(Received for publication, December 17, 1996)
From the Department of Biochemistry, School of Medicine, University
of North Carolina, Chapel Hill, North Carolina 27599-7260 and the
Department of Neurobiology, Becton Dickinson Research
Center, Research Triangle Park, North Carolina 27709
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 p59fyn 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 p59fyn from nonionic detergent (Brij 96) extracts of early postnatal mouse cerebellum and transfected rat B35 neuroblastoma and COS-7 cells. p59fyn 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. pp60c-src, a tyrosine kinase that promotes neurite growth on the neuronal cell adhesion molecule L1, did not interact with any NCAM isoform. Whereas p59fyn was constitutively associated with NCAM140, the focal adhesion kinase p125fak, a nonreceptor tyrosine kinase known to mediate integrin-dependent signaling, became recruited to the NCAM140-p59fyn 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 p125fak and p59fyn. These results suggest that NCAM140 binding interactions at the cell surface induce the assembly of a molecular complex of NCAM140, p125fak, and p59fyn and activate the catalytic function of these tyrosine kinases, initiating a signaling cascade that may modulate growth cone migration.
Neuronal migration, axon pathfinding, and fasciculation are fundamental processes that underlie the formation of accurate synaptic connections during development. These processes are governed by interactions of cell adhesion molecules, extracellular matrix, and neurotrophic factors with receptors on the neuronal cell surface. How these interactions are integrated and translated biochemically into molecular signals that guide growth cones to their synaptic targets is not well understood.
NCAM,1 a neural cell adhesion molecule of the immunoglobulin (Ig) superfamily, promotes axon growth, fasciculation, and cell adhesion by homophilic and heterophilic interactions (1). NCAM has a complex expression pattern due to alternative splicing, developmental regulation, and posttranslational processing, producing a number of isoforms. Alternative splicing of a single gene results in three major NCAM isoforms as follows: transmembrane forms of 140 and 180 kDa, and a 120-kDa glycophosphatidylinositol (GPI)-linked isoform (2). The cytoplasmic domains of the transmembrane isoforms lack catalytic activity but may mediate interactions with intracellular cytoskeletal and signaling molecules. NCAM180 is identical to NCAM140 except for a 261-amino acid insert in the cytoplasmic domain (2). This insert confers the potential for interaction with spectrin and reduces lateral mobility of NCAM in the plasma membrane (3). NCAM140 is present in free, migratory growth cones, whereas NCAM180 is found at sites of cell-cell contact, where it may be involved in stabilization of synapses (3, 4). NCAM120 is present in some neurons but is mainly found in glia (5). All three isoforms can be posttranslationally modified by addition of polysialic acid, a carbohydrate moiety that modulates axon guidance (6, 7). NCAM can also be expressed as an isoform with a 10-amino acid insertion in the 4th Ig domain (VASE isoform), resulting in down-regulation of its axon growth-promoting ability (8-10). NCAM functions in the adult as a modulator of learning, memory, and synaptic plasticity, as NCAM antibodies reduce long term potentiation in rat hippocampal slices (11), and mice with a total NCAM gene knockout have deficits in spatial memory (12). L1, another transmembrane glycoprotein of the Ig superfamily found in growth cones and axons, plays similar roles in adhesion and neurite growth and may also function in learning and memory (11, 13-15).
Stimulation of NCAM or L1 on the cell surface by homophilic binding or by binding of antibodies that recognize extracellular determinants of NCAM or L1 evokes changes in protein tyrosine phosphorylation (16, 17) and other intracellular signaling responses including calcium rise, pH changes, and altered phosphoinositide turnover (18-20). Atashi et al. (16) first demonstrated that such "triggering" of NCAM and L1 modulates tyrosine phosphorylation of proteins associated with growth cone membranes, and Klinz et al. (17) showed that this occurs by changes in the activities of both tyrosine kinases and tyrosine phosphatases. Studies with tyrosine kinase inhibitors such as genistein have demonstrated positive as well as negative effects of tyrosine phosphorylation on neurite outgrowth (21-24), and mutational studies in Drosophila have defined functions for certain transmembrane tyrosine phosphatases in regulating axon guidance (25, 26).
Two members of the src family of nonreceptor tyrosine kinases, p59fyn and pp60c-src, act as positive regulators of neurite growth (27, 28). During the major period of neuronal process outgrowth, these kinases are widely expressed on many axonal tracts, where they are localized to the plasma membrane of growth cones and axons (29-31). p59fyn and pp60c-src exhibit a remarkable specificity for stimulating neurite growth on NCAM and L1, respectively. Cerebellar and dorsal root ganglion neurons from fyn-minus mice display complete inhibition of NCAM-dependent neurite growth on NCAM140-expressing fibroblast monolayers, whereas src-minus and yes-minus neurons show unimpaired neurite growth on NCAM (27). On an L1 substrate, neurons from src-minus but not fyn- or yes-minus mice are impaired for neurite outgrowth, although in this case neurite growth is reduced by only 50% (28). These results suggest the existence of separately regulated pathways for NCAM and L1 signaling.
NCAM and L1 signaling pathways appear to be capable of some degree of functional compensation, as fyn-minus and src-minus mice do not have severe neurological phenotypes. However, the hippocampus of fyn-minus mice is mildly affected, showing loosely organized dendrites of CA1 pyramidal cells, increased neuronal number, and blunted long term potentiation (32). Another strain of fyn-minus mice displays partially impaired myelination (33). src-minus mice show no neurological abnormalities but instead develop osteopetrosis, a bone remodeling disease of osteoclasts (34). Because both NCAM and L1 are coexpressed on many neuroanatomical tracts, the effects of eliminating either p59fyn or pp60c-src might be minimized due to compensation by the other intact adhesion pathway. In support of this possibility, src/fyn double mutants show defective axonal growth in vivo (35) and die perinatally (36). Such separately regulated adhesion pathways may serve to minimize errors in axonal pathfinding.
A biochemical approach has been undertaken to identify the molecular components of the NCAM signaling pathway. Here it is reported that p59fyn but not pp60c-src associated preferentially with the NCAM140 isoform. In addition the focal adhesion kinase (p125fak) (37), a nonreceptor tyrosine kinase that mediates extracellular matrix signaling through integrin receptors, was recruited to the complex by NCAM cross-linking, and both p125fak and p59fyn became rapidly and transiently phosphorylated on tyrosine upon NCAM stimulation. These results suggest that NCAM140, p125fak, and p59fyn comprise a functional adhesion signaling complex that may modulate growth cone motility.
Cell Cultures, Antibodies, and NCAM Fusion Protein
Simian COS-7 cells were obtained from the American Type Culture Collection (ATCC, Rockville, MD). Rat B35 neuroblastoma cell lines stably expressing cDNAs encoding NCAM140 or -180 (9) were obtained from the laboratory of Dr. Richard Akeson (Children's Hospital Medical Center, Cincinnati, OH). All cell lines were cultured in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. The following monoclonal (mAb) and polyclonal (pAb) antibodies were used: p59fyn pAb FYN3 (Santa Cruz, CA), pp60c-src mAb 327 (38), NCAM mAbs 14.2 and 16.2 (Becton Dickinson, Inc.), NCAM mAb 5B8 (Developmental Studies Hybridoma Bank, University of Iowa), NCAM mAb 310 (Chemicon International Inc., Temecula, CA), NCAM pAb 1505 (Sigma), p125fak pAbs BC3 and HUB3 (M. Schaller, University of North Carolina; T. Parsons, University of Virginia), phosphotyrosine mAb 4G10 (Upstate Biotechnology Inc., Lake Placid, NY), normal rabbit and mouse IgG (Sigma), and goat anti-rat and anti-mouse IgG (Sigma).
Chinese hamster ovary cells expressing a fusion protein consisting of the entire NCAM extracellular domain fused to the Fc region of human IgG were provided by Melitta Schachner (Swiss Federal Institute of Technology). The NCAM-Fc protein was isolated from conditioned medium by affinity chromatography using Protein G-Sepharose.
Plasmids, Transfection, and Extract Preparation
The eukaryotic expression vector pcDNA3 containing the human cytomegalovirus promoter was used for all constructs. Plasmid constructs contained full-length cDNA clones encoding the B isoform of mouse p59fyn, which is expressed in brain and other tissues (39), the mouse c-src+ isoform containing a 6-amino acid insert in the SH3 domain (40), the chicken c-src isoform lacking that insert (M. Schaller, University of North Carolina), or the rat NCAM140 and NCAM180 isoforms with and without VASE sequences (R. Akeson laboratory, Children's Hospital Medical Center, Cincinnati, OH).
COS-7 or B35 neuroblastoma cells (2 × 106 cells/plate) were transferred to serum-free Opti-MEM (Life Technologies, Inc.) and transfected with indicated plasmid DNA (10 µg) for 8 h at 37 °C using lipofectamine (Life Technologies, Inc.). Fetal calf serum (10%) was added, and cells were incubated for a further 16 h. Medium was replaced with either Opti-MEM containing 10% fetal calf serum (COS-7 cells) or 0.5% fetal calf serum and 1 mM dibutyryl cAMP to induce neuronal differentiation (B35 cells). Cells were incubated for a further 24 h before lysis or NCAM cross-linking. For lysis the medium was removed and cells were incubated for 5 min at 4 °C with Opti-MEM containing 1 mM sodium pervanadate to inhibit tyrosine phosphatases. Cells were then solubilized at 4 °C in a nonionic detergent buffer (Brij lysis buffer) containing 1% Brij 96, 150 mM NaCl, 10 mM Tris-HCl, pH 7.4, 1 mM NaEDTA, 1 mM NaEGTA, 500 µg/ml Pefabloc (Boehringer Mannheim, Mannheim, Germany), 200 µM Na3VO4, 10 mM NaF, 0.01% leupeptin, 0.11 trypsin-inhibiting units/ml aprotinin. Lysates were passed through a 22-gauge syringe and clarified by centrifugation at 10,000 × g for 20 min at 4 °C. Cerebella were isolated from wild type C57Bl/6 mice (Harlan Sprague Dawley) at postnatal day 4 as described (27), and Brij extracts were prepared similarly. Protein concentrations were determined by the micro-BCA method (Pierce).
The efficiency of solubilization of p59fyn and pp60c-src was evaluated in pilot experiments with a membrane fraction from fetal rat brain using a variety of nonionic detergents including Brij 96, Triton X-100, digitonin, N-octylglucoside, CHAPS, and sodium deoxycholate. Only Brij 96 (0.1-1%) and to a lesser extent Triton X-100 (1%) afforded complete solubilization with retention of kinase activity, as judged by the distribution of p59fyn and pp60c-src protein and kinase activities in supernatant and pellet fractions (17).
Antibody-induced Cross-linking and Incubation with NCAM Fusion Protein
COS-7 or B35 neuroblastoma cells cultured in 100-mm tissue culture dishes were rinsed in serum-free Opti-MEM then incubated with NCAM mAb (25 µg/ml; Chemicon 310 or NCAM 16.2) or normal rat IgG for 30 min at 4 °C. Cells were rinsed with Opti-MEM and incubated with 5 µg/ml anti-rat or anti-mouse IgG for 20 min at 4 °C and then transferred to 37 °C. Alternatively, cells were incubated with the NCAM Fc-fusion protein (50 µg/ml) for 30 min at 4 °C. Prior to lysis, cells were incubated for 5 min in Opti-MEM containing 1 mM sodium pervanadate at 4 °C; then extracts were prepared in Brij lysis buffer (0.75 ml). NCAM immunoprecipitates were collected by the addition of 35 µl of a 1:1 (v/v) suspension of Protein G-Sepharose in Brij lysis buffer.
Immunoprecipitation, in Vitro Kinase Assays, and Immunoblotting
ImmunoprecipitationCell lysates were precleared by incubation with normal IgG for 30 min and Protein A- or Protein G-Sepharose beads (Sigma) for 30 min followed by centrifugation at 14,000 rpm. Primary antibody in excess amount (1 µg) or normal IgG was added to equal amounts of extract (500 µg to 1 mg) in a 1-ml volume, and extracts were incubated with gentle inversion for 1 h at 4 °C. Immune complexes were recovered by the addition of 40 µl of Protein A- or Protein G-Sepharose for polyclonal or monoclonal antibodies, respectively. Pellets were collected by centrifugation at 10,000 × g for 1 min and washed 4 times with Brij lysis buffer.
In Vitro Kinase AssaysImmunoprecipitates were incubated in
a kinase reaction buffer (30 µl) containing 7.5 µCi of
[-32P]ATP (3000 Ci/mmol), 50 mM Tris-HCl,
pH 7.4, 3 mM MnCl2, 3 mM MgCl2, 100 mM NaCl, 100 µM
Na3VO4, and 1 µM ATP. In
vitro kinase assays of immune complexes from cerebellum were
performed with 15 µCi of [
-32P]ATP (3000 Ci/mmol)
per reaction without additional unlabeled ATP. Phosphorylation was
allowed to proceed for 20 min at room temperature. Beads were washed
and proteins eluted by boiling for 10 min in 3% SDS, 50 mM
Tris-HCl, pH 7.4, 150 mM NaCl, and 100 µM
Na3VO4, with removal of beads by
centrifugation. For re-immunoprecipitation, supernatants were diluted
1/12 in Brij lysis buffer, incubated with antibody (1 µg) for 1 h, and Protein A- or Protein G-Sepharose for 30 min. The beads were
washed 3 times, heated to 95 °C for 3 min in SDS sample buffer, and
the supernatant proteins subjected to SDS-PAGE, pH 8.8. Autoradiography
on Dupont Chronex 6-plus x-ray film was carried out with intensifying
screens at
70 °C.
NCAM140 was immunoprecipitated from
Brij lysis buffer extracts of mouse cerebellum or cell cultures under
conditions of antibody excess, and in vitro kinase assays
were carried out as described above and in Fig. 1A. Proteins
were eluted from the immune complexes, incubated with Fyn or Fak
antibodies, and separated by SDS-PAGE under reducing conditions. The
amount of 32P-labeled p59fyn or p125fak was
quantitated by Cerenkov counting of excised gel bands and compared with
the total amount of 32P-labeled p59fyn or
p125fak immunoprecipitated from the same amount of extract
protein. In some experiments stoichiometry was estimated by
densitometric scanning of bands obtained by enhanced chemiluminescence
of immunoblots.
Immunoblotting
Proteins were separated under nonreducing
(p59fyn) or reducing (p125fak) conditions by SDS-PAGE,
pH 8.8, and transferred to nitrocellulose filters (Schleicher & Schuell). Filters were blocked for 2 h in 2% bovine serum albumin
and incubated overnight with primary antibodies. After 5 washes of 5 min each in TBS/Tween buffer (0.02 M Tris-HCl, pH 7.4, 150 mM NaCl, 0.05% Tween 20), filters were incubated with a
1:10,000 dilution of goat anti-rabbit secondary antibody conjugated to
horseradish peroxidase (Sigma) for 1 h at room temperature, followed by 5 washes in TBS/Tween buffer. Proteins were visualized by
enhanced chemiluminescence (Amersham Corp.). To reprobe immunoblots the
membranes were stripped for 30 min at 50 °C in 10% SDS, 0.02 M Tris, pH 6.8, 100 mM -mercaptoethanol, and
rinsed with TBS/Tween.
At postnatal day 4 (PND4) the mouse cerebellum
contains many developing neurons engaged in migration and axonal
growth, and it expresses p59fyn, pp60c-src
(27), and the three major NCAM isoforms (NCAM180, -140, and -120) (Fig.
1). To investigate a potential physical association between p59fyn and NCAM in the developing cerebellum, NCAM and
associated proteins were immunoprecipitated from extracts of PND4 mouse
cerebellum prepared in nonionic Brij 96 detergent-containing buffer
(Brij lysis buffer) using a pool of two NCAM monoclonal antibodies
recognizing all three NCAM isoforms. The resulting immunoprecipitates
were subjected to in vitro kinase assays with
[-32P]ATP to label active tyrosine kinases present in
the NCAM complexes by autophosphorylation. This method was used because
it afforded greater sensitivity and quantitation than immunoblotting.
Immune complexes were solubilized in 3% SDS, diluted 1/12 in Brij
lysis buffer, and re-immunoprecipitated with Fyn or Src antibodies. p59fyn was found to co-immunoprecipitate with NCAM from the
mouse cerebellar extracts, whereas pp60c-src did
not significantly co-immunoprecipitate with NCAM (Fig. 1A). p59fyn and pp60c-src kinase activities were
expressed at approximately equal levels in the PND4 mouse cerebellum
(Fig. 1A), indicating a selective association of NCAM with
p59fyn, which was in accord with the neurite outgrowth
properties of fyn- and src-minus neurons
(27).
The specificity of association between NCAM and p59fyn was ascertained by a reverse immunoprecipitation protocol in which p59fyn or pp60c-src was first immunoprecipitated from mouse cerebellar extracts (PND4), and the resulting complexes were assayed for the presence of NCAM by immunoblotting with an NCAM polyclonal antibody recognizing all three isoforms. This approach also revealed the particular isoform(s) of NCAM co-immunoprecipitating with p59fyn. Mouse cerebellum expressed equivalent amounts of NCAM180 and NCAM140 and a small amount of NCAM120 at this stage in development (Fig. 1B). The broad bands were most likely due to polysialylation of NCAM isoforms. However, only the NCAM140 isoform co-immunoprecipitated with p59fyn (Fig. 1B). The sharpening of the NCAM140 band probably resulted from desialylation during heat treatment in SDS. Neither the NCAM180 transmembrane isoform nor the GPI-linked NCAM120 isoform were present to a significant degree in the p59fyn immune complexes. In addition, none of the NCAM isoforms co-immunoprecipitated with pp60c-src.
By comparing the amount of 32P-labeled p59fyn in
the NCAM immunoprecipitates to the total amount of
32P-labeled p59fyn in immunoprecipitates from an
equivalent amount of mouse cerebellar extract, it was estimated that
1% of the p59fyn molecules in the PND4 mouse cerebellum
associated with NCAM140. Conversely, by comparing the amount of NCAM140
in p59fyn immunoprecipitates to total NCAM140
immunoprecipitated from mouse cerebellar extracts by densitometric
scanning (Fig. 1B), it was estimated that approximately 3%
of the NCAM140 in mouse cerebellum was associated with p59fyn.
This stoichiometry approximated that of p59fyn associated with
the T cell receptor subunit (1-5%) (41), myelin-associated
glycoprotein (5%) (33), and B cell receptor protein Ig-
(3-7%)
(42), and was suggestive of a low affinity interaction. However, the
actual stoichiometry could be higher, since some complexes might
dissociate during lysis.
Cell lines expressing transfected NCAM or
fyn cDNAs were used to further investigate the
specificity of association of p59fyn with NCAM isoforms. Simian
COS-7 cells were cotransfected for transient expression with pcDNA3
plasmids containing cDNAs encoding NCAM140 or NCAM180 (each lacking
the VASE sequence), together with plasmids encoding brain-enriched
forms of p59fyn (B form) or pp60c-src
(src+ isoform with a 6-amino acid insert). Pilot
enzyme-linked immunosorbent assay results showed that COS-7 cells
expressed very low levels of endogenous p59fyn and
pp60c-src, which did not contribute significantly
to the high levels expressed from transfected cDNAs. NCAM was
immunoprecipitated from Brij lysates of transfected COS-7 cells 48 h after transfection, and immune complexes were subjected to in
vitro kinase reactions with [-32P]ATP. Proteins
in the immune complexes were solubilized in 3% SDS, diluted 1/12 in
Brij lysis buffer, and re-immunoprecipitated with Fyn or Src
antibodies. This experiment showed that the B isoform of p59fyn
associated strongly with NCAM140 (without VASE) and to a much lesser
degree with NCAM180 (Fig. 2A). Conversely,
pp60c-src+ was not associated with either NCAM140
or NCAM180 (Fig. 2A). The alternative form of
pp60c-src lacking the insert in the SH3 domain also
failed to associate with NCAM140 or NCAM180 when transiently expressed
in COS-7 cells (not shown). An endogenous protein of approximately 125 kDa was phosphorylated in the NCAM140 immune complexes, along with a
59-kDa protein which most likely represented p59fyn, and
several proteins in the 45-55-kDa range (Fig. 2A, lane 7). Notably, the 125-kDa phosphoprotein persisted in the Fyn
re-immunoprecipitates after solubilization of the immune complexes in
SDS and dilution in Brij lysis buffer (lane 2). This most
likely resulted from reassociation after dilution of the detergent and
a further incubation for 1 h at 4 °C. The 125-kDa protein was
not present in either the Fyn re-immunoprecipitates from NCAM180 immune
complexes (lane 4) or the Src re-immunoprecipitates from
immune complexes containing either NCAM isoform (lanes 5 and
6). In the reverse immunoprecipitation, p59fyn was
first immunoprecipitated from the transfected COS-7 cells followed by
NCAM immunoblotting (Fig. 2B). These results confirmed that
NCAM140 and little NCAM180 co-immunoprecipitated with p59fyn,
whereas neither isoform of NCAM co-immunoprecipitated with
pp60c-src.
To investigate the p59fyn association in a neuronal cell type
known to express moderate levels of NCAM isoforms on the cell surface, the central nervous system-derived rat B35 neuroblastoma cell line was
used for similar co-immunoprecipitation experiments. Differentiated B35
neuroblastoma cells exhibit neuronal properties including membrane
excitability and expression of enzymes involved in neurotransmitter
metabolism (43). Stably transformed B35 cell lines have been developed
that express NCAM140 or NCAM180 (each with and without VASE) at
equivalent levels on their cell surface (9) (Fig. 2B, lanes 17 and 18). Because the cells express much lower levels of
p59fyn and pp60c-src as detected by
immunoprecipitation (not shown), the B35 cell lines were transiently
transfected with fyn (B form) or src
(c-src+) pcDNA3 plasmids and assayed 48 h later
for association with NCAM. NCAM was immunoprecipitated from Brij
lysates of transfected B35 neuroblastoma cells and immune complexes
subjected to in vitro phosphorylation with
[-32P]ATP. Immune complexes were solubilized and
re-immunoprecipitated with Fyn or Src antibodies. p59fyn was
found to co-immunoprecipitate with NCAM140 and not NCAM180 (each
without VASE), whereas pp60c-src did not
co-immunoprecipitate with either NCAM isoform (Fig. 2A). The
reverse immunoprecipitation confirmed these results (Fig. 2B).
Cerenkov counting of excised p59fyn bands revealed that 3% of the total p59fyn immunoprecipitated from COS-7 cells under conditions of antibody excess preferentially bound to NCAM140, whereas densitometric scanning of NCAM immunoblots indicated that 5% of the total NCAM140 immunoprecipitated from COS-7 cells bound to p59fyn. Equivalent levels of NCAM140 and -180 were expressed in the transfected COS-7 cells as shown by immunoblotting (Fig. 2B, lanes 9 and 10). Immunoperoxidase staining of formaldehyde-fixed cell cultures showed pronounced staining of NCAM140 and NCAM180 on the cell surface (not shown). Similarly, comparable levels of p59fyn and pp60c-src kinase activity were expressed in the transfected COS-7 cells (lanes 8 and 9).
A large portion of NCAM in the adult brain contains the alternatively spliced VASE exon in the fourth Ig domain, a modification that down-regulates neurite outgrowth (8-10). The presence of the VASE exon in NCAM140 or NCAM180 did not alter the association of p59fyn in transfected COS-7 cells (Fig. 2B, lanes 3 and 6) indicating that the neurite growth inhibitory effect of the VASE isoform was not due to p59fyn dissociation from the NCAM complex.
p59fyn has been reported to associate with the GPI-linked proteins F3/F11/contactin (44), Thy-1 (45), decay accelerating factor, or CD59 (46), but these associations can be disrupted in N-octylglucoside, a nonionic detergent resembling glycolipids (44, 45). The association of p59fyn with NCAM140 was stable in cell extracts prepared with N-octylglucoside (1%) or other nonionic detergents CHAPS and Triton X-100 (1%). Stability in 1% N-octylglucoside indicated that the p59fyn-NCAM140 association was not likely to be mediated by a GPI-linked molecule. Moreover, p59fyn did not co-immunoprecipitate with the GPI-linked NCAM120 isoform from mouse cerebellum (Fig. 1).
Identification of the p59fyn-associated Protein as p125fakBecause p59fyn was known to associate
with the focal adhesion kinase p125fak in nonneuronal cells
(47), it was logical to investigate whether the 125-kDa protein that
was phosphorylated in the NCAM140-p59fyn immune complexes from
COS-7 cells was p125fak. To this end NCAM140 was
immunoprecipitated from Brij extracts of COS-7 cells transiently
expressing NCAM140 and fyn cDNAs. The resulting NCAM
immune complexes were subjected to in vitro kinase assays
with [-32P]ATP, solubilized, and re-immunoprecipitated
with Fak antibodies. A 32P-labeled 125-kDa protein
specifically immunoprecipitated with Fak antibodies (Fig.
3A, lane 3). This protein was also evident in
re-immunoprecipitations with Fyn antibodies (Fig. 2A, lane 2; Fig. 3A, lane 2). p59fyn was not evident in
the Fak re-immunoprecipitations (Fig. 3A, lane 3) possibly
due to steric hindrance of p59fyn by the Fak antibody. COS-7
cells expressed high levels of endogenous p125fak (Fig.
3A, lane 4), which approximated the levels of p59fyn
transiently expressed from the transfected fyn plasmid
(lane 5). As shown in Fig. 2A, p125fak
did not co-immunoprecipitate with NCAM180 or with
pp60c-src from COS-7 cells expressing either
NCAM180 or NCAM140. In the reverse immunoprecipitation protocol,
p125fak was immunoprecipitated from Brij extracts of COS-7
cells transiently expressing NCAM140 and fyn cDNAs, and
the immune complexes were subjected to immunoblotting with NCAM
antibodies (Fig. 3B). NCAM140 was found to specifically
co-immunoprecipitate with p125fak.
To estimate the stoichiometry of the association, the amount of 32P-labeled p125fak re-immunoprecipitated from solubilized NCAM140 immune complexes was compared with the total amount of 32P-labeled p125fak immunoprecipitated from an equal amount of COS-7 cell extract. Approximately 1% of the p125fak expressed in COS-7 cells was present in NCAM140 immune complexes. In the reverse immunoprecipitation, densitometric scanning indicated that approximately 3% of the NCAM140 expressed in COS-7 cells associated with p125fak, a stoichiometry of similar magnitude to that of the NCAM140-p59fyn association. A 125-kDa protein in association with the NCAM140-p59fyn complex was not detected by co-immunoprecipitation from mouse cerebellar extracts (Fig. 1). This may be due to lower expression of p125fak in the cerebellum at PND4 or less recruitment of p125fak to NCAM complexes at this stage.
NCAM Binding Induces the Phosphorylation and Recruitment of p125fakTo examine the functional interaction between
p125fak and NCAM, tyrosine phosphorylation of p125fak
was assayed following antibody-induced cross-linking of NCAM140. COS-7
cells transiently expressing NCAM140 and p59fyn were incubated
for 30 min at 4 °C with an NCAM monoclonal antibody (mAb 16.2)
directed against an extracellular epitope in the homophilic binding
site of NCAM under conditions that allowed antibody binding but
prevented internalization (48). Secondary antibodies were then added
and cells transferred to 37 °C to cross-link NCAM molecules on the
cell surface. At various times after treatment cells were lysed in Brij
lysis buffer and p125fak was immunoprecipitated with Fak
antibodies. Immunoblotting was carried out with phosphotyrosine
antibodies and then the blot was stripped and reprobed with Fak
antibodies. This experiment revealed an increase in the tyrosine
phosphorylation of p125fak without a significant change in
p125fak protein (Fig. 4). Maximum
phosphorylation was observed 5 min following NCAM antibody treatment
and then diminished, possibly due to the action of tyrosine phosphatase
activity in cells. Omission of secondary antibodies did not stimulate
tyrosine phosphorylation of p125fak (not shown). By
densitometric scanning it was estimated that total p125fak
tyrosine phosphorylation was stimulated approximately 50-fold following
NCAM antibody ligation. To address whether these changes were also
induced by NCAM protein, COS-7 cells were treated with a soluble NCAM
fusion protein in which the NCAM extracellular region was fused to the
Fc portion of human Ig (NCAM-Fc). Immunoprecipitation of
p125fak followed by immunoblotting with phosphotyrosine or Fak
antibodies revealed an increase in specific p125fak tyrosine
phosphorylation (Fig. 4). However, p125fak phosphorylation
induced by NCAM-Fc was somewhat slower and less pronounced than
antibody-induced ligation of NCAM.
To address whether p125fak was recruited to NCAM complexes,
COS-7 cells transiently expressing NCAM140 and p59fyn were
treated with nonimmune IgG or NCAM monoclonal antibodies followed by
secondary antibodies. Cells were lysed and immune complexes containing
NCAM were collected by precipitation with Protein G-Sepharose.
Immunoblotting with Fak antibodies showed that p125fak was
strongly recruited into NCAM complexes within 5 min of stimulation (Fig. 4). Basal levels of p125fak complexed to NCAM at
t = 0 were not evident by immunoblotting with Fak
antibodies, although they were evident by the more sensitive in
vitro phosphorylation assay (Figs. 2 and 3). p125fak
appeared to dissociate from the NCAM complexes during antibody-induced ligation of NCAM in COS-7 cells, and this occurred concomitant with the
observed dephosphorylation of p125fak shown above. This may
indicate that p125fak recruitment to the NCAM complex was
dependent on p125fak tyrosine phosphorylation. The
p125fak doublet observed in the recruitment experiment was not
reproducibly seen, but often the band appeared broad. A similar Fak
doublet has been previously reported (49) and may be a consequence of differential phosphorylation (50, 51) or proteolytic degradation. The
possibility of cross-reactivity of the Fak antibodies with a
Fak-related kinase cannot be ruled out, but preliminary experiments with antibodies against PYK2/CADTK/CAK (52) (from S. Earp, University of North Carolina) did not show NCAM antibody-induced tyrosine phosphorylation of this kinase.
Unlike p125fak, p59fyn was not recruited to the NCAM complexes upon antibody-induced ligation of NCAM140 in COS-7 cells under the same conditions but instead appeared to be constitutively bound to NCAM140. NCAM complexes isolated at various times after antibody treatment of cells were subjected to immunoblotting with Fyn antibodies and showed levels of p59fyn protein in NCAM complexes that remained more or less unchanged during the 20 min of antibody treatment (Fig. 4).
The 125-kDa phosphoprotein identified as p125fak in NCAM140
immune complexes from COS-7 cells was not evident in NCAM140
immunoprecipitates from unstimulated B35 neuroblastoma cells (Fig. 3).
This may have been due to lower levels of p125fak and to fewer
cell-cell contacts in B35 cultures, which would reduce the basal levels
of p125fak recruited to NCAM. However, NCAM-induced tyrosine
phosphorylation of total immunoprecipitated p125fak could be
measured. Accordingly, B35 cells expressing NCAM140 and p59fyn
were treated with primary NCAM antibodies and secondary antibodies, and
then tyrosine phosphorylation of total cell proteins and
p125fak was examined by immunoblotting with phosphotyrosine
antibodies (Fig. 5A). Transient tyrosine
phosphorylation was observed in several proteins, including those of
125, 110, 85 (doublet), and 59 kDa within 5 min of antibody treatment
(Fig. 5A). Immunoprecipitation with Fak antibodies followed
by immunoblotting with phosphotyrosine antibodies identified the
125-kDa protein as p125fak and demonstrated that
p125fak was rapidly (within 5 min) and transiently
tyrosine-phosphorylated upon antibody-mediated NCAM ligation (Fig.
5B). The amount of p125fak protein was unchanged as
shown by immunoblotting with Fak antibodies (Fig. 5C). The
relative increase in p125fak tyrosine phosphorylation could not
be calculated because basal levels of phosphotyrosine-modified
p125fak were undetectable. Furthermore, p125fak was not
detectable in NCAM immune complexes from B35 cell extracts by
immunoblotting with Fak antibodies after antibody-induced ligation of
NCAM. Because 1-3% of the p125fak expressed in B35 cells
would not be detectable, p125fak could be recruited to NCAM
complexes with a stoichiometry similar to that occurring in COS-7
cells.
NCAM Binding Activates Tyrosine Phosphorylation of p59fyn
Treatment of B35 cells expressing NCAM140 and
p59fyn with primary NCAM antibodies followed by secondary
antibodies caused a rapid and transient tyrosine phosphorylation
of p59fyn (Fig. 6A).
Immunoprecipitation with Fyn antibodies followed by phosphotyrosine
immunoblotting revealed maximal activation of p59fyn
tyrosine phosphorylation (approximately 4-fold) at 5-10 min after cross-linking. The amount of p59fyn protein was unchanged as
shown by immunoblotting with Fyn antibodies (Fig. 6A).
Treatement of cells with two different NCAM antibodies (mAb 16.2 and
mAb 310) elicited the same extent and kinetics of phosphorylation.
p59fyn tyrosine phosphorylation was of similar magnitude to
that induced by antibody-mediated ligation of myelin-associated
glycoprotein (33). Clustering of NCAM on the cell surface appeared to
be necessary for maximal p59fyn phosphorylation, since
phosphotyrosine levels of p59fyn were unchanged in cells
treated with primary NCAM antibodies alone (Fig. 6B,
non-crosslinked). Triggering of B35 cells with the soluble
NCAM-Fc fusion protein also resulted in a transient elevation in
p59fyn tyrosine phosphorylation with maximum phosphorylation at
5-10 min and then declining (Fig. 6C); however, the extent
of phosphorylation was not so great as with antibody-mediated
cross-linking possibly due to a lower state of oligomerization. The
presence of the VASE exon in the NCAM140 isoform did not alter the
extent or kinetics of p59fyn phosphorylation upon NCAM antibody
treatment of B35 neuroblastoma cells stably expressing this isoform
(not shown). One explanation for the smaller increase in tyrosine
phosphorylation of p59fyn compared with p125fak is that
p59fyn may be initially phosphorylated to some degree on its
terminal tyrosine residue (Tyr-531). This residue is known to be
phosphorylated by the tyrosine kinase Csk, which negatively regulates
p59fyn activity (53). Although peptide mapping studies have not
been performed, NCAM binding interactions may induce dephosphorylation of Tyr-531 of p59fyn by activating a tyrosine phosphatase,
followed by activation of autophosphorylation (Tyr-420).
The results reported here demonstrate a physical and functional interaction of the 140-kDa isoform of the neural cell adhesion molecule NCAM with the focal adhesion tyrosine kinase p125fak and the src-related tyrosine kinase p59fyn. Antibody-mediated ligation of cell surface NCAM140 or stimulation with soluble NCAM fusion protein led to a transient increase in tyrosine phosphorylation of both p125fak and p59fyn, suggesting that activation of these nonreceptor tyrosine kinases is a proximal event in the NCAM signal transduction pathway.
The interaction of p59fyn with NCAM was consistent with the impaired NCAM-dependent neurite growth displayed by fyn-minus neurons in culture and the widespread distribution of p59fyn in developing axonal tracts and nerve growth cones (31). The binding preference of NCAM for p59fyn and not pp60c-src was in accord with the impaired NCAM-dependent neurite outgrowth of fyn-minus but not src-minus neurons (27). The distinct but overlapping molecular associations mediated by the SH2 and SH3 domains of src family members provides a molecular basis for substratum-specific cellular responses displayed by growth cones (79). Neurons display different growth cone morphologies when plated on NCAM, L1, N-cadherin, laminin, and p84 suggesting that adhesive contacts and cytoskeletal structure are differentially modulated on these substrates (54-56). A molecular basis for these differences is not easily explained by a model in which a single tyrosine kinase, the fibroblast growth factor receptor, is responsible for neurite outgrowth on NCAM, L1, and N-cadherin (57). Unlike p59fyn, we have not been able to detect tyrosine phosphorylation of the fibroblast growth factor receptor upon NCAM antibody binding in our assays. However, the fibroblast growth factor receptor may provide trophic support that is permissive for neurite extension in certain neuronal cell types.
Strict specificity was displayed by p59fyn and p125fak for the NCAM140 transmembrane isoform. Since NCAM140 is preferentially found in free migratory growth cones in contrast to NCAM180, which is associated with stable cell contacts (3, 4), these kinases may be necessary for migration of growth cones toward their targets. A developmentally regulated isoform switch from NCAM140 to NCAM180 in neurons could facilitate the transition from growth cone to synapse by terminating p59fyn and p59fak signaling, and down-regulation of expression of p59fyn and p59fak during maturation (31, 58) could contribute to such a transition. A primary role for p59fyn compared with pp60c-src in p125fak activation is also indicated by reduced p125fak tyrosine phosphorylation in the brains of fyn-minus but not src-minus mice (59), and preferential complex formation between p125fak and p59fyn in nontransformed, nonneural cells (47).
p59fyn appeared to be constitutively bound to NCAM140, either
directly or indirectly, whereas p125fak was recruited to the
NCAM complex. The binding site for p59fyn (or an adaptor
protein) may reside within the cytoplasmic domain of NCAM140, because
the 261-amino acid insert in the corresponding region of NCAM180
effectively disrupts the association. The NCAM140 "tail" does not
contain any known phosphorylated tyrosine residues for the binding of
the p59fyn SH2 domain (60, 61) or polyproline motifs for the
binding of an SH3 domain (62), and hence the interaction is unlike that of p59fyn with myelin-associated glycoprotein, which is
mediated by the p59fyn SH2 and SH3 domains (33). Instead,
NCAM140 might interact with the amino-terminal unique domain of
p59fyn, which is the most divergent region among src
family kinases and is responsible for low affinity interactions of
p59fyn with the T cell receptor , CD3
, and CD3
subunits (41), and the B cell receptor Ig-
subunit (42). However,
the tail of NCAM140 lacks an immunoreceptor tyrosine-based activation
motif, which mediates the association of p59fyn with the Ig-
subunit (42) and is present in each of these receptors. Differences in
polysialylation within the extracellular region of NCAM140 and -180 might also influence p59fyn binding by altering the
conformation of the cytoplasmic tail or modulating the cis
interaction of NCAM with a possible transmembrane coreceptor.
Polysialylation of NCAM has been shown to be important for tangential
migration of olfactory bulb interneurons (63), and its removal mimics
the phenotype of NCAM-minus mice (64). Mice with gene knockouts for
NCAM180 or total NCAM display similar phenotypes, suggesting that a
putative function of NCAM140 (and p59fyn) in axonal growth or
guidance may be partially compensated by other adhesion signaling
pathways.
A possible mechanism for NCAM signaling based on the results presented
here is that NCAM140 binding interactions in the membrane induce
autophosphorylation of constitutively associated p59fyn,
possibly through activation of a tyrosine phosphatase that
dephosphorylates p59fyn at its negative regulatory site
(Tyr-531). Indeed, NCAM antibodies have been shown to stimulate a
tyrosine phosphatase activity in growth cone-enriched membranes (65).
The p59fyn SH2 domain would then become available to bind and
recruit p125fak. Subsequently, p59fyn may phosphorylate
p125fak at additional tyrosine residues that could then recruit
other SH2 domain-containing signaling or cytoskeletal proteins. A
similar mechanism occurs in nonneuronal cells where antibody-induced
ligation or ligand stimulation of integrins by extracellular matrix
proteins such as fibronectin increases p125fak
autophosphorylation on tyrosine residue 397, creating a binding site
for p59fyn or pp60c-src (47, 66-68).
p125fak is then phosphorylated at additional tyrosine residues
providing a binding site for Grb2 and activating the
Ras-mitogen-activated protein kinase pathway (66, 67, 69). Thus the
involvement of p125fak in NCAM signaling thus raises the
interesting prospect that NCAM may regulate gene transcription.
Additionally, p125fak is known to recruit the p85 subunit of
phosphatidylinositol 3-kinase (70) and the GTPase-activating protein
Graf, a negative regulator of RhoA and Cdc42, which are GTP binding
proteins regulating lamellipodial and filopodial formation (71). In
another model, p125fak may join the NCAM complex indirectly
through association with an integrin (72). Although preliminary
experiments have not demonstrated an association of NCAM and
1-integrin by co-immunoprecipitation from Brij lysates
of mouse cerebellum (PND4) or fetal (E18) rat brain, such interactions
could be weak or involve another subclass of integrin.
The identification of p125fak as a component of the NCAM140-p59fyn complex expands the potential role of the focal adhesion tyrosine kinase from integrin signaling to signal transduction by a cell adhesion molecule of the Ig superfamily. p125fak is expressed in all regions of the central nervous system and is enriched in the growth cones of developing neurons, although its function there has yet to be identified (58, 59). A functional interaction of NCAM and integrins also offers a potential means of convergence of NCAM and integrin-dependent adhesion pathways. Such convergence could provide a means of integrating growth cone guidance cues from extracellular matrix and cell surface adhesion molecules. Such interactions may be permissive for axonal outgrowth or instructive for growth cone guidance at boundaries or guidepost cells. Although p125fak can be activated by clustering of cell surface NCAM140 molecules, it remains to be established whether such activation contributes to axon growth or guidance.
The recent development of Fak knockout mice has provided evidence that p125fak functions in regulating focal adhesive contacts induced by integrin binding of extracellular matrix proteins (73). Fak-minus mice are impaired in the development of the anterior-posterior axis and in mesodermal structures (73, 74). Inhibition of p125fak by the Fak-related nonkinase delays the formation of new focal adhesions (75). Regulated adhesion is expected to be essential for the migration of neuronal growth cones, which establish focal contacts on some substrates but mainly rely upon the making and breaking of transient point contacts during axonal migration (76, 77). Activation of p125fak and p59fyn by signaling clusters of NCAM in the growth cone membrane may regulate the dynamics of adhesive contacts, allowing rapid neuronal migration. In nonneuronal cells, p125fak is activated by integrin cross-linking and is recruited to sites of focal contacts on the cell surface together with a host of cytoskeletal and signaling proteins including p59fyn (78). In an analogous manner, NCAM140 clustering in the growth cone may recruit and activate not only p125fak but other focal contact-associated cytoskeletal and signaling proteins with net effects on filopodial and lamellipodial dynamics.
Dr. Stefan Klinz is gratefully acknowledged for his generous advice and assistance with experiments and many helpful discussions regarding work related to this project. We thank Ron Graff for the NCAM Fc purification, and Terri Worley, Jim Fiordalisi, and Wendy Morse for helpful comments on the manuscript. We are indebted to Drs. Mike Schaller, Shelton Earp, Roger Perlmutter, and Tom Parsons for antibodies and cDNA clones.