Tyrosine Phosphorylation at a Site Highly Conserved in the L1 Family of Cell Adhesion Molecules Abolishes Ankyrin Binding and Increases Lateral Mobility of Neurofascin

SDS–polyacrylamide gel electrophoresis and Western blotting, after transfer of resolved proteins to nitrocellulose, were performed as previously described (Davis and Bennett, 1983, 1984). Briefly, immunoblots were incubated overnight at 4°C with each primary antibody, washed, and then incubated with ¹²⁵I-radiolabeled protein A for 2 h at room temperature. Blots were visualized via autoradiography. The primary antibodies used included the anti-phosphotyrosine polyclonal (1:1,000; Upstate Biotechnology Inc., Lake Placid, NY), a mucin-specific neurofascin polyclonal (1:1,000; Davis et al., 1996), the Nr-CAM–specific polyclonal (1:1,000; Davis et al., 1993), and the brain ankyrin-specific polyclonal antibodies (1: 1,000; Davis and Bennett, 1984).

A 600-bp ApaI–EcoRI fragment of wild-type neurofascin cDNA, which encodes most of the cytoplasmic domain sequence, was subcloned into pBluescript vector (Stratagene, La Jolla, CA). All deletions and point mutations were carried out using the Exsite PCR-based mutagenesis kit (Stratagene). The HA epitope was inserted by PCR six amino acids downstream of the signal peptide. Mutated constructs were confirmed by DNA sequencing before the wild-type ApaI–EcoRI sequence was substituted by the mutated fragment. Both HA epitope-tagged wild-type and cytoplasmic domain-truncated neurofascin were subcloned into pCMV vector at HindIII and BamHI sites. The pCMV vector was derived from pBC12 (Cullen, 1986) with the following modifications: (a) the human IL-2 cDNA sequence (760–1437 bp) was deleted and replaced by a multiple cloning site containing NheI, SmaI, SalI, and XhoI restriction sites; (b) the BamHI fragment of the Neo^R gene with an SV40 promotor was inserted into the BglII site. All the mutated neurofascin cDNA constructs were subcloned into the pC13 CAT vector at HindIII and NotI sites. The pC13 CAT vector was derived from pOP13 CAT (Stratagene) with the RSV promotor of pOP13 CAT being replaced by the CMV promotor from pBC12 at SacII and BglII sites.

Transfection of B104 cells was carried out using lipofectamine (GIBCO BRL, Gaithersburg, MD). For each 35-mm dish, 5 μg of DNA and 10 μl (2 mg/ml) of lipofectamine were used following the instructions provided by GIBCO BRL. 2 d after transfection, 1.5 μg/ml of geneticin (GIBCO BRL) was added to the media, and colonies of neoresistance cells were visible 1–2 wk afterward. Approximately 24 single colonies for each construct were chosen and expanded for Western blot analysis to examine levels of neurofascin expression. Clones with highest expression levels were used for subsequent studies.

After lysis of cells with a modified RIPA buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.5% sodium deoxycholate, 1 mM vanadate, 1 mM benzamidine, 1 mM PMSF, aprotinin [10 μg/ml], and leupeptin [10 μg/ml]), neurofascin was immunoprecipitated overnight at 4°C from 250–500 μg crude cell lysate using an HA-specific monoclonal antibody (BAbCO, Berkeley, CA). Immune complexes were captured with either protein A–Sepharose (Pierce, Rockford, IL) for SDS-PAGE analysis or with protein A–latex beads for use in the in vitro ankyrin-binding assays.

The ability of neurofascin to bind ankyrin was quantitated using an in vitro ankyrin-binding assay. Beads coated with protein A were used to immunoadsorb HA-tagged neurofascin and subsequently used as substrates for binding of radiolabeled ankyrin. Antibody-coated beads were prepared by coupling 1 mg protein A (Pierce, Rockford, IL) to 100 μg latex beads (epoxy-activated, 0.4-μm-diam MMA/GMA copolymer beads from Bangs Laboratories, Carmel, IN) for 72 h at 4°C with end over end mixing in a buffer containing 10 mM Hepes, pH 8.5, 1 mM EGTA, 1 M NaCl, and 1 mM NaN₃. Unreacted sites were quenched by adding Tris, pH 8.0, to 100 mM concentration and incubating overnight at 4°C. Nonspecific protein sites were then blocked by the addition of 10 mg/ml bovine serum albumin for 2 h. Neurofascin was immunoprecipitated as described above, and the immune complexes were captured with 20 μl protein A–latex beads. The amount of each expressed neurofascin isoform used for an assay was normalized after Phosphorimage scanning (Molecular Dynamics, Inc., Sunnyvale, CA) of crude neurofascin immunoblots of cell extracts for each respective transfected cell line.

The 82-kD membrane binding domain of ankyrin_B was expressed and purified as described (Davis et al., 1993). Purified 82-kD ankyrin was then radiolabeled with ¹²⁵I–Bolton-Hunter reagent (ICN Biochemicals, Irvine, NY) as previously described for ankyrin and other proteins (Davis et al., 1993; Davis and Bennett 1983, 1984). Individual neurofascin immunoprecipitates were incubated with 20 nM radiolabeled ankyrin plus a given amount of unlabeled 82-kD ankyrin in a buffer containing 10 mM sodium phosphate, pH 7.0, 100 mM NaCl, 1 mM EDTA, 2 mg/ml bovine serum albumin, 0.1% Triton X-100, and 1 mM NaN₃ for 3 h at 4°C. After incubation the beads were centrifuged over a 10% sucrose cushion for 15 min at 10,000 g to separate bound from free radioactivity, and the beads were assayed for ¹²⁵I in a γ counter. Each plot depicts one representative complete assay, whereas individual points were determined from the average of duplicate assays for two individual immunoprecipitates.

Transfected and untransfected B104 cells were cultured in high glucose DMEM containing 10% heat-inactivated fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin. All cell culture reagents were purchased from GIBCO BRL. Transfected cells were additionally cultured in 250 μg/ml neomycin (Sigma Chemical Co., St. Louis, MO). For pharmacological treatments, cells were grown to 50–70% confluency. Cells were cultured on serum-free media for 12 h before treatment with either NGF or bFGF. Immediately preceding lysis, cells were washed with ice-cold phosphate-buffered saline.

HA epitope-tagged full-length neurofascin, full cytoplasmic-truncated neurofascin, and the FIGQY to F tyrosine mutant were either untreated or treated with 100 ng/ml NGF for 30 min and then immunolabeled with the HA-specific monoclonal antibody and an FITC-tagged secondary antibody. Cells were subsequently monitored using an Odyssey confocal imaging system (Noran Inc., Middleton, WI) mounted on an inverted Nikon Diaphot microscope in combination with an imaging board (Bitflow, Woburn, MA) and eye imager calculator software (IO Industries, ON, Canada), a system that captures images every 16.7 ms. These instruments were purchased and assembled by Dr. Tobias Meyer (Duke University Medical Center, Durham, NC). FITC was excited at 488 nm and was monitored above 515 nm. FITC was photobleached using a focused UV laser at 365 nm. UV laser estimated intensity is 1 mW with an opening time of 10 ms. Fluorescence measurements were determined by the National Institutes of Health Image method of analysis. Triplicate measurements were taken from each image and averaged followed by subtraction of average background fluorescence.

Denoted neurofascin-expressing B104 cells were either untreated or treated with 100 ng/ml NGF for 90 min. Control cells were first fixed with 2% paraformaldehyde for 20 min at 4°C and then treated with a PBS blocking buffer (150 mM NaCl, 30 mM sodium phosphate, pH 7.4, 1% BSA, and 10% normal goat serum) for 30 min. Control cells were subsequently treated with 0.2% Triton X-100 in the PBS blocking solution for 10 min at room temperature and washed with a 0.2% Tween-20 solution (in the PBS blocking buffer). Alternatively, cells were first treated with 0.2% Triton X-100 in DMEM cell culture media (GIBCO BRL) for 10 min at room temperature and then fixed with 2% paraformaldehyde for 20 min at 4°C. Subsequently, cells were treated with PBS blocking buffer for 30 min and washed with a 0.2% Tween-20 solution.

After fixation, all cells were incubated overnight at 4°C with the HAspecific monoclonal antibody (1:1,000). Cells were washed in 0.2% Tween-20 buffer and then incubated with anti–rabbit FITC-conjugated secondary antibody (1:2,000) for 2 h at 4°C. Finally, cells were washed with the 0.2% Tween-20 solution. Confocal images were obtained with an LSM Zeiss confocal microscope. Fluorescence measurements (intensity/ μm²) were obtained from entire cells, and data was analyzed with a Zeiss analysis software package. Fluorescence intensities for cells treated with 0.2% Triton X-100 before paraformaldehyde fixation are presented relative to fluorescence intensities for cells that were paraformaldehyde fixed before Triton X-100 extraction.

Rat brain lysates were prepared for embryonic days 15, 17, and 20, postnatal days 1, 2, 5–8, 10, 14, and 15, as well as adult rat. Brains were sonicated in the modified RIPA buffer previously mentioned. Crude lysates (100 μg) from each day were subjected to SDS-PAGE, and subsequent immunoblots were probed with either the mucin-specific neurofascin polyclonal or an anti–Nr-CAM polyclonal anitbody. Immunoblots were scanned via a Phosphorimager to determine the amount of neurofascin and Nr-CAM expressed each day over the developmental period. Subsequently, lysates were immunoprecipitated either for normalized amounts of neurofascin or Nr-CAM, and resulting blots were probed with the anti-phosphotyrosine antibody. Alternatively, lysates were immunoprecipitated with the anti-phosphotyrosine antibody (after normalization for neurofascin or NrCAM levels), and subsequent blots were probed with either neurofascin or Nr-CAM–specific polyclonal antibodies.

Neurofascin and other members of the L1 family of nervous system cell adhesion molecules have four highly conserved tyrosine residues in their cytoplasmic domains. The possibility that one or more of these tyrosines are substrates for phosphorylation by protein kinases was evaluated using HA epitope-tagged neurofascin stably transfected into the B104 rat neuroblastoma cell line. Transfected cells were first treated with activators of receptor tyrosine kinases (NGF and bFGF) or inhibitors of protein tyrosine phosphatases (vanadate and dephostatin), after which neurofascin was immunoprecipitated using the HA-specific monoclonal antibody. Subsequent immunoblots were probed with an anti-phosphotyrosine antibody (Fig. 1,A). The protein tyrosine phosphatase inhibitor dephostatin (10 μM for 30 min) resulted in strong anti-phosphotyrosine immunoreactivity (Fig. 1,A, lane 2), as did sodium metavanadate (data not shown). Furthermore, incubation of the transfected cells with either nerve growth factor (100 ng/ml for 30 min; Fig. 1,A, lane 3) or basic fibroblast growth factor (50 ng/ml for 30 min; Fig. 1,A, lane 4) also generated a strong immunoreactive pool of tyrosine-phosphorylated neurofascin. Fig. 1 B quantitates the extent of tyrosine phosphorylation of neurofascin after treatment of transfected cells with either protein tyrosine phosphatase inhibitors (vanadate and dephostatin) or tyrosine kinase activators (NGF and bFGF) over an extended concentration range. These results demonstrate that neurofascin is a target for both protein tyrosine kinases and phosphatases in vivo when the relevant enzymes are pharmacologically perturbed in a dose-dependent manner.

The question of whether tyrosine phosphorylation modulates ankyrin-binding activity of neurofascin was evaluated by determining the effects of phosphorylation on the ability of neurofascin to coimmunoprecipitate endogenous ankyrin expressed in B104 cells (Fig. 2,A). Neurofascin immunoprecipitates were simultaneously enriched for polypeptides of 210 and 190 kD as well as a doublet at 102 and 104 kD that crossreact with a brain ankyrin-specific polyclonal antibody (ankyrin_B; Fig. 2,A, lane 2). The 210- and 190-kD crossreacting polypeptides are of the expected size for ankyrin with membrane-binding, spectrin-binding, and COOH-terminal domains. The smaller polypeptides presumably represent either alternatively spliced variants missing one or more of these domains or proteolytically cleaved forms of the 210- and 190-kD polypeptides. Activation of tyrosine kinases (Fig. 2,A, NGF or bFGF; lanes 4 and 5) or inactivation of tyrosine phosphatases (Fig. 2 A, dephostatin; lane 3) completely eliminate the ability of neurofascin to coimmunoprecipitate 210- and 190-kD ankyrin as well as the 102 and 104-kD crossreacting polypeptides.

Coimmunoprecipitation experiments provide qualitative evidence that tyrosine phosphorylation of neurofascin abolishes association with endogenous ankyrin in the context of intact cells. These results do not, however, address whether other cellular factors such as phosphotyrosinebinding proteins are required for inhibition. Effects of tyrosine phosphorylation of neurofascin on binding to ankyrin were evaluated quantitatively in isolation from other cytosolic proteins in an in vitro ankyrin-binding assay. Neurofascin was immunoprecipitated using the HA-specific monoclonal antibody and adsorbed to protein A–coated beads, extensively washed, and then incubated with radiolabeled ankyrin in a binding assay described in Materials and Methods.

Neurofascin isolated from cells treated with NGF and dephostatin exhibits an ∼40% reduction in ankyrin-binding activity compared to neurofascin isolated from untreated cells (Fig. 2,B). Scatchard plots (Fig. 2 C) demonstrate that the inhibition is due to a reduction in capacity (intercept with the X axis) with no change in affinity (slope). Tyrosine phosphorylation thus decreased the total ankyrin-binding capacity, while the ankyrin-binding affinity was unaltered. Such a result can be interpreted to indicate that tyrosine phosphorylation of a given neurofascin molecule completely abolishes its ability to bind ankyrin, but only a subpopulation of neurofascin molecules are tyrosine phosphorylated. Possible explanations for only partial phosphorylation and, thus, only a partial reduction in the ankyrin-binding capacity of the total neurofascin population include dephosphorylation during isolation, expression of neurofascin in amounts that exceed the capacity of tyrosine kinases, and/or the presence of a subpopulation of neurofascin in a cellular compartment that does not contain appropriate tyrosine kinase activity. All three of these possibilities could be operative and therefore contribute to partial phosphorylation of isolated neurofascin.

The finding that tyrosine phosphorylation abolished ankyrinbinding activity suggested the possibility that definition of the ankyrin-binding determinants of neurofascin would also lead to identification of the specific site of tyrosine phosphorylation. The ankyrin-binding site of neurofascin has been mapped previously to a 35–amino acid stretch in the COOH-terminal portion of the cytoplasmic domain encompassing two highly conserved tyrosine residues (Davis and Bennett, 1994). Deletion constructs targeted to this general location, as well as the full length and fully truncated constructs (Fig. 3,A), were evaluated for ankyrinbinding activity as determined by coimmunoprecipitation of endogenous ankyrin (Fig. 3,B) and by direct measurement of ankyrin-binding activity of isolated neurofascin (Fig. 3 C).

Neurofascin with a complete truncation of the cytoplasmic domain exhibits no detectable coimmunoprecipitation with endogenous ankyrin isoforms (Fig. 3,B, lane 4) and has no binding activity in ankyrin-binding assays (Fig. 3,C). Deletion of the COOH-terminal 28 residues up to, but not including, the FIGQ sequence had no effect on ankyrin binding, while deletion of an additional 11 residues resulted in a major loss of ankyrin-binding activity (Garver, T., and Q. Ren, unpublished observations). These results were the basis for a second generation of internally deleted constructs. Deletion of FIGQY region alone leads to loss of ability to coimmunoprecipitate ankyrin (Fig. 3,B, lane 3) and an 80% decrease in the level of in vitro ankyrin binding after immunoadsorption of HA-tagged neurofascin to protein A–coated beads (Fig. 3,C). Neurofascin with a deletion of an additional seven residues (QFNEDFIGQY) exhibited a similar 80% reduction in the ankyrinbinding activity as compared to deletion of FIGQY alone (Fig. 3,C). However, an internal deletion of 25 residues encompassing SDDSLVDY . . . FIGQY (56–81) resulted in a complete loss of ankyrin-binding activity (Fig. 3 C).

The results with deletion mutants implicated the five residues, FIGQY, as essential for full ankyrin binding, with the NH₂-terminal residues SLVDY potentially playing a secondary role, and suggested that phosphorylation of one or both of these tyrosines could be the basis for regulation of ankyrin-binding activity. The role of these tyrosines in regulation was evaluated by site-directed mutagenesis of each to phenylalanine (Fig. 3,A) followed by evaluation of tyrosine phosphorylation and ankyrin-binding activity. These constructs were stably transfected in B104 cells that were subsequently treated with NGF, bFGF, or dephostatin. The LVDY to F tyrosine mutant has a tyrosine phosphorylation pattern (Fig. 4,A) similar to that seen for the full length, native neurofascin (Fig. 1,A). However, the FIGQY to F tyrosine mutant is significantly less phosphorylated under the same pharmacological paradigm. Fig. 4 B quantitatively depicts the effects of the treatments on tyrosine phosphorylation. These results identify the FIGQY tyrosine residue as the primary site of tyrosine phosphorylation under the given treatment conditions.

Upon establishing differential phosphorylation patterns for the two tyrosine residues, we next inspected whether these two tyrosines are differentially responsible for the modulation of ankyrin binding. After treatment of the respective transfected cells, neurofascin was immunoprecipitated and used for the in vitro ankyrin-binding assays. The ankyrin-binding competence of the LVDY to F tyrosine mutated form of neurofascin is negatively regulated by tyrosine phosphorylation to a similar extent as native neurofascin, as determined by the ability to coimmunoprecipitate endogenous ankyrin (Fig. 4,C) and in ankyrin-binding assays (Fig. 4,D). However, the FIGQY to F neurofascin mutant isolated from cells treated with NGF, bFGF, or dephostatin retains full ankyrin-binding competence based on the ability to coimmunoprecipitate endogenous ankyrin (Fig. 4,C) and in ankyrin-binding assays (Fig. 4 D). Therefore, the FIGQY tyrosine residue appears to be both the primary site of tyrosine phosphorylation as well as the phosphorylated site responsible for inhibition of ankyrinbinding activity.

Consequences of ankyrin binding and tyrosine phosphorylation on the dynamic behavior of neurofascin in living cells were evaluated using the technique of FRAP. HAtagged neurofascin constructs transfected in B104 cells were labeled for fluorescence measurements using a monoclonal antibody against the HA epitope and an FITC-labeled secondary antibody. The extent and rate of recovery of fluorescence at the site of bleaching provide insight about the mobile fraction and the diffusion properties of immunolabeled neurofascin molecules.

Native neurofascin exhibits essentially no recovery of fluorescence after photobleaching, implying its involvement in highly constraining intermolecular interactions (Fig. 5,A). In contrast, neurofascin with a fully truncated cytoplasmic domain exhibits a 10–15% recovery after photobleaching (Fig. 5 A), indicating that the cytoplasmic domain of neurofascin makes a small but significant contribution to the restriction in lateral mobility. The effect of the cytoplasmic domain presumably results from connections to the spectrin skeleton through ankyrin binding. Interactions involving the ectodomain and/or membrane-spanning domain of neurofascin, however, probably participate in the majority of restriction in the lateral mobility of neurofascin.

Treatment of cells expressing full length neurofascin with NGF under conditions that promote phosphorylation of the FIGQY tyrosine releases all of the restrictions attributable to the cytoplasmic domain (Fig. 5,A). The fraction that recovers increased from 0 to ∼10–15% of original fluorescence levels. The FIGQY tyrosine is responsible for effects of NGF-promoted tyrosine phosphorylation since the FIGQY to F mutant is resistant to NGF treatment (Fig. 5 B) and exhibits the same parameters of fractional recovery of fluorescence as neurofascin in untreated cells.

Effects of tyrosine phosphorylation on the static in vivo interaction between forms of expressed neurofascin with detergent-insoluble components of the spectrin-based cytoskeleton were evaluated by analyzing the changes in rhodamine-labeled, HA epitope-tagged neurofascin intensity after Triton X-100 extraction. Detergent-extraction experiments allow one to examine the steady state interaction between neurofascin and ankyrin and the factors that may modulate this interaction. B104 cells expressing either native neurofascin, cytoplasmic domain-deleted neurofascin, or the FIGQY to F tyrosine mutant form of neurofascin were either untreated or treated with 100 ng/ml NGF for 90 min (Fig. 6, A–D). Cells (extracted with 0.2% Triton X-100 before fixation) expressing full length neurofascin exhibited a 65% retention of fluorescence, as compared to cells fixed before detergent treatment. The fluorescence was reduced to 28% after treatment with NGF. Neurofascin lacking the cytoplasmic domain was retained at only 10% in the presence of detergent, indicating that NGF abolished most, but not all, of the interactions of the cytoplasmic domain with detergent-insoluble proteins. Interestingly, the reduction in fluorescence due to NGF-induced tyrosine phosphorylation of native neurofascin is quantitatively similar to the reduction in the in vitro binding capacity of neurofascin after tyrosine phosphorylation (Fig. 2 B). In contrast to native neurofascin, cells expressing the FIGQY to F tyrosine mutant neurofascin were unaffected by NGF treatment and retained ∼65% of the label in the presence and absence of NGF, once again indicating that this site is the primary one responsible for phosphotyrosine-dependent regulation of neurofascin–ankyrin interactions.

We wished to determine whether neurofascin and the related molecule Nr-CAM are tyrosine phosphorylated in vivo to establish a physiological correlation for what we have observed in vitro. Consequently, we looked at each molecule during neuronal development in rat to see if either or both of these molecules exhibit a period of phosphotyrosine immunoreactivity. Fig. 7, A (neurofascin) and B (Nr-CAM) demonstrate that both of these molecules are tyrosine phosphorylated in vivo in a time-dependent fashion, with the maximal level of phosphotyrosine immunoreactivity present during the embyronic period. Phosphorimage scanning (data not shown) indicates there is nearly a threefold reduction in Nr-CAM tyrosine phosphorylation from embryonic day 15 to adult brain. In comparison, neurofascin phosphotyrosine immunoreactivity is intense at embryonic day 15. However, this signal is dramatically reduced (an ∼20-fold decline from embryonic day 15 to adult) over the denoted developmental period. With respect to neurofascin, it is interesting to note that the 186-kD isoform is the preferential splice form to be tyrosine phosphorylated, even when minimally expressed, as in embryonic day 15. Immunoprecipitation first with the anti-phosphotyrosine antibody followed by immunoblotting with either neurofascin or Nr-CAM antibodies rendered qualitatively similar results (data not shown).

This paper presents evidence that a member of the L1 family of cell adhesion molecules is a substrate for protein tyrosine kinase(s) and phosphatase(s), identifies the highly conserved FIGQY tyrosine in the cytoplasmic domain as the site of phosphorylation, and demonstrates that phosphorylation of the FIGQY tyrosine abolishes ankyrinbinding activity. Phosphorylation of the FIGQY tyrosine increases lateral mobility of neurofascin expressed in neuroblastoma cells to the same extent as removal of the cytoplasmic domain and increases detergent extractability of neurofascin. Ankyrin binding thus modulates the dynamic behavior of neurofascin and is the target for regulation by tyrosine phosphorylation–dephosphorylation in response to external signals. We have further demonstrated that two L1 family members, neurofascin and Nr-CAM, are both tyrosine phosphorylated in vivo during embryonic development, indicating that the functions attributed to tyrosine phosphorylation, as determined from our in vitro studies, are likely to be relevant in vivo as well.

The FIGQY sequence is present in cytoplasmic domains of all members of the L1 family of nervous system cell adhesion molecules, including a recently described Caenorhabditis elegans partial cDNA sequence. Moreover, ankyrinbinding activity has been directly demonstrated for mammalian neurofascin, L1 and Nr-CAM (Davis and Bennett, 1994), as well as Drosophila neuroglian (Dubreuil et al., 1996). The findings of this study, therefore, may reflect a highly conserved mechanism, used by the entire class of L1-related cell adhesion molecules, of ankyrin-dependent connections to the spectrin skeleton regulated via site-specific tyrosine phosphorylation.

The interaction of neurofascin with ankyrin is the first example of an ankyrin–membrane protein association that is reversible and subject to dynamic control. The potential for localized activation or inhibition of cell adhesion molecule interactions with ankyrin could be important for processes such as neuronal development, synaptic plasticity, and localized assembly of cell adhesion molecule–based structures in axons. The observation that Nr-CAM and, more dramatically, neurofascin are preferentially tyrosine phosphorylated during fetal development indicates phosphorylation may play a key role in modulating cytoskeletal and membrane dynamics during assembly of the nervous system. Neurofascin and Nr-CAM are targeted to nodes of Ranvier in myelinated axons where the voltage-dependent sodium channel and an isoform of ankyrin are highly concentrated (Davis et al., 1996). Neurofascin and Nr-CAM are detected early during the development of nodes of Ranvier and may direct the subsequent targeting of ankyrin and the voltage-dependent sodium channel to this site (Lambert, S., J. Davis, P. Michael, and V. Bennett. 1995. Mol. Biol. Cell. 6:98a). Dephosphorylation by a protein tyrosine phosphatase localized at nodes of Ranvier is an attractive mechanism for localized activation of ankyrin-binding activity of neurofascin and Nr-CAM. The biological significance of a reversible ankyrin-binding activity for L1like cell adhesion molecules could be evaluated using FIGQY to F mutants, which are resistant to inhibition of ankyrin binding by tyrosine phosphorylation.

Consequences of ankyrin association with neurofascin include the potential for stabilizing laterally associated neurofascin oligomers in addition to providing a connection to spectrin. The membrane-binding domain of ankyrin contains two independent binding sites for neurofascin, and, in principle, could promote assembly of neurofascin dimers (Michaely and Bennett, 1995). Transcellular interactions of neurofascin, by analogy with cadherins, could be promoted by lateral association of neurofascin molecules. These considerations suggest the testable hypotheses that ankyrin binding could enhance cell–cell interactions mediated by neurofascin and other L1 family members and that elevation of tyrosine phosphorylation would reverse activation of cell adhesion by ankyrin.

A question remains how the phosphorylation event prohibits neurofascin–ankyrin association. The fact that the key tyrosine residue is directly adjacent to the primary amino acid sequence established in this study as required for ankyrin binding may mean that tyrosine phosphorylation of neurofascin blocks ankyrin binding directly through steric hinderance or by the introduction of electrostatic repulsion. Another possibility is that ankyrin associates preferentially with neurofascin dimers and that phosphorylation inhibits neurofascin dimerization. These issues could be addressed by characterizing the stoichiometries of neurofascin–ankyrin complexes in solution.

A possible consequence of tyrosine phosphorylation of neurofascin, in addition to inhibition of ankyrin binding, is the activation of a phosphotyrosine-specific binding protein. However, consensus sequences for SH2 and PTB proteins differ from the neurofascin sequence flanking the FIGQY region. Putative phosphotyrosine-binding proteins specific for phosphorylated members of the L1 family, therefore, may be distinct from currently established candidates.

Previous studies (Williams et al., 1994a,b,c,d; Doherty et al., 1995; Dubreuil et al., 1996) have recognized that L1 family members can directly act as signal transducers. We are providing evidence of an L1 family member as a substrate for enzyme(s) in tyrosine signaling cascade(s). Therefore, it will be important in future work to elucidate the specific tyrosine protein kinases and phosphatases that control the phosphorylation state of neurofascin. Both NGF and bFGF promoted phosphorylation of neurofascin, suggesting either neurofascin is recognized by at least two distinct receptor protein kinases or that other protein kinase(s) activated downstream are involved. Another point of interest includes determining which protein tyrosine phosphatase(s) are involved in dephosphorylating neurofascin. Recent work has focused on a class of receptor protein tyrosine phosphatases (for review see Brady-Kalnay et al., 1995). Some integral protein tyrosine phosphatases (LAR-PTP, for example) are homologous to cell adhesion molecules of the Ig superfamily. Interestingly, Krypta et al. (1996) have shown that a direct association between LARPTP and cadherins helps to control cadherin adhesive functions. Although the cadherins are distinct from the L1 family, this direct association may represent a common mechanism by which cell adhesion molecules, through lateral association with a structurally related receptor protein tyrosine phosphatase, may be dephosphorylated.

Modulation of the interaction between ankyrin and the various members of L1 cell adhesion molecules may play a critical role in mediating reversible assembly of a variety of transcellular complexes in the developing and adult nervous systems. The next step will be to extend the findings of this study to more physiologically relevant systems. Both tyrosine phosphorylation-resistant neurofascin mutants and forms of neurofascin lacking ankyrin-binding activity should provide precisely defined molecular tools for future work.