Calcium influx into neurons can solely account for cell contact-dependent neurite outgrowth stimulated by transfected L1.

We have used monolayers of control 3T3 cells and 3T3 cells expressing transfected human L1 as a culture substrate for rat PC12 cells and rat cerebellar neurons. PC12 cells and cerebellar neurons extended longer neurites on human L1 expressing cells. Neurons isolated from the cerebellum at postnatal day 9 responded equally as well as those isolated at postnatal day 1-4, and this contrasts with the failure of these older neurons to respond to the transfected human neural cell adhesion molecule (NCAM). Human L1-dependent neurite outgrowth could be blocked by antibodies that bound to rat L1 and, additionally, the response could be fully inhibited by pertussis toxin and substantially inhibited by antagonists of L- and N-type calcium channels. Calcium influx into neurons induced by K+ depolarization fully mimics the L1 response. Furthermore, we show that L1- and K+(-)dependent neurite outgrowth can be specifically inhibited by a reduction in extracellular calcium to 0.25 microM, and by pretreatment of cerebellar neurons with the intracellular calcium chelator BAPTA/AM. In contrast, the response was not inhibited by heparin or by removal of polysialic acid from neuronal NCAM both of which substantially inhibit NCAM-dependent neurite outgrowth. These data demonstrate that whereas NCAM and L1 promote neurite outgrowth via activation of a common CAM-specific second messenger pathway in neurons, neuronal responsiveness to NCAM and L1 is not coordinately regulated via posttranslational processing of NCAM. The fact that NCAM- and L1-dependent neurite outgrowth, but not adhesion, are calcium dependent provides further evidence that adhesion per se does not directly contribute to neurite outgrowth.

Induction of neurite outgrowth through contactin and Nr-CAM by extracellular regions of glial receptor tyrosine phosphatase beta.

Receptor protein tyrosine phosphatase beta (RPTPbeta) is expressed as soluble and receptor forms with common extracellular regions consisting of a carbonic anhydrase domain (C), a fibronectin type III repeat (F), and a unique region called S. We showed previously that a recombinant Fc fusion protein with the C domain (beta C) binds to contactin and supports neuronal adhesion and neurite growth. As a substrate, betaCFS was less effective in supporting cell adhesion, but it was a more effective promoter of neurite outgrowth than betaCF. betaS had no effect by itself, but it potentiated neurite growth when mixed with betaCF. Neurite outgrowth induced by betaCFS was inhibited by antibodies against Nr-CAM and contactin, and these cell adhesion molecules formed a complex that bound betaCFS. NIH-3T3 cells transfected to express betaCFS on their surfaces induced neuronal differentiation in culture. These results suggest that binding of glial RPTPbeta to the contactin/Nr-CAM complex is important for neurite growth and neuronal differentiation.

Alternatively spliced mRNAs code for different polypeptide chains of the chicken neural cell adhesion molecule (N-CAM).

Rabbit polyclonal antibodies directed against the chicken neural cell adhesion molecule (N-CAM) were used to isolate four overlapping cDNA clones from a chicken cDNA expression library in bacteriophage gamma gt11. These clones collectively accounted for 3.8 kilobases of N-CAM mRNA sequence and hybridized specifically to two 6-7-kilobase brain polyadenylated RNA species that co-migrated with previously identified N-CAM mRNAs. DNA fragments derived from an internal region of the cloned cDNA sequences hybridized to the larger but not to the smaller N-CAM mRNA species, while fragments on either side of this region hybridized to both mRNAs. A cDNA fragment that recognized only the larger mRNA was subcloned into gamma gt11, and the expressed fusion protein was used to affinity-purify rabbit polyclonal antibodies; the antibodies recognized only the larger of the two structurally related N-CAM polypeptides. In contrast, when several cDNA clones that recognized both mRNAs were used to purify antibodies, the antibodies recognized both polypeptides. The results, in conjunction with other data indicating that there is one gene specifying N-CAM, suggest that different N-CAM polypeptides are synthesized from multiple N-CAM messages generated by alternative splicing of transcripts from a single N-CAM gene.

Neural cell adhesion molecule: structure, immunoglobulin-like domains, cell surface modulation, and alternative RNA splicing.

The neural cell adhesion molecule, N-CAM, appears on early embryonic cells and is important in the formation of cell collectives and their boundaries at sites of morphogenesis. Later in development it is found on various differentiated tissues and is a major CAM mediating adhesion among neurons and between neurons and muscle. To provide a molecular basis for understanding N-CAM function, the complete amino acid sequences of the three major polypeptides of N-CAM and most of the noncoding sequences of their messenger RNA's were determined from the analysis of complementary DNA clones and were verified by amino acid sequences of selected CNBr fragments and proteolytic fragments. The extracellular region of each N-CAM polypeptide includes five contiguous segments that are homologous in sequence to each other and to members of the immunoglobulin superfamily, suggesting that interactions among immunoglobulin-like domains form the basis for N-CAM homophilic binding. Although different in their membrane-associated and cytoplasmic domains, the amino acid sequences of the three polypeptides appear to be identical throughout this extracellular region (682 amino acids) where the binding site is located. Variations in N-CAM activity thus do not occur by changes in the amino acid sequence that alter the specificity of binding. Instead, regulation is achieved by cell surface modulation events that alter N-CAM affinity, prevalence, mobility, and distribution on the surface. A major mechanism for modulation is alternative RNA splicing resulting in N-CAM's with different cytoplasmic domains that differentially interact with the cell membrane. Such regulatory mechanisms may link N-CAM binding function with other primary cellular processes during the embryonic development of pattern.

Further studies of elevated cerebrospinal fluid neuronal cell adhesion molecule in schizophrenia.

BACKGROUND: The purposes of the present study were to attempt to replicate a previous finding of increased cerebrospinal fluid (CSF) neuronal cell adhesion molecule (N-CAM) in schizophrenia, and to assess whether the increases could be related to medication, clinical state effects, or brain structural measures. METHODS: CSF N-CAM was measured by the Western blot technique in 45 DSM-III-R diagnosed male schizophrenic patients both on and off haloperidol treatment and in 20 healthy male control subjects. RESULTS: CSF N-CAM was significantly increased in schizophrenic patients, with no overlap in the ranges, when compared to controls. There were no significant effects of medication or exacerbation on CSF N-CAM. No associations with measures of brain structure were found. CONCLUSIONS: Because N-CAM levels were not shown to be different on and off treatment or in exacerbated versus nonexacerbated patients, the higher levels seen in schizophrenic patients may be inherent to the disorder and possibly related to neurodevelopment.

Novel fluorescent technology platform for high throughput cytotoxicity and proliferation assays.

We have developed a novel fluorescent Oxygen BioSensor technology platform adaptable to many applications in the area of drug discovery and development, particularly cell-based assays. This biosensor technology requires no additional reagents or incubations, and affords continuous real-time readout of dissolved oxygen concentrations. Since the level of oxygen dissolved in an assay's medium correlates to the number and viability of the cells in the medium, this technology is ideally suited for monitoring cell viability, proliferation, or death. The technology is particularly well suited to investigating cells' kinetic responses to proliferative or toxic stimuli, such as drugs. When incorporated into a 96- or 384-well microplate format, it is compatible with standard laboratory automation systems. Here we present data illustrating the application of the Oxygen BioSensor technology for rapid, homogeneous detection and evaluation of metabolic activity of a variety of eukaryotic and prokaryotic cells, including mammalian cells, insect cells, yeast, and bacteria. In the absence of toxic substances, we find a good correlation between cell number and signal over a wide range of cell concentrations and growth times. To evaluate the usefulness of the Oxygen BioSensor for cytotoxicity assays, we have performed a series of experiments using a range of toxic agents and cell types, including both bacteria and mammalian cell lines. In a side-by-side comparison to standard MTT assays using HL60 cells, comparable IC(50) values were found with the Oxygen BioSensor for five different toxins or drugs. This assay method does not have the need for additional reagents, handling steps, or incubation periods required by the MTT assays.

Variants of human L1 cell adhesion molecule arise through alternate splicing of RNA.

The L1 cell adhesion molecule was initially identified and characterized in mouse as a cell-surface glycoprotein that mediates neuron-neuron and neuron-Schwann cell adhesion. We have characterized L1 in humans using cDNA structural and mRNA expression analyses. We present the entire coding sequence for human L1, which predicts a 1253-amino acid protein displaying a signal sequence, transmembrane segment, RGD sequence, and potential glycosylation and phosphorylation sites. Nucleotide and deduced amino acid sequence identities between human and mouse L1 are 85% and 87%, respectively. In contrast, the amino acid identity between human L1 and the L1-related molecule chicken Ng-CAM is only 45%. Using Northern blot analyses, a single L1 transcript of 5.5 kb is detected in human fetal brain and in neuroblastoma (IMR-32) and retinoblastoma (Y-79) cell lines. L1 is also expressed in the rhabdomyosarcoma cell lines RD and A-204, which display several muscle characteristics. Two forms of L1, which differ by the presence or absence of a 12-bp cytoplasmic segment, are expressed in both human and mouse. This segment is encoded by a single exon that can be alternately spliced to give rise to the two forms, which appear to be expressed in tissue-specific patterns.

Characterization of cDNA clones defining variant forms of human neural cell adhesion molecule N-CAM.

The neural cell adhesion molecule N-CAM has been identified in a number of species and comprises at least three major cell surface polypeptides of different molecular structures and tissue distributions. We report here the isolation and characterization of cDNA clones encoding two of the three major forms of N-CAM from a human neuroblastoma cDNA library. One of the clones, NII-6, provides the first complete sequence of a small cytoplasmic domain (140 kDa) form of the molecule in humans and differs in a number of respects from cDNA clones derived from human muscle. These differences include the presence of a 30-bp insert in the fourth immunoglobulin-like domain of N-CAM, a 3-bp insert in the extracellular portion of the molecule, and an additional 6 bp in the middle of the membrane-spanning segment. Based on the analysis of a genomic DNA clone spanning these regions of N-CAM, the first two differences arise by alternate splicing of RNA and occur in some, but not all clones; the additional 6 bp may reflect a genetic polymorphism. A second cDNA clone, NI-10, encodes the complete sequence of a segment that is specific to the large cytoplasmic domain (180 kDa) polypeptide of human N-CAM and is very similar to corresponding segments of mouse, chicken, and rat N-CAM. This sequence also arises by alternative splicing of RNA. In addition, we have identified a genomic DNA segment encoding sequences specific to the third, small surface domain (120 kDa) polypeptide of N-CAM. The data presented here and previously define the DNA sequences of the membrane-bound forms and known variants of human N-CAM. From these sequences, a wide variety of probes can be generated for investigating the expression of particular N-CAM polypeptides in normal and pathological tissues.

Identification and characterization of the human cell adhesion molecule contactin.

We have prepared a monoclonal antibody, Neuro-1, that recognizes the human homolog of the chicken contactin/F11 and mouse F3 cell adhesion molecules. The Neuro-1 antigen, structurally characterized as a 135 kDa glycosylphosphatidylinositol-linked glycoprotein, was immunoaffinity purified and partially sequenced. Comparison of an internal peptide sequence to that predicted from the chicken contactin/F11, mouse F3 and human contactin (reported herein) cDNA sequence identifies the Neuro-1 antigen as human contactin. Moreover, a polyclonal antisera generated against the purified Neuro-1 antigen was immunoreactive with a fragment of human contactin expressed in bacteria. The complete coding and deduced amino acid sequences of human contactin were determined and are 86% and 95% identical to the respective mouse F3 sequences. Structural features shared with contactin/F11/F3 include six immunoglobulin type C2 and four fibronectin type III-like domains, multiple sites for asn-linked glycosylation and a COOH-terminal signal peptide presumably removed during the generation of a phosphatidylinositol cell surface linkage. The potential for glycosylation and GPI-linkage is also consistent with protein chemical studies of human contactin. Contactin mRNA expression was characterized using Northern blot analyses of human tissues and cell lines. High level expression of a single contactin transcript in adult brain, and low level expression of multiple transcripts in lung, pancreas, kidney and skeletal muscle are observed. Highly expressed multiple transcripts, similar in pattern to that of pancreas, lung, kidney and skeletal muscle, are also observed in human neuroblastoma and retinoblastoma cell lines.

CSF N-CAM in neuroleptic-naïve first-episode patients with schizophrenia.

An increased concentration of neural cell adhesion molecule (N-CAM) 105-115 kDa has been reported in patients with schizophrenia in both CSF and in post-mortem brain samples. To determine whether increased N-CAM is integral to the disease process or, alternatively, results from early treatment, CSF N-CAM was measured in a blind study of first episode (FE) patients, who were either neuroleptic-naïve (NN) or neuroleptic-treated (NT, < 100 mg Haldol equivalents), multi-episode (ME) patients, and controls. Overall, the FE patients displayed lower N-CAM concentrations as compared to controls (p = 0.043). This decrease in N-CAM in FE patients was seen only in the FE-NT group as compared to both controls (p = 0.0006). The FE-NT group also showed a lower CSF N-CAM compared to that in the FE-NN (p = 0.025) group. No difference in CSF N-CAM between the FE-NN and control group was found. ME patients showed an increased N-CAM as compared with FE patients (p = 0.018), but not as compared to controls (p = 0.93). Neuroleptic-naïve first-episode patients do not display a phenotypic increase in N-CAM. Thus, N-CAM is altered in first-episode patients following acute neuroleptic treatment and withdrawal, as compared to neuroleptic-naïve first-episode patients.

Elevated concentration of N-CAM VASE isoforms in schizophrenia.

Neural cell adhesion molecule (N-CAM) is a cell recognition molecule, four major isoforms (180, 140, 120, and 105-115 kDa) of which are present in brain. N-CAM has several roles in cellular organization and CNS development. Previously we have found an elevation in CSF N-CAM 120 kDa in the CSF of patients with schizophrenia, bipolar disorder, and depression. We now report an increase in the variable alternative spliced exon (VASE), a 10 amino acid sequence inserted into the fourth N-CAM domain, in the CSF of patients with schizophrenia, but not in bipolar disorder or depression. VASE-immunoreactive (VASE-ir) bands were measured in CSF from patients with schizophrenia (n = 14), bipolar disorder I (n = 7), bipolar disorder II (n = 9), unipolar depression (n = 17) and matched controls (n = 37) by Western immunoblotting. Three VASE-ir bands were distinguished in lumbar CSF corresponding to heavy (165 kDa), medium (155 kDa) and low (140 kDa) MW. A logarithmic transformation was applied to the VASE protein units and analyzed with a MANOVA. There was a 51% and 45% increase in VASE heavy (p = 0.0008) and medium (p = 0.04) MW protein, respectively, in patients with schizophrenia as compared with normal controls. Current neuroleptic treatment in patients with schizophrenia had no effect on CSF VASE concentrations. VASE concentration correlated significantly with behavioral ratings in patients with schizophrenia but not affective disorders. Thus, VASE immunoreactivity is increased in schizophrenia but not in affective disorders. These results provide further evidence of an abnormality of N-CAM protein in chronic schizophrenia and suggest differences between schizophrenia and affective disorders in regulation of N-CAM.

Monozygotic twins discordant for schizophrenia are discordant for N-CAM and L1 in CSF.

While schizophrenia has a genetic component, its pathogenesis is unknown. Abnormal concentrations of two cell recognition molecules (CRMs), neural-cell adhesion molecule (N-CAM) and L1 antigen have been described in the cerebrospinal fluid (CSF) of patients with schizophrenia. Studies of monozygotic twins discordant for schizophrenia may help separate genetic and environmental contributions to the disease. In the present study of monozygotic twins discordant for schizophrenia, the affected twins had increased N-CAM and decreased L1 antigen in their CSF. Non-affected twins were not different from normals. Although processes related to genetic instability cannot be entirely ruled out, these results suggest that these abnormalities are not a part of the genetic predisposition to become schizophrenic. Thus the changes in N-CAM and L1 antigen may reflect either the events which precipitated the onset of schizophrenia, or events which are associated with the experience of having the disease.

N-cadherin expression and function in cultured oligodendrocytes.

N-Cadherin is a major cell adhesion molecule that is expressed in the developing nervous system where it has been implicated in neural migration and axon growth. Recently, a role for N-cadherin in oligodendrocyte differentiation has been identified [23]. Oligodendrocyte precursors adhere to N-cadherin and mature rapidly to produce myelin sheets. Since this implies that oligodendrocytes express N-cadherin, we examined the expression of N-cadherin by oligodendrocytes in culture. N-Cadherin was expressed by O-2A progenitors, immature oligodendrocytes and mature oligodendrocytes, but at a lower level than in type 1 astrocytes in the same cultures. On mature oligodendrocytes, the N-cadherin was concentrated on the major processes emerging from the soma. The ability of N-cadherin and merosin to promote oligodendrocyte precursor migration was also studied. Average migration rates were significantly higher on merosin (11.2 microns/h) than on N-cadherin (5.6 microns/h). These results suggest that N-cadherin is not likely to function predominantly as a substrate that stimulates migration of O-2A progenitors, but may be more important in initiating early oligodendrocyte-axon interactions that promote the process of myelination.

Expression of neural cell adhesion molecule in normal and neoplastic human neuroendocrine tissues.

The neural cell adhesion molecule (N-CAM) is a group of cell surface glycoproteins involved in direct cell--cell adhesion. N-CAM expression in normal and neoplastic tissues was examined with specific antibodies and oligonucleotide probes by immunohistochemistry and in situ hybridization. Most neuroendocrine cells and tumors with secretory granules expressed N-CAM protein and mRNA. Parathyroid adenomas (4) were somewhat unusual, because N-CAM mRNA, but not protein, was detected in some of these benign neoplasms. Most non-neuroendocrine cells and tumors did not express N-CAM, although uterine smooth muscle and an adrenal cortical carcinoma were both positive. Western blots disclosed proteins of 180, 140, and 120 kd in normal adult brain, whereas two pheochromocytomas, a null cell adenoma, and a gastrinoma had proteins of approximately 180 and 140 kd. These results indicate that N-CAM protein and mRNA are widely expressed in neuroendocrine cells and neoplasms. N-CAM oligonucleotide probes as well as antibodies against N-CAM can be used as broad-spectrum neuroendocrine markers. In addition, these molecular probes can be used to examine the role of N-CAM in the development and regulation of neuroendocrine tissues.

In vitro translation and processing of a precursor form of favin, a lectin from Vicia faba.

Favin, the glucose- and mannose-binding lectin isolated from fava (Vicia faba) beans, consists of two polypeptide chains (alpha, Mr = 5,571; beta, Mr = 20,700). Translation of fava bean mRNA in vitro in a wheat germ-derived system yields a single favin polypeptide chain of Mr = 29,000. This molecule appears to consist of a hydrophobic 29-amino acid residue signal sequence at the NH2 terminus followed by the beta chain sequence; it also includes the alpha chain sequence. These results suggest that the alpha and beta chains arise by post-translational cleavage of a single precursor polypeptide: signal-beta chain-alpha chain. The signal peptide is similar in sequence to those seen in animal and prokaryotic systems, suggesting that translocation mechanisms are highly conserved. Translation of favin mRNA in the presence of dog pancreas microsomal membranes yields at least three polypeptides in addition to the presumed precursor chain. The largest of these molecules is translocated into the lumen of the membrane vesicles and glycosylated but its signal sequence remains intact. The two other species are translocated and glycosylated, but their signal sequences have been removed; they appear to differ from each other in that one begins with the beta chain sequence and the other begins one residue after the NH2-terminal threonine of the beta chain. These three variants could reflect normal features of the processing of the favin precursor but more likely result from aberrant processing of the plant protein by dog pancreas membranes.

Amino acid sequence and variant forms of favin, a lectin from Vicia faba.

We have determined the complete amino acid sequence (182 residues) of the beta chain of favin, the glucose-binding lectin from fava beans (Vicia faba), and have established that the carbohydrate moiety is attached to Asn 168. Together with the sequence of the alpha chain previously reported (Hemperly, J. J., Hopp, T. P., Becker, J. W., and Cunningham, B. A. (1979) J. Biol. Chem. 254, 6803-6810), these data complete the analysis of the primary structure of the lectin. We have also examined minor polypeptides that appear in all preparations of favin. Two lower molecular weight species (Mr = 9,500-11,600) appear to be fragments of the beta chain resulting from cleavage following Asn 76, whereas six high molecular weight forms (Mr = 25,000 or greater) appear to include aggregates of the beta chain and possibly some alternative products of chain processing.

cDNA clones of the neural cell adhesion molecule (N-CAM) lacking a membrane-spanning region consistent with evidence for membrane attachment via a phosphatidylinositol intermediate.

In embryonic chicken brains, the neural cell adhesion molecule N-CAM is expressed mainly as two polypeptides, the large intracellular-domain polypeptide (ld) (Mr = 160,000) and the small intracellular-domain polypeptide (sd) (Mr = 130,000) chains, that differ in their cytoplasmic domains and that arise by alternative splicing of RNA transcribed from a single gene. There is evidence for a minor N-CAM polypeptide of Mr = 120,000 that is similar to the ld and sd chains for most of its amino-terminal sequence, but which lacks a cytoplasmic domain. We report here the isolation and characterization of a cDNA clone, lambda N151, that appears to encode this third N-CAM polypeptide, which we designate the ssd (small surface-domain) polypeptide chain. The cDNA insert of lambda N151 consists of 2437 base pairs (bp). DNA hybridization and sequencing indicate that the first 1721 bp are nearly identical to the corresponding sequences of clone lambda N208, which encodes the ld chain. Following in the same reading frame, lambda N151 encodes 25 amino acids not present in lambda N208. The rest of lambda N151 consists of a 637-bp noncoding region containing an AATACA polyadenylylation sequence and a 55-bp poly(A) tract. Messenger RNAs complementary to lambda N151 appear later in development than those complementary to the ld and sd chains, and their appearance is correlated with the appearance of the ssd polypeptide. Although the polypeptide encoded by lambda N151 lacks a membrane region that would define a cytoplasmic domain, it does contain at its carboxyl end a relatively hydrophobic stretch of amino acids similar to those seen in precursors of membrane proteins that are attached to membranes via the lipid phosphatidylinositol. We show here that the ssd chain of chicken N-CAM can be released from brain vesicles by treatment with phospholipase C, suggesting that it too may have a phosphatidylinositol anchor. These results define two additional modes by which N-CAM expression can be modulated: by RNA splicing at a new site and by differential membrane attachment of the resulting polypeptide through a lipid intermediate.

Structure of the gene for the liver cell adhesion molecule, L-CAM.

The liver cell adhesion molecule, L-CAM, mediates calcium-dependent cell-cell adhesion in early embryos and in nonneural epithelia in adult tissues. Earlier studies of cDNAs for chicken L-CAM established the amino acid sequence of the mature protein. The sequence has now been extended in the 5' direction through the precursor and signal sequences and past a consensus translation initiation site. The combined cDNAs were used to isolate genomic clones covering the entire L-CAM coding sequence. The structural gene for chicken L-CAM contains 16 exons ranging in size from 115 to over 1045 base pairs with an average size of 222 base pairs. Single exons do not correspond to known structural elements such as the signal sequence, precursor segment, internal repeats, or membrane-spanning region of L-CAM. Hybridization of restriction digests of chicken genomic DNA with cDNA and genomic probes indicated that there is a single L-CAM gene in the chicken. In contrast to genes for other cell-cell or cell-substrate adhesion molecules, there is no evidence for alternative splicing of exons in this gene.

Monovalent derivatives of concanavalin A.

Monovalent dimers of concanavalin A (Con A) have been prepared by a combination of succinylation and photoaffinity labeling. Partial derivatization of native Con A using the photoaffinity label, p-azidophenyl-alpha-D-mannopyranoside, followed by affinity chromatography yielded a fraction that consisted of dimers with a single saccharide-binding site at pH 5. These monovalent dimers formed divalent tetramers at pH 7. In order to achieve a monovalent dimer at this pH, the divalent tetramers were succinylated by previously developed methods. Ultracentrifugation, equilibrium dialysis, and chromatographic experiments indicated that the resultant preparations consisted mainly of monovalent dimers which showed subunit exchange to yield about 15% divalent dimers after 12 hr at physiological pH. Freshly prepared material failed to agglutinate sheep erythrocytes at concentrations 500-fold higher than native tetravalent Con A. In addition, they showed saturating dose-response curves of mitogenic stimulation of mouse splenic lymphocytes. These curves resembled those of divalent succinyl-Con A but not those of the native molecule. Further development of methods for preparing stable monovalent derivatives of Con A should allow a refined analysis of the effects of lectin valence at the cell surface.

NCAM140 interacts with the focal adhesion kinase p125(fak) and the SRC-related tyrosine kinase p59(fyn).

Axonal growth cones respond to adhesion molecules and extracellular matrix components by rapid morphological changes and growth rate modification. Neurite outgrowth mediated by the neural cell adhesion molecule (NCAM) requires the src family tyrosine kinase p59(fyn) in nerve growth cones, but the molecular basis for this interaction has not been defined. The NCAM140 isoform, which is found in migrating growth cones, selectively co-immunoprecipitated with p59(fyn) from nonionic detergent (Brij 96) extracts of early postnatal mouse cerebellum and transfected rat B35 neuroblastoma and COS-7 cells. p59(fyn) did not associate significantly with the NCAM180 isoform, which is found at sites of stable neural cell contacts, or with the glycophosphatidylinositol-linked NCAM120 isoform. pp60(c-)src, a tyrosine kinase that promotes neurite growth on the neuronal cell adhesion molecule L1, did not interact with any NCAM isoform. Whereas p59(fyn) was constitutively associated with NCAM140, the focal adhesion kinase p125(fak), a nonreceptor tyrosine kinase known to mediate integrin-dependent signaling, became recruited to the NCAM140-p59(fyn) complex when cells were reacted with antibodies against the extracellular region of NCAM. Treatment of cells with a soluble NCAM fusion protein or with NCAM antibodies caused a rapid and transient increase in tyrosine phosphorylation of p125(fak) and p59(fyn). These results suggest that NCAM140 binding interactions at the cell surface induce the assembly of a molecular complex of NCAM140, p125(fak), and p59(fyn) and activate the catalytic function of these tyrosine kinases, initiating a signaling cascade that may modulate growth cone migration.