kilobase message and a Mr 220,000 polypeptide were present in brain but not gizzard. These results indicate that differential splicing of primary transcripts of the cytotactin gene yields various site-specific polypeptides. Sequence analyses of lambda C801 indicated that it specifies a region with extensive similarities to other proteins: the sequence begins with four consecutive epidermal growth factor-like repeats that are followed by eight segments that closely resemble each other and the type III repeats in fibronectin, and it ends with a 66 amino acid sequence similar to part of the beta and gamma chains of fibrinogen. One fibronectin-like repeat contains a single Arg-Gly-Asp sequence. The similarities with all three of these apparently unrelated proteins are extensive, suggesting that cytotactin has an evolutionary and possibly a functional relationship to each.

The high-molecular-weight J1 glycoproteins are immunochemically related to tenascin

The J1 glycoproteins have been shown to mediate neuron-astrocyte adhesion and appear in the nervous system as four species of Mr 160,000 (J1-160), 180,000 (J1-180), 200,000 (J1-200), and 220,000 (J1-220), respectively. Tenascin is a disulfide-linked oligomeric, extracellular matrix glycoprotein of subunit Mr 170,000, 190,000, 200,000, and 220,000, which has been proposed to promote epithelial cell proliferation. In view of the structural similarities of the molecules we have used immunohistochemical and immunochemical techniques to compare them. Immunohistochemically, polyclonal J1 and tenascin antibodies yielded identical staining patterns in non-nervous-system tissues, and staining could be completely blocked by preincubating the sera with purified tenascin. In the central nervous system all structures expressing tenascin immunoreactivity were also recognized by J1 antibodies. However, not all J1-positive structures were also tenascin-positive, indicating that J1 antibodies recognized additional epitopes not present on tenascin. Western-blot experiments performed with affinity-purified polyclonal J1 antibodies showed that J1 glycoproteins can be subdivided into two separate pairs, J1-160/180 and J1-200/220, which share a small degree of homology. Western-blot experiments and sequential immunoprecipitations on biosynthetically [35S]methionine- or 125I-radiolabeled J1 glycoproteins carried out with polyclonal J1 and tenascin antibodies demonstrated that J1-200/220 is immunochemically indistinguishable from tenascin. These observations suggest that one set of extracellular glycoproteins is associated with processes as different as neural histogenesis and carcinogenesis of mammary glands.

Molecular forms, binding functions, and developmental expression patterns of cytotactin and cytotactin-binding proteoglycan, an interactive pair of extracellular matrix molecules

Cytotactin is an extracellular matrix protein that is found in a restricted distribution and is related to developmental patterning at a number of neural and non-neural sites. It has been shown to bind specifically to other extracellular matrix components including a chondroitin sulfate proteoglycan (cytotactin-binding [CTB] proteoglycan) and fibronectin. Cell binding experiments have revealed that cytotactin interacts with neurons and fibroblasts. When isolated from brain, both cytotactin and CTB proteoglycan contain the HNK-1 carbohydrate epitope. Here, specific antibodies prepared against highly purified cytotactin and CTB proteoglycan were used to correlate the biochemical alterations and modes of binding of these proteins with their differential tissue expression as a function of time and place during chicken embryo development. It was found that, during neural development, both the levels of expression of cytotactin and CTB proteoglycan and of the molecular forms of each molecule varied, following different time courses. In addition, a novel Mr 250,000 form of cytotactin was detected that contained chondroitin sulfate. The intermolecular binding of cytotactin and CTB proteoglycan and the binding of cytotactin to fibroblasts were characterized further and found to be inhibited by EDTA, consistent with a dependence on divalent cations. Unlike the molecules from neural tissue, cytotactin and CTB proteoglycan isolated from non-neural tissues such as fibroblasts lacked the HNK-1 epitope. Nevertheless, the intermolecular and cellular binding activities of cytotactin isolated from fibroblast culture medium were comparable to those of the molecule isolated from brain, suggesting that the HNK-1 epitope is not directly involved in binding. Binding experiments involving enzymatically altered molecules that lack chondroitin sulfate suggested that this glycosaminoglycan is also not directly involved in binding. Although they clearly formed a binding couple, the spatial distributions of cytotactin and CTB proteoglycan in the embryo were not always coincident. They were similar in tissue sections from the cerebellum, gizzard, and vascular smooth muscle. In contrast, CTB proteoglycan was present in cardiac muscle where no cytotactin is present, and it was seen in cartilage throughout development unlike cytotactin, which was present only in immature chondrocytes. Cell culture experiments were consistent with the previous conclusion that cytotactin was specifically synthesized by glia, whereas CTB proteoglycan was specifically synthesized by neurons.(ABSTRACT TRUNCATED AT 400 WORDS)

Neuron-glia cell adhesion molecule interacts with neurons and astroglia via different binding mechanisms

The neuron-glia cell adhesion molecule (Ng-CAM) is present in the central nervous system on postmitotic neurons and in the periphery on neurons and Schwann cells. It has been implicated in binding between neurons and between neurons and glia. To understand the molecular mechanisms of Ng-CAM binding, we analyzed the aggregation of chick Ng-CAM either immobilized on 0.5-micron beads (Covaspheres) or reconstituted into liposomes. The results were correlated with the binding of these particles to different types of cells as well as with cell-cell binding itself. Both Ng-CAM-Covaspheres and Ng-CAM liposomes individually self-aggregated, and antibodies against Ng-CAM strongly inhibited their aggregation; the rate of aggregation increased approximately with the square of the concentration of the beads or the liposomes. Much higher rates of aggregation were observed when the ratio of Ng-CAM to lipid in the liposome was increased. Radioiodinated Ng-CAM on Covaspheres and in liposomes bound both to neurons and to glial cells and in each case antibodies against Ng-CAM inhibited 50-90% of the binding. Control preparations of fibroblasts and meningeal cells did not exhibit significant binding. Adhesion between neurons and glia within and across species (chick and mouse) was explored in cellular assays after defining markers for each cell type, and optimal conditions of shear, temperature, and cell density. As previously noted using chick cells (Grumet, M., S. Hoffman, C.-M. Chuong, and G. M. Edelman. 1984 Proc. Natl. Acad. Sci. USA. 81:7989-7993), anti-Ng-CAM antibodies inhibited neuron-neuron and neuron-glia binding. In cross-species adhesion assays, binding of chick neurons to mouse astroglia and binding of mouse neurons to chick astroglia were both inhibited by anti-Ng-CAM antibodies. To identify whether the cellular ligands for Ng-CAM differed for neuron-neuron and neuron-glia binding, cells were preincubated with specific antibodies, the antibodies were removed by washing, and Ng-CAM-Covasphere binding was measured. Preincubation of neurons with anti-Ng-CAM antibodies inhibited Ng-CAM-Covasphere binding but similar preincubation of astroglial cells did not inhibit binding. In contrast, preincubation of astroglia with anti-astroglial cell antibodies inhibited binding to these cells but preincubation of neurons with these antibodies had no effect. Together with the data on Covaspheres and liposome aggregation, these findings suggested that Ng-CAM-Covaspheres bound to Ng-CAM on neurons but bound to different molecules on astroglia.(ABSTRACT TRUNCATED AT 400 WORDS)

Expression of markers shared between human natural killer cells and neuroblastoma lines

Neuroblastoma is a tumor of neuroectodermal origin arising most commonly from the adrenal medulla. We have examined the ability of several monoclonal antibodies which recognize markers predominantly expressed on human natural killer (NK) cells to react with neuroblastoma cell lines in vivo derived sections of tumor. HNK-1 (Leu 7) is a monoclonal IgM antibody which recognizes a carbohydrate epitope on NK cells and a wide range of tumor cell types. We have shown that HNK-1 recognizes the human neuroblastoma lines SMS-KCNR, SMS-KAN, NMB/N7, and IMR/5. Expression of this antigen on cell lines can be slightly increased by retinoic acid-induced differentiation of the cells. N901 (NKH1), a monoclonal antibody raised against interleukin 2-dependent human NK cell lines also recognizes all human neuroblastoma cell lines examined. This expression is independent of differentiation induction and levels remain unaltered following retinoic acid treatment of the cell lines. Lastly, with monoclonal antibody 49H.8, it has been found that reactivity of the lines is weak until induction of differentiation, after which highly significant increases of reactivity are seen. 49H.8 recognizes several cryptic carbohydrate antigens with varying affinities, shown to identify mouse and rat NK cells. In contrast to other NK markers, human neuroblastoma cell lines did not express significant reactivity with B73.1, Leu 11b, or Leu 18. Immunohistochemical staining of sections of human neuroblastoma tumors correlated with the in vitro findings; however, staining with N901 and 49H.8 was only seen on frozen sections, not paraffin-embedded. The significance of shared NK cell-neuroblastoma/neuron antigens is currently under investigation.

The perinodal astrocyte

Several studies have demonstrated the presence of perinodal astrocyte processes at nodes of Ranvier in the central nervous system, suggesting that, in addition to the axon and oligodendrocyte, astrocytes participate in the formation of mature central nodes. The specific association between perinodal astrocyte processes and nodal membrane develops at the time of, or soon after, the appearance of relatively differentiated nodes of Ranvier. This interaction is likely to be mediated by cell adhesion molecules. J1 is a member of a family of glycoproteins that share a common carbohydrate epitope, designated L2/HNK-1, and that have been implicated in cell-cell interactions. This glycoprotein is concentrated at the interface between perinodal astrocyte processes and the nodal region of the axon. Moreover, N-CAM, which is a member of the same family as J1, and cytotactin, an extracellular matrix component produced by glia, are localized at the interface between the axon and perinodal astrocyte processes at nodes of Ranvier. The association of perinodal astrocyte processes with nodal membrane in the central nervous system is similar to that exhibited by perinodal Schwann cell processes at peripheral nodes, and similar functional properties have been suggested for these two glial cell processes, including production of nodal gap substance, buffering of perinodal extracellular ion concentration, and development and/or maintenance of nodal specializations in the axon membrane. Perinodal astrocyte and Schwann cell processes may also function as extraneuronal sites for the synthesis of voltage-sensitive sodium channels, to complement neuronal perikaryal synthesis and axonal transport. Ultrastructural studies on specialized patches of axon membrane within some unmyelinated, demyelinated, and dysmyelinated axons support the hypothesis of a specific role for perinodal astrocyte processes in the assembly, stabilization, and/or maintenance of axolemma with nodal characteristics. These observations suggest a multiplicity of functions for perinodal astrocyte processes at central nodes and implicate the astrocyte as an important component of the node of Ranvier.

CAMs and Igs: cell adhesion and the evolutionary origins of immunity

The lymphoid system and cells of immunity are as morphologically well defined as those of any complex organ but in addition they show dynamic long-range interactions between the fluid tissues (lymphocytes, monocytes, etc.) and the solid, vascular and generative tissues and organs which they comprise. Given the observation that CAMs are present in epithelial components of lymphoid organs, it appears that, in their ontogeny, the organs of immunity will share a common principle of morphoregulation by CAMs with brains, feathers and other parts of the phenotype. As discussed here, this principle is a regulatory one operating across many levels of organization from the genes to tissues and back again (see Fig. 1). At some early point in the evolution of the immune system, a gene corresponding to an N-CAM precursor must have duplicated to provide a basis for the Ig superfamily with its increasing specializations for recognition and for cellular regulation during the immune response. Lymphocyte cellular families also developed later specializations (along with other leukocytes) for adhesive functions accessory to specific recognition. As far as we can see, the molecules for these accessory functions only remotely resemble CAMs, but closely resemble receptors for matrix molecules and SAMs. What CAMs and Ig superfamily members have in common is an evolutionary path and important roles in mediating complex regulatory responses that arise from cell-cell interactions. In the one case, this regulation leads to morphology, and in the other, to immune recognition. The first depends directly upon pattern (the formation of definite tissue structure); the regulation of the second also depends upon pattern to the extent that its function is dependent upon the morphology of lymphoid organs and vasculature. But although specific immune recognition depends locally upon adhesion through special mechanisms, it does not lead to morphology. One must not therefore impute too much in the physiological sense to the resemblance among brain molecules and molecules of the immune system. CAMs themselves are not directly histotypic at the level of individual differentiated cells but rather are used to link early tissue boundaries in induction and function in a wide variety of different tissues. As a consequence, N-CAM is central to the formation and maintenance of neural tissue but has a much wider tissue distribution and a fundamental role in very early embryogenesis as is the case with other primary CAMs. Thus, the immune system did not evolve from the nervous system, but from a cell adhesion system essential to both.(ABSTRACT TRUNCATED AT 400 WORDS)

Asymmetric expression in somites of cytotactin and its proteoglycan ligand is correlated with neural crest cell distribution

The development of the vertebrate neural crest presents a particularly challenging problem in pattern formation. Several studies have revealed that a population of neural crest cells penetrates the sclerotomal mesenchyme of the somite only in its rostral half. In a search for molecular correlates of this pattern, we have observed that cytotactin and a chondroitin sulfate proteoglycan, two interactive extracellular matrix molecules, show a specialized distribution within the sclerotome. Cytotactin was localized in the rostral half of the sclerotome at about the time of neural crest cell invasion. The proteoglycan was initially diffuse throughout the sclerotome but became restricted to the caudal half after the appearance of cytotactin and invasion of neural crest cells in the rostral half. These distributions were crest cell-independent; they occurred on the same schedule even when all crest cells were removed by surgical extirpation of the neural tube. Furthermore, in tissue culture, somite cells synthesized high levels of both molecules. In vitro, crest cells rounded up in the presence of these molecules and cell migration assays revealed that neither cytotactin nor proteoglycan alone was as good a substratum for crest cell migration as fibronectin. In combination with fibronectin, however, cytotactin or proteoglycan only restricted cell movement but did not prevent it. Taken together, these observations support the hypothesis that cytotactin and the chondroitin sulfate proteoglycan may contribute to pattern formation during embryogenesis by means of their site-restricted distribution, their ability to alter migration on other substrates such as fibronectin, and their ability to induce cell-surface modulation.