Heterophilic interactions of sodium channel beta1 subunits with axonal and glial cell adhesion molecules.

Voltage-gated sodium channels localize at high density in axon initial segments and nodes of Ranvier in myelinated axons. Sodium channels consist of a pore-forming alpha subunit and at least one beta subunit. beta1 is a member of the immunoglobulin superfamily of cell adhesion molecules and interacts homophilically and heterophilically with contactin and Nf186. In this study, we characterized beta1 interactions with contactin and Nf186 in greater detail and investigated interactions of beta1 with NrCAM, Nf155, and sodium channel beta2 and beta3 subunits. Using Fc fusion proteins and immunocytochemical techniques, we show that beta1 interacts with the fibronectin-like domains of contactin. beta1 also interacts with NrCAM, Nf155, sodium channel beta2, and Nf186 but not with sodium channel beta3. The interaction of the extracellular domains of beta1 and beta2 requires the region 169TEEEGKTDGEGNA181 located in the intracellular domain of beta2. Interaction of beta1 with Nf186 results in increased Nav).2 cell surface density over alpha alone, similar to that shown previously for contactin and beta2. We propose that beta1 is the critical communication link between sodium channels, nodal cell adhesion molecules, and ankyrinG.

The axo-glial complex in myelinated axons is composed of a nodal gap region that contains high density clusters of voltagegated sodium channels and a juxtaparanodal region that contains voltage-gated potassium channels (1)(2)(3)(4)(5)(6). The nodal gap and the juxtaparanode are separated by the paranode, a region containing septate-like junctions composed of cell adhesion molecules (CAMs) 1 that act as diffusion barriers to ion channel movement (2, 6 -8). This specific arrangement of voltage-gated sodium and potassium channels results in rapid and efficient saltatory conduction of action potentials (9 -12).
Sodium channel ␤1 and ␤2 subunits are also members of the Ig superfamily of CAMs. ␤1 and ␤2 colocalize with sodium channel ␣ subunits at nodes of Ranvier, and ␤1 interacts with contactin and with Nf186 in vitro (9,17,19,20). Nf186, Nr-CAM, ␤1, and ␤2 each interact in vitro with a key cytoskeletal anchoring protein, ankyrin G , that is also localized to nodes of Ranvier (21)(22)(23). ␤1or ␤2-mediated homophilic cell-adhesive interactions result in ankyrin recruitment in Drosophila S2 cells, and the interaction of sodium channel ␣ subunits with ankyrin G is greatly enhanced in the presence of ␤1 subunits in vitro (22,24). We propose that the sodium channel signaling complex at nodes of Ranvier includes the CAMs contactin, NrCAM, and Nf186 as well as ankyrin G and that sodium channels interact with these proteins via ␤1 subunits.
In sciatic nerve, Nf186 and NrCAM are early markers of nodal formation that cluster at presumptive nodes prior to sodium channels and ankyrin G (2,25). In the peripheral nervous system (PNS), Schwann cell contact at the beginning stages of myelin ensheathement is required for sodium channel clustering (26,27). In optic nerve, developing nodes of Ranvier are defined by clustering of ankyin G prior to the arrival of CAMs and sodium channels (28,29), and this process is independent of paranodal axoglial cell adhesion (29). In the central nervous system (CNS), conditioned medium from optic nerve glia induces clustering of sodium channels in retinal ganglion cell axons in vitro (30). In vivo, the presence of oligodendrocytes is necessary for sodium channel clustering and nodal formation in the CNS (30).
Contactin is expressed at paranodes in the PNS, where it interacts with Caspr on the axonal membrane and neurofascin 155 (Nf155) on the Schwann cell membrane to form septatelike junctions that separate the nodal gap from the juxtaparanode (31)(32)(33). In the CNS, contactin is expressed in the nodal gap, where it interacts with sodium channels (17,32), as well as in the paranode. Contactin increases the level of sodium channel expression at the cell surface from 4-to 6-fold in vitro (17), and this interaction is dependent on presence of ␤1 subunits (9,17). In ␤1 (Ϫ/Ϫ) mice, contactin and sodium channels co-localize at optic nerve nodes of Ranvier but do not associate, confirming that interaction of these two molecules is dependent on the presence of ␤1 (9). The number of mature nodes of Ranvier is significantly reduced in ␤1 (Ϫ/Ϫ) optic nerves, and we have proposed that the loss of sodium channel-contactin interactions at the node may be responsible for this effect (9).
The purpose of the present study was to test the hypothesis that sodium channel ␤1 subunits participate in heterophilic cell-adhesive interactions with CAMs found in the axo-glial complex. We show that ␤1, expressed in the absence of sodium channel ␣ subunits, interacts with NrCAM, Nf186, contactin, and Nf155. We observed an interaction of the ␤1 and ␤2 subunit extracellular domains only in the presence of a region of the ␤2 intracellular domain, suggesting that cytoskeletal interactions may affect ␤2-mediated heterophilic cell adhesion. We did not detect ␤1 interactions with sodium channel ␤3 subunits in our system. ␤1-Nf186 interactions result in increased sodium channel ␣ subunit cell surface expression in vitro, similar to what we showed previously for contactin and ␤2. In contrast, although ␤1 is required for NrCAM interactions with Na v 1.2, coexpression of ␤1 and NrCAM did not increase ␣ subunit cell surface expression. These results suggest that ␤1 subunits interact with multiple CAMs to form the basis of the sodium channel complex at nodes of Ranvier. In addition, the present results showing ␤1 interaction with Nf155, coupled with our previous observation that ␤1 (Ϫ/Ϫ) mice lack septate-like junctions at the periphery of the nodal gap in a subset of CNS and PNS axons (9), support our hypothesis that ␤1 subunits may also participate in axo-glial communication at the nodal-paranodal boundary.
Generation of ␤ Subunit Fc Fusion Proteins-␤1, ␤2, or ␤3 cDNAs (24, 38 -40) encoding each extracellular domain up to but not including the transmembrane segment were amplified by polymerase chain reaction using Taq polymerase (Invitrogen) and oligonucleotide primers that contained appropriate restriction sites to facilitate directional cloning into the polylinker region of pIG1 (provided by Dr. Genviève Rougon). A splice donor site was incorporated at the 3Ј end of the PCR product as previously described (37,41). Briefly, the 3Ј end of each ␤ subunit cDNA fragment was defined by nucleotides A and G, which represent the first half of the introduced splice donor site (AG/GT). The amplified region with the splice donor site was then fused to the genomic sequence of the Fc region (hinge, CH2, and CH3) of the human IgG1, which bears a 3Ј acceptor site into the pIG1 vector. The PCR fragment and the intron were sequenced and were determined to be free of mutations. Following transfection, the resulting pre-mRNAs were spliced by the cells to yield mature mRNAs that produce chimeric molecules containing the Ig-like domains of the ␤ subunits and the hinge, CH2, and CH3 regions of the human IgG1. The resulting chimeras are called ␤1-Fc, ␤2-Fc, and ␤3-Fc, respectively.
Binding and Immunofluorescence-Fc fusion protein binding assays were performed as previously described (35,36). 1610 Chinese hamster lung fibroblasts (CHL) (CRL-1657; American Type Culture Collection, Manassas, VA) grown in 8-well chamber slides (BD Biosciences) were transiently transfected with 0.5 g of a cDNA encoding a particular CAM, as indicated in the figure legends. Independent CHL cultures were transfected in parallel with 1 g of an Fc fusion cDNA, as indicated in the figure legends. 24 -36 h post-transfection, conditioned medium containing a 0.5-1 g/ml concentration of the Fc fusion protein was incubated overnight with the now confluent 8-well chamber slides containing CHL cells transfected with the CAM of interest. The next morning, cells were washed with PBS and fixed with 4% paraformaldehyde. The cells were washed twice with PBS, followed by blocking with PBS containing 1% glycine and 10% goat serum (PBGG) for 30 min at room temperature as previously described (35). Primary antibodies were diluted in PBGG and incubated for 2 h at room temperature. Fc fusion proteins were detected using a Cy3-conjugated anti-human IgG Fc fragment-specific antibody (Jackson ImmunoResearch Laboratories). Cells were washed three times with PBS and incubated with secondary antibodies diluted in PBGG for 45 min at room temperature. Cells were washed once in PBS and mounted using Pro Long® anti-fade kit (Molecular Probes). Immunofluorescence was detected and analyzed using an Olympus BX51 inverted light microscope using a 40ϫ UPla-nApo objective lens (Olympus, Melville, NY) located in the Microscopy and Image Analysis Laboratory at the University of Michigan. Images were captured using an Olympus DP70 camera and processed using DP Controller version 1.1.1.65 and DP Manager version 1.1.1.71 software.
[ 3 H]Saxitoxin (STX) Binding Analysis-Whole cell [ 3 H]STX saturation binding analysis of each cell line was performed using a vacuum filtration method as previously described (42) at a concentration of 5 nM [ 3 H]STX with the addition of 10 M unlabeled tetrodotoxin (Calbiochem) to assess nonspecific binding. [ 3 H]STX (28 Ci/mmol) was obtained from Amersham Biosciences. Binding data were normalized to protein concentrations using Advanced Protein Assay Reagent (Cytoskeleton, Inc., Denver, CO).
For each binding experiment, CHL cells were transiently transfected with the indicated CAM as described in the figure legends. To investigate binding interactions, cells were incubated with conditioned medium containing a particular Fc fusion protein. 36 h post-transfection and 12 h postincubation, the cells were fixed and stained for the expressed CAM using specific antibodies and for the Fc fusion protein of interest using a Cy3conjugated anti-human IgG antibody. None of the Fc fusion proteins showed binding to untransfected CHL cells above background (Fig. 1C). CHL cells transfected with ␤1 only did not bind the Fc protein backbone (Fig. 1C, c). Thus, any interaction that was observed between CHL cells transfected with ␤1 and incubated with the Fc fusion proteins was interpreted as a specific interaction between ␤1 and the Fc fusion protein.
␤1 Interacts with the Fibronectin-like Domains of Contactin-We have shown previously that contactin is part of the sodium channel complex at CNS nodes of Ranvier and may be involved in nodal formation during development and remyelination (17). The association of contactin with sodium channels in vitro is dependent on the presence of ␤1, and this interaction results in increased sodium channel ␣ subunit expression at the cell surface (17). The interaction between ␤1 and contactin requires the Ig loop domain but not the juxtamembrane, transmembrane, or intracellular domains of ␤1 (24). We further investigated this interaction by examining the regions of contactin that are important for interaction with ␤1. ␤1 interacts with Cn-Fc, a fusion protein containing full-length contactin ( Fig. 2A). These results are similar to the interaction previ- Conditioned medium from CHL cells transiently transfected with the following fusion constructs was incubated with untransfected CHL cells as follows. a, pIG1; b, Cn-Fc; c, ␤1 and pIG1; d, Nr-Fc; e, ␤1-Fc; f, ␤2-Fc; g, Nf155-Fc; h, ␤3-Fc. Cells were stained with Cy3-conjugated anti-human Fc (red, to detect Fc) and 4Ј,6-diamidino-2-phenylindole (blue, insets a, b, d, f, g, and h) or rabbit anti-␤1 subunit antibodies (green, inset c). A differential interference contrast image of CHL cells is shown in inset e. Scale bar, 50 m.
ously described for contactin with Nf155 and confirmed in Fig.  2C (36). In contrast, a Fc fusion construct consisting only of the six Ig domains of contactin and not the fibronectin-like domains, Cn Ig -Fc, did not bind ␤1 (Fig. 2B) but did interact with NrCAM as previously described (45) (Fig. 2D). These results suggest that the fibronectin-like repeats of contactin are required for its interaction with ␤1.
␤1 Associates with Nf186, Nf155, and NrCAM-We next investigated whether the neurofascin isoforms interact with ␤1. Previous studies have shown that Nf186 is important for determining node of Ranvier location (2,18). Ratcliffe et al. (20) showed that the extracellular domain of ␤1 interacts with Nf186 in a heterologous system using coimmunoprecipitation techniques. As shown in Fig. 3A, ␤1-Fc incubated with CHL cells that were transfected with full-length NF186 showed fluorescent binding, supporting the previous evidence that ␤1 and NF186 interact in this configuration (Fig. 3A).
The 155-kDa form of neurofascin is expressed in oligodendrocytes and is important for the formation of axo-glial septatelike junctions separating nodes of Ranvier from the juxtaparanodes (6 -8, 33). Axo-glial septate-like junctions are formed through interactions between contactin and Caspr on the axonal membrane (6 -8, 33). Nf155 may be the glial receptor for the contactin-Caspr complex, although recent evidence has shown that Caspr may negatively regulate interactions between contactin and Nf155 (36). Our next step was to test for ␤1 interactions with Nf155. We were interested in investigating interactions between ␤1 and Nf155, as ␤1 (Ϫ/Ϫ) mice show disrupted paranodal loops. As shown in Fig. 3B, the extracellular domain of ␤1 interacts with the extracellular domain of Nf155 (Nf155-Fc), suggesting that ␤1 and Nf155 may interact in a trans manner between adjacent cells. As a positive control, and consistent with previous data, we show that Nf155-Fc interacts with contactin ( Fig. 3C) (36).
We next investigated whether the nodal protein NrCAM could interact with ␤1. When ␤1-transfected cells were incubated with Nr-Fc, we detected only a weak interaction between these two proteins (Fig. 4A). Using the opposite configuration, we detected a strong signal when NrCAM-transfected CHL cells were incubated with ␤1-Fc (Fig. 4B). Our results suggest that the NrCAM intracellular domain may influence its ability to participate in extracellular cell-adhesive interactions with ␤1.
As a positive control, CHL cells transfected with full-length ␤1 were incubated with ␤1-Fc. We observed ␤1-Fc binding in this assay (Fig. 5A), consistent with our previous results (22). In contrast, we could not detect an interaction between ␤1and ␤3-Fc in parallel experiments (Fig. 5B), suggesting that ␤1 and ␤3 do not interact. To perform the reverse experiment, CHL cells were transfected with full-length ␤3 and then incubated with ␤1-Fc (Fig. 5C). Again, no interaction between ␤1 and ␤3 was observed in this configuration. Ratcliffe et al. (20) showed previously that ␤3 interacts heterophilically with Nf186. Consequently, as a positive control, CHL cells were transfected with Nf186 and incubated with ␤3-Fc. Under these conditions, ␤3-Fc interacted with Nf186 (Fig. 5D). Thus, our ␤1-Fc and ␤3-Fc constructs were functional in terms of binding interactions, but they did not interact with each other in our system.
When CHL cells were transfected with full-length ␤1 and incubated with ␤2-Fc, we observed no interaction between these two sodium channel subunits (Fig. 6B). As a positive control, we showed that full-length ␤2 binds ␤2-Fc (Fig. 6C), consistent with our previous results showing ␤2-␤2 interactions in Drosophila S2 cells (22). Interestingly, performing the reverse experiment, cells transfected with full-length ␤2 bound ␤1-Fc (Fig. 6D), suggesting that the ␤2 intracellular domain may play a role in the interaction of the ␤2 extracellular domain with the extracellular domain of ␤1. To test this hypothesis, CHL cells were transfected with ␤2 STOP , a protein lacking the intracellular domain of ␤2 (22). ␤2 STOP -transfected CHL cells incubated with ␤1-Fc showed no binding above background (Fig. 6E), suggesting that the intracellular domain of ␤2 is required for interaction with the ␤1 extracellular domain. Alternatively, the inability of ␤2 STOP to interact with ␤1-Fc may be due to a conformational change in the ␤2 protein due to the lack of an intracellular domain. This possibility is unlikely, however, since ␤2 STOP expression in S2 cells results in cellular aggregation (22), indicating that this construct is able to function as a homophilic CAM (Fig. 6F). Our results suggested that ␤2-mediated inside-out signaling and/or ␤2-cytoskeletal interactions may be required for ␤2-mediated heterophilic cell-adhesive interactions with ␤1.
To further investigate the role of the ␤2 intracellular domain in the extracellular interaction of ␤2 with ␤1, we created a series of ␤2 truncation mutants in which increasing numbers of amino acids were deleted from the intracellular domain (Fig.  6A). These truncation mutants are termed ␤2⌬ 181-186 , ␤2⌬ 174 -186 , ␤2⌬ 169 -186 , and ␤2⌬ 164 -186 , respectively. When ␤2⌬ 181-186 was transfected into CHL cells and incubated with ␤1-Fc, it interacted with ␤1-Fc to a similar extent as wild type ␤2 (Fig. 6G). ␤2⌬ 174 -186 , however, showed an impaired ability to interact with ␤1-Fc, since only a small portion of the transfected cells exhibited positive binding for ␤1-Fc (Fig. 6, H and  I). Further deletion of the ␤2 intracellular domain, ␤2⌬ 169 -186 , and ␤2⌬ 164 -186 resulted in the abolishment of ␤2 binding to ␤1-Fc despite robust expression of these mutant proteins at the plasma membrane, indicating that the region 169 TEEEGKT-DGEGNA 181 in the intracellular domain of ␤2 is critical for interaction of the ␤2 extracellular domain with ␤1. ␤1, Contactin, and Nf186, but Not Nf155 or NrCAM, Increase Cell Surface Sodium Channel Expression-␤1 increases Na v 1.2 cell surface expression in transfected cells by ϳ2-fold over that observed with ␣ alone (17,42). Coexpression of ␤2 or ␤3 with Na v 1.2 in the absence of ␤1 resulted in no significant increases in sodium channel cell surface density over ␣ alone, whereas the presence of both contactin and ␤1 increased Na v 1.2 at the cell surface by 6-fold (17,24). We next investigated whether Nf186, Nf155, or NrCAM affected Na v 1.2 cell surface expression levels. Stable CHL cell lines expressing Na v 1.2 in combination with ␤1, Nf186, Nf155, and/or NrCAM were generated by stably transfecting Nf186, Nf155, or NrCAM into the previously established Na v 1.2 or Na v 1.2/␤1 stable cell lines (17,24,42). Two independent stable clones showing robust expression of each respective CAM by Western blot were selected for the binding studies (Fig. 7A). Cell surface sodium channel expres- sion was measured using whole cell [ 3 H]STX binding as previously described (24). When expressed with Na v 1.2 in the absence of ␤1, neither Nf186, Nf155, nor NrCAM increased Na v 1.2 cell surface levels over ␣ alone (Fig. 7B). In fact, these CAMs produced statistically significant decreases in [ 3 H]STX binding compared with ␣ alone. We have observed similar results for ␤2 and contactin (17) but do not as yet have an explanation for these effects. In the presence of ␤1, neither Nf155 nor NrCAM increased cell surface Na v 1.2 expression over that observed with the Na v 1.2/␤1 cell line (Fig. 7B). In contrast, Nf186 did increase Na v 1.2 density at the cell surface by ϳ2-fold in the presence of ␤1 compared with Na v 1.2/␤1. We next investigated whether NrCAM or Nf186 might increase the level of sodium channel cell surface expression observed in the presence of ␤1 and contactin. The cell lines Na v 1.2/␤1/Cn/ Nf186 and Na v 1.2/␤1/Cn/NrCAM were generated from the previously established Na v 1.2/␤1/Cn line (17). Neither Nf186 nor NrCAM increased sodium channel cell surface expression over that observed in the presence of ␤1 and contactin (Fig. 7B), suggesting that the level of Na v 1.2 expressed at the cell surface in the presence of ␤1 and contactin may be the maximal cell surface density of channels attainable in this heterologous system. [ 3 H]STX binding to Na v 1.2/␤1/Cn/Nf186 was actually decreased from that observed for Na v 1.2/␤1/Cn; however, we suspect that this could be due to variations in individual cells in the original Na v 1.2/␤1/Cn cell line.
We propose that ␤1 plays a key role in the interaction of sodium channels with CAMs at nodes of Ranvier. Our previous results showed that contactin and sodium channels do not interact in the absence of ␤1 in vitro or in vivo (9,17). Ratcliffe et al. showed that the interaction of Nf186 with the sodium channel complex is dependent on the presence of the ␤1 subunit in transfected cells (20). We were next interested in determining whether ␤1 is required for the interaction of NrCAM with Na v 1.2. Lysates from the Na v 1.2/NrCAM and Na v 1.2/␤1/Nr-CAM cell lines used in the binding studies were immunoprecipitated with either nonimmune IgG or the anti-Na v 1.2 antibody, Sp 11 II. Immunoblotting with anti-NrCAM antibodies showed that Na v 1.2 and NrCAM associate in the Na v 1.2/␤1/ NrCAM cell line but not in the absence of ␤1 (Fig. 7C, lane 3). The control IgG and Na v 1.2/NrCAM lanes (Fig. 7C, lanes 1 and   2) were negative for NrCAM staining despite similar NrCAM expression levels in the two cell lines (Fig. 7C, lanes 4 and 5). This result indicates that, similar to contactin and Nf186, ␤1 is required for the interaction of NrCAM with the sodium channel ␣ subunit. DISCUSSION Sodium channel ␤1 subunits are multifunctional proteins that are involved in channel modulation as well as cell adhesion (17,(22)(23)(24)42). ␤1 subunits interact homophilically in vitro, resulting in cellular aggregation and ankyrin recruitment to points of cell-cell contact, as well as heterophilically in vitro with contactin and Nf186 (17,20,22,24). The present data identify the CAMs Nf155, NrCAM, and sodium channel ␤2 subunits as additional heterophilic binding partners of ␤1 in vitro. In contrast, sodium channel ␤3 subunits do not appear to interact with ␤1. Nf155 and NrCAM are important for formation and maintenance of both CNS and PNS nodes of Ranvier (2,3,6). ␤2 is critical for sodium channel cell surface expression in vitro and in vivo (19,46,47). We show that contactin, Nf186 and NrCAM increase channel cell surface expression in the presence but not in the absence of ␤1, similar to that shown previously for contactin and ␤2 (17,24). Thus, we propose that ␤1 functions as the critical communication link between sodium channels, ankyrin G , and other CAMs at the node of Ranvier and may participate in paranodal septate-like junction formation via interaction with Nf155.
We have previously described the interaction between sodium channel ␤1 subunits and the CAM contactin (17,24). This interaction depends on the presence of the ␤1 Ig loop domain and results in a 4 -6-fold increase in Na v 1.2 expression at the cell surface in vitro (24). In the present study, we show that the fibronectin-like repeats of contactin, but not the Ig loop domains, are essential for ␤1 interactions. In ␤1 (Ϫ/Ϫ) mice, the association of sodium channels with contactin is disrupted, confirming the interpretation of our in vitro results that ␤1 subunits are required for formation of the sodium channelcontactin complex (9). We have proposed that the loss of sodium channel-contactin interactions in vivo results in destabilization of sodium channel cell surface expression at nodes and the subsequent decrease in the number of mature CNS nodes of Ranvier observed in ␤1 (Ϫ/Ϫ) mice (9). Our present results suggest that the "horseshoe" configuration of contactin (48 -50) is not required for its interaction with ␤1, leaving the Ig domains of contactin free to interact with other CAMs while simultaneously interacting with ␤1. This proposed arrangement would thus increase the number and diversity of CAMs that can be involved in the sodium channel complex.
Recent evidence has shown a role for NrCAM and Nf186 in delineating presumptive PNS nodes of Ranvier prior to sodium channel and ankyrin G clustering (2,18). NrCAM (Ϫ/Ϫ) mice exhibit a delay in nodal development as well as a delay in the localization of ankyrin G to PNS nodal areas (2). In CNS myelinated axons, ankyrin G has been shown to define nodal regions prior to CAM and sodium channel clustering (28,29). Ankyrin G (Ϫ/Ϫ) mice exhibit disrupted localization of sodium channels, neurofascin, and NrCAM at initial segments of unmyelinated cerebellar granule cell axons and myelinated Purkinje neurons (28,51). Thus, coordinated temporal and spatial adhesive events between NrCAM, Nf186, sodium channels, and ankyrin G are critical to sodium channel clustering and may exhibit different mechanisms in the PNS compared with the CNS. In the present study, we show that sodium channel ␤1 subunits interact with the nodal, ankyrin-binding CAMs NrCAM, Nf186, and ␤2, in addition to contactin. Furthermore, we show that, like contactin and Nf186, the interaction of NrCAM with the sodium channel complex is dependent on ␤1. The interac- FIG. 7. Coexpression of ␤1 and Nf186 increases while coexpression of ␤1 with Nf155 or NrCAM has no effect on Na v 1.2 cell surface density. A, Western blot analysis of Na v 1.2 or Na v 1.2/␤1 or Na v 1.2/␤1/Cn cell lines that were transfected with Nf155, Nf186, or NrCAM as indicated. Immunblots were probed with anti-Nf155 (1: 2000), anti-HA.11 (to detect Nf186, 1:1000), and anti-NrCAM (1:200), respectively. B, [ 3 H]STX binding to intact CHL cells expressing either Na v 1.2 alone or Na v 1.2 with a combination of contactin, Nf155, Nf186, NrCAM, and/or ␤1. The error bars indicate means Ϯ S.E. for Na v 1.2 (n ϭ 6), Na v 1.2/Nf155 (n ϭ 3), Na v 1.2/Nf186 (n ϭ 3), Na v 1.2/NrCAM (n ϭ 3), Na v 1.2/␤1 (n ϭ 6), Na v 1.2/␤1/Nf155 (n ϭ 3), Na v 1.2/␤1/Nf186 (n ϭ 6), Na v 1.2/␤1/NrCAM (n ϭ 3), Na v 1.2/␤1/Cn (n ϭ 5), Na v 1.2/␤1/ Cn/Nf186 (n ϭ 3), and Nav1.2/␤1/Cn/NrCAM (n ϭ 3) cells. For all cell lines, binding was performed on at least two different stable clones. 10 M tetrodotoxin was used to determine nonspecific binding as previously described (8). [ 3 H]STX binding was normalized to protein using Advanced protein assay reagent (Cytoskeleton, Inc., Denver, CO). *, statistically significant compared with Na v 1.2 (p Ͻ 0.05). **, statistically significant compared with Na v 1.2/␤1 (p Ͻ 0.05). ***, statistically significant compared with Na v 1.2/␤1/Cn (p Ͻ 0.05). C, immunoprecipitation (IP) of Na v 1.2 with NrCAM from stably expressing Na v 1.2/NrCAM or Na v 1.2/␤1/ NrCAM CHL cell lines. Equal amounts of cell lysates were immunoprecipitated with 5 l of either nonimmune IgG or Sp 11 II for Na v 1.2. Immunoblots were probed with anti-NrCAM antibodies (1:200). Cell lysates indicate equal expression of NrCAM in both cell lines (lanes 4 and 5). tion of ␤1 with contactin, with Nf186, or with ␤2 (17) results in significant increases in Na v 1.2 cell surface expression in vitro. We showed previously that Na v 1.2 channels interact weakly with ankyrin G in CHL cells, and coexpression of ␤1 greatly enhances this interaction (24). In the present study, interaction of ␤1 with the ankyrin-binding CAMs NrCAM, Nf186, and ␤2 could result in increased interaction of sodium channels with the endogenous ankyrin B protein in CHL cells (23), stabilizing sodium channels at the cell surface. We propose that ␤1 may act to stabilize sodium channels at the nodal plasma membrane in vivo through multiple extracellular interactions with ankyrin-binding CAMs as well as through its own intracellular ankyrin-binding domain. Furthermore, we propose that the reduction in action potential conduction velocity, reduction in the integral of the optic nerve compound action potential, loss of sodium channel cell surface expression, and reduction in the number of mature nodes of Ranvier observed in ␤1 (Ϫ/Ϫ) and/or ␤2 (Ϫ/Ϫ) mice (9,19) result from the loss of these critical interactions between nodal CAMs and sodium channels.
Our results show that sodium channel ␤1 and ␤2 subunits interact through heterophilic cell-adhesive interactions in addition to the interaction of each of these proteins with the pore-forming ␣ subunit (10,52). Our observation that a region of the ␤2 intracellular domain is required for heterophilic cell adhesion between the ␤2 and ␤1 extracellular domains raises some interesting possibilities. ␤2 subunits participate in homophilic cell adhesion, resulting in the recruitment of ankyrin to points of cell-cell contact in Drosophila S2 cells (22). ␤2 STOP , a ␤2 truncation mutant lacking the intracellular domain, maintains homophilic adhesion yet does not recruit ankyrin (22). In the present study, when full-length ␤1 was incubated with soluble ␤2-Fc, no interaction was detected between these two subunits. However, when full-length ␤2 was incubated with soluble ␤1-Fc, interactions between the extracellular domains of ␤1 and ␤2 were observed. These results suggested that a region in the intracellular domain of ␤2 must be present for heterophilic interaction with the extracellular domain of ␤1 to occur. Using a series of ␤2 truncation mutants, we found that the intracellular 169 TEEEGKTDGEGNA 181 region of ␤2 is critical for extracellular, heterophilic adhesion with ␤1. We propose that an interaction in the ␤2 intracellular domain, possibly with ankyrin, stabilizes extracellular, heterophilic cell adhesive events. We showed previously that a single residue in the ␤1 intracellular domain, Tyr 181 , is critical for ankyrin binding and that phosphorylation of this residue negatively regulates ankyrin interactions (22). Whereas we have not yet defined the ankyrin-binding domain of ␤2, the region that we have identified as critical for ␤2 heterophilic adhesion is located in a region of the protein similar in position to the ankyrin-binding domain of ␤1. The ␤2 intracellular domain does not contain tyrosine residues and is not similar to other ankyrin-binding domains reported in the literature (22). However, it does contain three possible casein kinase II phosphorylation sites according to the PhosphoBase program (available on the World Wide Web at www.cbs.dtu.dk/databases/Phospho-Base/) provided by the Technical University of Denmark. These putative sites include QKLS 163 TDD, DLKT 169 EEE, and EGKT 175 DGE. The residues Thr 169 and Thr 175 are located in the region determined to be required for heterophilic adhesion. It will be interesting in the future to investigate whether this is indeed the location of the ankyrin-binding domain of ␤2 and, if so, whether it is regulated by phosphorylation.
␤1 (Ϫ/Ϫ) mice exhibit disrupted axo-glial contacts in a subset of axons (9). In these axons, the septate-like junctions at the nodal-paranodal boundaries are everted or pulled away from the axon, leaving all other septate-like junctions, seen as trans-verse bands by transmission electron microscopy, intact. We proposed that axonal ␤1 subunits located at the nodal-paranodal border may interact in cis with axonal contactin molecules in the paranode or in trans with glial Nf155 molecules at the nodal-paranodal border region. Alternatively, because we have also shown ␤1 expression in oligodendrocytes (53), it is possible that glial ␤1 subunits may interact in trans with axonal ␤1 subunits, in cis with glial Nf155, or in trans with axonal contactin molecules. In the present study, we tested the hypothesis that ␤1 interacts with Nf155 and showed that these molecules interact in our assay. Thus, we propose that ␤1 is associated with sodium channels at the node of Ranvier, where it serves a dual function as a channel modulator and as a CAM. In addition, we propose that ␤1 may be expressed in the absence of sodium channels in the nodal-paranodal border region of axons or in oligodendrocytes, where it functions as a CAM only. Interaction of ␤1 with sodium channels at nodes of Ranvier results in modulation of channel function and stabilization of channel cell surface expression through interactions with ␤2, contactin, Nf186, NrCAM, and ankyrin G . Axonal ␤1 subunits located at the nodal-paranodal border may also interact in cis with paranodal contactin or in trans with glial Nf155. ␤1 subunits expressed in oligodendrocytes may participate in the formation of septate-like junctions at the nodal-paranodal border through cell-adhesive interactions in cis with Nf155 and/or in trans with contactin. It will be necessary to perform in vivo experiments using ␤1 mutant constructs to investigate the importance of the different functional domains of ␤1 on the maintenance and stabilization of nodes of Ranvier.