Sodium Channel β1 Subunits Promote Neurite Outgrowth in Cerebellar Granule Neurons*

Many immunoglobulin superfamily members are integral in development through regulation of processes such as growth cone guidance, cell migration, and neurite outgrowth. We demonstrate that homophilic interactions between voltage-gated sodium channel β1 subunits promote neurite extension in cerebellar granule neurons. Neurons isolated from wild-type or β1(-/-) mice were plated on top of parental, mock-, or β1-transfected fibroblasts. Wild-type neurons consistently showed increased neurite length when grown on β1-transfected monolayers, whereas β1(-/-) neurons showed no increase compared with control conditions. β1-Mediated neurite extension was mimicked using a soluble β1 extracellular domain and was blocked by antibodies directed against the β1 extracellular domain. Immunohistochemical analysis suggests that the β1 and β4 subunits, but not β2 and β3, are expressed in cerebellar Bergmann glia as well as granule neurons. These results suggest a novel role for β1 during neuronal development and are the first demonstration of a functional role for sodium channel β subunit-mediated cell adhesive interactions.

Many immunoglobulin superfamily members are integral in development through regulation of processes such as growth cone guidance, cell migration, and neurite outgrowth. We demonstrate that homophilic interactions between voltage-gated sodium channel ␤1 subunits promote neurite extension in cerebellar granule neurons. Neurons isolated from wild-type or ␤1(؊/؊) mice were plated on top of parental, mock-, or ␤1-transfected fibroblasts. Wild-type neurons consistently showed increased neurite length when grown on ␤1transfected monolayers, whereas ␤1(؊/؊) neurons showed no increase compared with control conditions. ␤1-Mediated neurite extension was mimicked using a soluble ␤1 extracellular domain and was blocked by antibodies directed against the ␤1 extracellular domain. Immunohistochemical analysis suggests that the ␤1 and ␤4 subunits, but not ␤2 and ␤3, are expressed in cerebellar Bergmann glia as well as granule neurons. These results suggest a novel role for ␤1 during neuronal development and are the first demonstration of a functional role for sodium channel ␤ subunit-mediated cell adhesive interactions.
Intercellular communications mediate critical developmental events in neurons. Interactions between integrins, cadherins, and immunoglobulin superfamily cell adhesion molecules (IGSF CAMs) 1 on opposing cells result in events such as growth cone guidance and neurite extension. For example, NCAM-and L1-CAM-mediated cell adhesive interactions result in signal transduction pathways involving kinase activation, modulation of local, submembraneous calcium concentrations, gene transcription, and ultimately, neurite extension (1)(2)(3)(4). Some IGSF CAMs such as myelin-associated glycoprotein and contactin balance the growth promoting activity of other molecules through inhibition of neuritogenesis (5,6). Thus, it is the concerted effort of growth-promoting and growth-inhibitory molecules on neuronal and non-neuronal cells that act to influence the developing nervous system.
Postnatal cerebellar development involves migration of cerebellar granule neurons from the external germinal layer to the rapidly developing granule cell layer. During migration, a granule neuron develops several neurites, two of which ultimately become parallel fibers of the cerebellar molecular layer (7). Whereas most cell migration and neuritogenesis in the cerebellum is complete within the second postnatal week, migration of granule neurons, growth of the granule cell layer, and extension of parallel and vertical fibers continues through postnatal day 21 (P21) (7,8).
Voltage-gated sodium channels are composed of a central, pore forming ␣-subunit and one or two ␤ subunits (9). Whereas ␣ alone is sufficient to form the ion-conducting pore, current density, channel kinetics, gating mode, and channel cell surface density are influenced by ␤ subunit expression (9,10). There are five known ␤ subunits: ␤1, ␤1A, ␤2, ␤3, and ␤4. ␤1, ␤1A, and ␤3 are non-covalently linked to the pore-forming ␣ subunit, while ␤2 and ␤4 are disulfide linked to ␣. Based on structural and amino acid homologies, ␤ subunits are IGSF CAMs (11). ␤1 and ␤2 exhibit homophilic and heterophilic adhesion and interact with extracellular matrix molecules (12)(13)(14)(15)(16). ␤1and ␤2-mediated homophilic cell adhesion results in recruitment of ankyrin to points of cell-cell contact, which in ␤1 can be inhibited by phosphorylation of a single, intracellular tyrosine residue (12,17). ␤1 and ␤2 mRNAs are expressed as early as P1 in brain. Although ␤1 and ␤2 mRNA expression is initially low in cerebellum, their expression is more robust from P14 through adulthood (18). ␤3 mRNA expression is detected in brain as early as embryonic day 10. ␤3 then decreases in brain after P3 in certain areas (including the cerebellum) but remains high in the hippocampus and striatum (18). In adult rats, ␤4 mRNA is expressed in the cerebral cortex, cerebellar purkinje cells, hippocampus, caudate putamen, and globus pallidus (19). However, the expression profile of ␤4 at earlier time points is not known.
Here we examine the role of ␤-subunits in postnatal cerebellar granule cell neurite extension. We examine the effects of ␤1, ␤2, and ␤4 presented on the surface of fibroblast monolayers, or in soluble form, on neurite growth promotion in cerebellar granule cells isolated from ␤1(ϩ/ϩ) or ␤1(Ϫ/Ϫ) mice. We show that ␤1 promotes, whereas ␤2 reduces the level of basal neurite extension. The ␤4 subunit has no measurable effect on neurite outgrowth. Previous studies used heterologous expression systems, such as Drosophila S2 cells, to show ␤ subunit-mediated cell adhesive interactions (12). The present study is the first functional demonstration of ␤-subunit-mediated cell adhesion in neurons.
The observation that an IGSF CAM can promote neurite outgrowth is far from novel. What is novel, is that proteins that participate in voltage-dependent ion channel gating can also function as CAMs that participate in extracellular and intracellular signal transduction leading to neurite extension. We propose that ␤ subunits, as CAMs, participate in inter-and intracellular communication and thus may play critical roles in neuronal development.

EXPERIMENTAL PROCEDURES
Cerebellar Dissociation-␤1(ϩ/ϩ) and ␤1(Ϫ/Ϫ) mice were generated and maintained as previously described, in accordance with the guidelines of the University of Michigan Committee on the Use and Care of Animals (20). Animals used in this study were bred from ␤1(ϩ/Ϫ) mice that had been repeatedly backcrossed to C57Bl/6 mice for at least 10 generations, creating congenic strains. The ␤1(ϩ/ϩ) and ␤1(Ϫ/Ϫ) mice used in each individual experiment were age-matched littermates. Following cervical dislocation, cerebella from ␤1(ϩ/ϩ) or ␤1(Ϫ/Ϫ) mice at the ages indicated in the figure legends were quickly excised and placed in ice-cold Hibernate A/B-27 with L-glutamine. Tissue was cut into small pieces and cerebellar granule cells were dissociated according to the method described by Brewer with slight modifications (21). Briefly, cerebellar tissue was subjected to trypsin treatment (1 mg/ml) in Hibernate A (Brain Bits) for 30 min at 37°C with shaking at 250 rpm. Following trypsinization, tissue was allowed to settle for 2-5 min and Hibernate A/trypsin was aspirated. Tissue was resuspended in 2 ml of Hibernate A/B-27 and triturated using a fire-polished glass pipette or P 1000 pipette tip. Neurons were isolated by centrifugation for 15 min at 800 ϫ g using a 4-ml Opti-Prep (Axis-Shield) density gradient (35,25,20, and 15%) made with Hibernate A/B-27. Fraction 3, which is highly enriched in neurons, was collected and resuspended in 12 ml of Neurobasal A/B-27 with 0.5 mM L-glutamine. Cells were centrifuged at 800 ϫ g for 5 min, resuspended in Neurobasal A/B-27 with 0.5 mM L-glutamine, 10 g/ml gentamycin, and 5 ng/ml FGF-␤, then plated at 2 ϫ 10 4 cells/well. Cell viability was measured using the trypan blue exclusion assay.
Immunocytochemical Analysis of CHL Cells-Confluent monolayers of 1610 Chinese hamster lung (CHL) (13) or CHL-␤1 (13) cells were fixed in 4% paraformaldehyde for 40 min at room temperature. Following fixation, cells were blocked with a solution containing 5% nonfat dry milk and 1% bovine serum albumin. Cells were then incubated for 1 h with rabbit polyclonal anti-␤1 ex antibody (12) at 1:100 dilution. Goatanti rabbit Alexa Fluor 594 (Molecular Probes) at 1:500 dilution was used as the secondary antibody. Monolayers were visualized using a Zeiss Axiophot fluorescent microscope at the Microscopy and Image Analysis Core facility at the University of Michigan. Digital images were processed using Adobe Photoshop.
Creation of CHL␤4 Stable Cell Line-Rat brain total RNA was extracted using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Reverse transcriptase-PCR was performed using the Titan One RT-PCR kit (Roche Diagnostics) with primers corresponding to the 5Ј and 3Ј ends of the rat ␤4 cDNA sequences (GenBank TM accession numbers BK001030 and XM_236199): 5Ј-CT-GTGCACACTGTCCTATCCAAGC and 3Ј-CACATCTCAGACAGGAC-TCGGCATC. The resulting reverse transcriptase-PCR product was used as the template in a second PCR using nested ␤4 primers: 5Ј-CAACTCGAGCGCTCCGGAGAGAACAGGAC and 3Ј-ATCGAATTC-ACCATCAGAAAGTGAGGCTC. The resulting PCR product was gelpurified, subcloned into pcDNA3.1/hygro(Ϫ) (Invitrogen), and sequenced to confirm identity with ␤4. CHL cells were transfected with pcDNA3.1/hygro(Ϫ)/␤4 using FuGENE (Roche) and stable cell colonies were selected in the presence of 500 g/ml hygromycin (Invitrogen). Northern blot analysis of the resulting CHL-␤4 cell lines was performed using full-length ␤4 antisense cDNA as the probe label with digoxigenin. 10 g of total RNA was extracted from each cell line, and loaded onto an RNA gel. Following transfer, hybridization and detection was as previously described (20). These results were confirmed by Western blot analysis with an anti-␤4 antibody (1:200, gift from Dr. W. A. Catterall) followed by an anti-rabbit horseradish peroxidase-conjugated secondary.
Neurite Outgrowth Assay-Parental (CHL), pcDNA3 mock-transfected (CHL-mock), or stably transfected ␤1 (CHL-␤1), ␤2 (CHL-␤2), ␤1␤2 (CHL-␤1␤2), Na v 1.2␤1 (CHL-␣␤1), Na v 1.2␤2 (CHL-␣␤2), or Na v 1.2␤1␤2 (CHL-␣␤1␤2) subunit expressing 1610 CHL cells were plated at 4 ϫ 10 4 cells/well in 8-well chamber slides and grown for 24 h (13,14). Upon the establishment of confluent monolayers, freshly dissociated cerebellar granule cells were plated (2 ϫ 10 4 cells/well) on top of cell monolayers and allowed to grow in Neurobasal A/B-27 with L-glutamine, 10 g/ml gentamycin, and 5 ng/ml FGF-␤ for 19 -22 h. Cells were fixed with 4% paraformaldehyde and visualized using a primary monoclonal antibody to GAP 43 (Chemicon, 1:500) followed by and Alexa Fluor 488-conjugated anti-mouse antibody (Molecular Probes, 1:500). Images were captured on a Zeiss Axiophot florescent microscope using a ϫ40 objective in the Microscopy and Image Analysis Core facility at the University of Michigan. Digital images were processed using Adobe Photoshop. Images used for analysis were randomly selected by NIH Image or Image J software for all conditions. The longest neurite from each of the first 50, randomly selected, isolated granule cells were measured in at least five experiments per condition. This process of analysis eliminated user bias. A neurite was defined as a process protruding from GFAP positive cells, with neuronal morphology, ranging in length from ϳ0.1 m to greater than 200 m in some instances.
Construction and Analysis of the Soluble ␤1 Subunit-A vector constructed to express an inducible, truncated form of the ␤1 subunit was prepared using PCR with pcDNA3.1.␤1 encoding the rat ␤1 subunit (GenBank TM accession code NM_017288) as template, forward primer: 5Ј-GGATCCTTGCGCGGCCATGGGGAC-3Ј, and reverse primer: 5Ј-GGAATTCTCTGACACGATGGATGC-3Ј. These primers introduced BamHI and EcoRI sites at the 5Ј and 3Ј ends, respectively, of the extracellular ␤1 domain. PCR products were ligated in-frame into the pcDNA 4/TO vector (Invitrogen), transformed, and minipreps were performed. The pcDNA 4/TO vector is extremely useful because it is tetracycline responsive and provides carboxyl-terminal c-myc and His 6 tags. Successful cloning was confirmed by restriction digest with EcoRI and BamHI as well as dideoxy sequencing. The newly constructed pcDNA4/TO-␤1 vector was stably cotransfected into CHL cells with pcDNA6/TR that encodes a tetracycline repressor protein under the control of the human cytomegalovirus promotor (CHLSol␤1 cells).
Western blot analysis was carried out on 5 l of tissue culture medium exposed for 22 h to a single confluent well of either uninduced CHLSol␤1 cells or 1 mg/ml tetracycline-induced CHL or CHLSol␤1 cells (200 l total volume). Blots were probed with mouse anti-c-myc antibody (Santa Cruz, 1:500) for 1 h at room temperature. Secondary horseradish peroxidase-conjugated, anti-mouse antibody (Cell Signaling, 1:2,000) was incubated with blots for 1 h prior to washing and chemiluminescent detection using West Dura reagent (Pierce).
Affinity Purification of the CHLSol␤1 Subunit-Confluent T-75 flasks of CHLSol␤1 cells were induced to express the soluble ␤1 subunit using 1 mg/ml tetracycline overnight. Collected media was centrifuged at 500 ϫ g to remove cells and debris. Media was then collected and incubated for 1-2 h with nickel metal hydride-agarose (Qiagen). Nonspecific interactions with the nickel-agarose were disrupted using a washing buffer containing 500 mM NaCl, 20 mM NaHPO 4 , and 20 mM imidazole (pH 8.0) prior to elution. The CHLSol␤1 protein was eluted from nickel-agarose using 200 mM imidazole in the wash buffer.
Immunohistochemistry--P12 or P18 C57Bl/6 mice were anesthetized by administering an intraperitoneal injection of 0.2 ml of sodium pentobarbital (50 mg/ml, Abbott Laboratory) and perfused intracardi- ally with phosphate-buffered saline followed by 4% paraformaldehyde. Brains were removed and post-fixed in 4% paraformaldehyde for 2-5 h followed by immersion in 30% sucrose overnight. Brains were frozen in Ϫ20°C 2-methylbutane for 1 min and then stored at Ϫ80°C until sectioning. Brain sections were cut to a thickness of 10 m, mounted on glass slides, and stored at Ϫ20°C until use.
Immunohistochemistry was performed using protein A-Sepharosepurified anti-␤1 ex (1:200) or anti-␤4 (1:25). The anti-␤4 (19) antibody was a gift from the laboratory of Dr. W. A. Catterall. The monoclonal anti-GFAP antibody (Molecular Probes) used to identify cerebellar Bergmann glia was used at a dilution of 1:250. Anti-rabbit Alexa Fluor 488 nm and anti-mouse Alexa Fluor 594 nm secondary antibodies were obtained from Molecular Probes and used as secondary antibodies. Images were captured on a Zeiss Axiophot florescent microscope using a ϫ40 objective in the Microscopy and Image Analysis Core facility, University of Michigan. Digital images were processed using Adobe Photoshop.

␤1 Promotes Neurite Extension from Granule Neurons-To
determine whether sodium channel ␤ subunits play a role in neurite outgrowth, as described for other members of the IGSF, we examined the growth behavior of acutely dissociated P14 -P21 mouse cerebellar granule neurons plated on monolayers of non-transfected CHL cells, mock-transfected cells (CHL-mock), or ␤1-stably transfected cells (CHL-␤1). CHL cells offer a benefit over other cell lines in that they do not express endogenous sodium channel ␤ subunits (22). Immunohistochemical analysis of P18 mouse cerebellum shows ␤1 staining in the granule neurons as well as in the Purkinje cell layer (Fig. 1A). Immunocytochemical and Western blot analyses of CHL-␤1 cells indicate that they express moderate levels of ␤1 subunits (Fig.  1, B and C). Neurite outgrowth experiments were performed whereby 2 ϫ 10 4 acutely dissociated cerebellar granule neurons were plated on confluent fibroblast monolayers and allowed to grow for 19 -22 h prior to fixation and neurite length determination. Staining for GAP 43 ( Fig. 2A) revealed neurons that typically contained two or three processes. Sodium channel ␤1 subunits presented by the monolayer cells (CHL-␤1) resulted in increased neurite extension by the overlying granule neurons compared with neurons grown on CHL cells. As shown in Fig.  2B, growth of cerebellar granule neurons on CHL or CHL-mock monolayers yielded an average neurite length of 41.5 Ϯ 1.8 m and 46.3 Ϯ 4.2 m, respectively. In contrast to granule neurons grown on CHL or CHL-mock monolayers, neurons grown on CHL-␤1 fibroblasts consistently showed a 1.4 -1.7-fold increase in neurite extension with an average neurite length of 68.4 Ϯ 4.3 m. This increase was highly reproducible, and could not be attributed to individual outliers because there was always a greater percentage of neurons at any given neurite length when granule neurons were grown on CHL-␤1 monolayers versus control monolayers (Fig. 2C). As shown in the neurite length distribution (Fig. 2C), granule cells grown on CHL or CHL-mock monolayers had 9.2 or 9.8% of neurites with a length of 100 m or greater, respectively. This is in contrast to granule neurons grown on CHL-␤1 monolayers where 23.8% of neurites had a length of 100 m or greater. next used an antiserum specific to the Ig loop region of ␤1 (anti-␤1 ex ) to examine the specificity of the observed increases in neurite outgrowth. We reasoned that if the observed effects were truly ␤1 specific, involving extracellular, cell adhesive events, then neurite outgrowth on CHL-␤1 monolayers should be blocked by the addition of antibodies that recognize the ␤1 Ig loop domain. When 5 g/ml ␤1 ex antibody, directed against the A-AЈ face of the ␤1 Ig loop domain (12) (illustrated in Fig. 5A), was added to co-cultures, we observed a blockade of ␤1-stimulated neurite extension from granule neurons grown on CHL-␤1 monolayers compared with neurons grown on CHL-␤1 monolayers but in the absence of antibody (Fig. 3). Comparison of the ␤1 amino acid sequence with the crystal structure of myelin P o predicts that the anti-␤1 ex antibody epitope lies in a region that is critical for trans-homophilic cell adhesion (23, 24) (Fig. 5A). We observed no changes in neurite length when ␤1 ex antibodies were added to granule neurons grown on CHL monolayers. The addition of non-immune IgG had no effect on granule neurons grown on either CHL or CHL-␤1 monolayers. These results, taken together with our previous results, suggest that ␤1 may play a role in neurite extension from acutely dissociated P14 -P21 cerebellar granule cells.
The ␤1 Extracellular Domain Promotes Neurite Outgrowth-The results obtained thus far suggest that the sodium channel ␤1 subunit is able to promote neurite extension from cerebellar granule cells when expressed by monolayers of CHL fibroblasts. Our results also suggest that ␤1-enhanced neurite outgrowth requires ␤1 expression by both the neuron and monolayer. Although CHL fibroblasts do not express endogenous ␤ subunits, they may express a heterogeneous population of cell surface molecules that aid in the co-activation of neuritogenesis when ␤1 is present. Alternatively, ␤1-␤1 trans-homophilic adhesion may result in signal transduction and secretion of a growth-promoting molecule from the monolayer fibroblasts. To test these possibilities, we took advantage of our observation that untransfected CHL monolayers do not enhance neurite length in our system and developed a stable cell line expressing a tetracycline-inducible, c-myc and His 6 -tagged ␤1 subunit con- struct truncated at the extracellular juxtamembrane region (CHLSol␤1, Fig. 5A). This truncation allows for secretion of the ␤1 extracellular domain (c-myc-␤1) into the tissue culture media. Treatment of cells with 1 g/ml tetracycline for 22 h followed by Western blotting with anti-c-myc antibody shows that this cell line expresses high levels of c-myc-␤1 subunits (Fig. 5B). Uninduced cells express a low level of c-myc-␤1 protein. However, as shown below, this level of expression was not sufficient to promote neurite extension above control levels. Utilizing CHLSol␤1 cells as a monolayer, we examined the effect of the recombinant ␤1 extracellular domain on neurite outgrowth. Fig. 5C illustrates that c-myc-␤1 increased granule cell neurite length to a similar level as full-length ␤1 subunits. We obtained similar results when granule cells were grown on untransfected CHL monolayers in the presence of 100 nM nickel-nitrilotriacetic acid-purified CHLSol␤1 subunit added to the cell culture medium (Fig. 5D). Together, these results indicate that the ␤1 extracellular domain, including the Ig loop and juxtamembrane region, is sufficient to induce cerebellar granule cell neurite outgrowth in this system.

␤2 Reduces Basal Neurite Extension, whereas ␤4 Has No Effect on Neurite Extension in Granule
Neurons-To date, there are five known ␤-subunits ␤1, ␤1A, ␤2, ␤3, and ␤4 (9,19,27). ␤1 and ␤2 share similar functions as CAMs because both mediate homophilic adhesion, recruit ankyrin, and interact with tenascin-C and -R. Because ␤1 promotes neuritogenesis in our assay system, we were interested to investigate whether ␤2 or ␤4 might also function in a similar capacity. Utilizing monolayers of CHL cells stably transfected with ␤2 or ␤4 subunits (CHL-␤2, CHL-␤4), we examined whether the expression of ␤2 or ␤4 affected neurite length. CHL cells transfected with ␤4 subunit cDNA showed robust mRNA and protein expression, as assessed by Northern and Western blot, respectively (Fig. 6A). As shown in Fig. 6B, expression of ␤4 by CHL cells had no observable effects on granule cell neurite extension.
In contrast to cerebellar granule neurons grown on CHL-␤1 or CHL-␤4 monolayers, those grown on CHL monolayers expressing ␤2 consistently had shorter neurite lengths than neurons grown on untransfected CHL monolayers (Fig. 7). This apparent inhibitory effect of the ␤2 subunit was not nearly as robust as that observed for ␤1-enhanced outgrowth. Relative to neurons grown on CHL monolayers, those grown on CHL-␤2 monolayers showed an average 1.2-fold decrease in neurite length from control with a mean neurite length of 28.3 Ϯ 2.5 versus 34.5 Ϯ 3.2 m for granule cells grown on CHL monolayers. These results suggest that the ␤1 and ␤2 subunits may be able to act antagonistically to regulate neuronal development.
Effect of Sodium Channel ␣ Subunit Coexpression-We demonstrated previously that ␤1 and ␤2 subunits function as CAMS in vitro, both in the presence and absence of the poreforming ␣ subunit, Na v 1.2 (12). The results presented above show that ␤ subunits expressed on a cell monolayer in the absence of ␣ subunits regulate neurite extension from cerebellar granule neurons. We next examined whether ␣␤ subunit coexpression by the monolayer would affect the observed ␤ subunit-mediated effects on cerebellar neurite outgrowth, for example, by masking a critical site on ␤. Using acutely dissociated granule neurons from P14 -P21 mice, we observed significant increases in neurite length from granule neurons grown on CHL-␣␤1 cells compared with untransfected monolayers. This effect was comparable with that observed for ␤1 alone (Table I). Consistent with our previous results in S2 cells, the present results show that ␤1-␤1 homophilic interactions are not modulated by ␣ subunit association. In contrast, we found no significant differences in neurite length for CHL-␣␤2, CHL-␣␤1␤2, or CHL-␤1␤2 cells compared with CHL monolayers. Interestingly, these results suggest that ␤1-␤2 interactions may prevent ␤1-␤1 homophilic adhesion and subsequent signal transduction. Alternatively, the effect of ␤2 subunits to inhibit neurite extension may be dominant over ␤1-mediated promotion of neurite outgrowth.
Developmental Regulation-Our results thus far demonstrate that sodium channel ␤1 and ␤2 subunits modulate neurite extension in P14 -P21 cerebellar granule neurons via trans-homophilic cell adhesive interactions. We next investigated whether neurons isolated from older animals exhibited similar responses. Table I shows the results of ␤ subunitmediated neurite extension from cerebellar granule neurons isolated from P22-P26 mice. In these cells, ␤1 subunits presented by the monolayer in the absence of ␣ or ␤2 subunits (CHL-␤1) produced a ϳ1.5-fold increase in neurite length, similar to that observed for P14 -P21 neurons. However, in contrast to experiments performed with neurons isolated from P14 -P21 mice, CHL-␣␤1 monolayers did not promote neurite extension, suggesting that the response of the neurons to sodium channel subunits presented by the substrate changes with development.
Immunolocalization of ␤ Subunits in Cerebellum-During late migration from the external germinal layer to the developing granule cell layer, granule neurons appose Bergmann glia and travel en masse to their final destination. In situ hybridization studies have suggested that ␤1 mRNA is present in the molecular layer of the cerebellum from P14 through adulthood in rats (18). If sodium channel ␤1 subunits were expressed on Bergmann glial plasma membranes or on the plasma membranes of adjacent granule or Purkinje neurons, then they might serve as in vivo equivalents of the ␤1-transfected fibroblast monolayers used in our co-culture system. To test this, we performed immunohistochemistry on P12 or P18 mouse cerebellum to determine whether Bergmann glia express sodium channel ␤ subunits. RC-2 is the marker of choice for Bergmann glia in the early phases of development prior to birth, however, immunoreactivity for this antigen is absent following the second postnatal week (27). From P14 to adult- hood, Bergmann glia express moderate amounts of GFAP (28). Sagittal sections double stained for GFAP and ␤1 subunits showed co-localization of these two proteins in Bergmann glia (Fig. 8). This staining pattern is similar to that recently observed for Na v 1.6 in these cells (29). In addition to staining in Bergmann glia, ␤1 was also expressed in granule neurons and Purkinje cells (Figs. 1A and 8). Immunohistochemical localization of ␤2, ␤3, and ␤4 subunits revealed that ␤4 is also present in Bergmann glia during this time period, but ␤2 and ␤3 are not present (data not shown). Granule neurons were immunopositive for both ␤1 and ␤4 subunits. DISCUSSION Sodium channel ␤1 and ␤2 subunits regulate neurite extension from cerebellar granule neurons. We observed a level of ␤1-enhanced neurite outgrowth that was consistently on the order of 1.4 -1.7-fold when compared with control monolayers, levels similar to those published for a variety of IGSF molecules (3,30). This ␤1-mediated effect was specific and could be blocked by addition of an antibody directed against the A-AЈ face of the ␤1 Ig domain, suggesting that trans-homophilic adhesion involving this region may be responsible for our observed results. The soluble ␤1 extracellular domain was as effective in promotion of neurite extension as full-length ␤1 subunits. Neurons isolated from ␤1(Ϫ/Ϫ) mice did not respond to ␤1-expressing monolayers. Thus, we propose that, similar to L1-CAM and NCAM, ␤1-␤1 trans-homophilic cell adhesive in-teractions stimulate neurite extension via activation of second messenger cascades in the neuron (4).
In contrast to ␤1, ␤2, when presented by the substrate, reduced the extent of basal neurite outgrowth from wild-type cerebellar granule neurons. This is an important finding because neuronal development involves a balance between growth-promoting and growth-inhibiting events. The observation that ␤2 reduces neurite extension may be related to its homology to the third Ig loop and juxtamembrane region of F3/contactin, an IGSF CAM capable of strong neurite growth inhibition in monolayer assays (6,31). Whereas mRNA for ␤2 is found within the molecular layer of the cerebellum in adult rats, in situ hybridization studies as well as the present results indicate that P14 -P21 GFAP positive cells, such as Bergmann glia, are devoid of detectable ␤2 expression (18,32). We suggest that ␤1 and ␤2 may act in an antagonistic manner if presented as substrates to the same neuronal cell.
Co-expression of Na v 1.2 with either or both ␤ subunits revealed that there are differences in the response of cerebellar granule neurons to ␤1, depending on their level of maturation. Granule neurons isolated from P14 -P21 mice responded similarly to ␤1-expressing and to ␣␤1-expressing monolayers, whereas neurons isolated from P22-P26 mice displayed enhanced neurite extension in response to cells expressing ␤1 alone but not to those expressing the ␣␤1 complex. Perhaps different domains of ␤1 are required for neurite extension as  We observed that neurons isolated from ␤1(Ϫ/Ϫ) mice do not extend neurites in response to ␤1 presented by the monolayer. These results demonstrate that the mechanism responsible for ␤1-mediated neurite extension requires ␤1 expression on the neuron as well as on the substrate cell. Studies using NCAM(Ϫ/Ϫ) or L1-CAM(Ϫ/Ϫ) mice indicate that neurons lacking these CAMs exhibit higher levels of pathfinding errors and greater degrees of defasiculation compared with wild-type (33)(34)(35). ␤1(Ϫ/Ϫ) mice exhibit a dramatic behavioral phenotype including seizures, ataxia, and lethality by approximately P21 (20). The molecular phenotype of ␤1(Ϫ/Ϫ) mice includes decreases in action potential conduction velocity, disruption of nodal architecture, decreases in the number of mature nodes of Ranvier in the optic nerve, and altered patterns of sodium channel expression in the hippocampus. Although the observed seizure activity may be caused by altered sodium channel isoform expression in the hippocampus, our present results suggest that errors in neurite extension and pathfinding may also contribute. A human mutation in SCN1B that causes generalized epilepsy with febrile seizures plus type 1 results in the replacement of a key cysteine residue in the ␤1 Ig loop domain with tryptophan (36,37). We have shown that, in addition to inhibiting ␤1-mediated channel modulation, this mutation abolishes ␤1-mediated homophilic cell adhesive interactions (37). Thus, neuronal pathfinding errors may also underlie the generalized epilepsy with febrile seizures plus 1 epileptic phenotype. Finally, pathfinding errors in the cerebella of ␤1(Ϫ/Ϫ) mice may contribute to the observed ataxic phenotype.
Our observations that sodium channel ␤ subunits promote neurite outgrowth suggest the existence of ␤ subunit signaling complexes in neurons. In addition to the pore-forming ␣ subunits, sodium channel ␤1 subunits have been shown to associate with tenascin-R, contactin/F3, neurofascin-186, ankyrin, and the receptor protein-tyrosine phosphatase ␤ (12-14, 16, 38). ␤2 associates with tenascin-C, tenascin-R, and ankyrin, but not with contactin/F3 (12,15). Signaling mechanisms responsible for ␤1-mediated promotion of neurite extension may involve a variety of pathways previously proposed for other CAMs such as mitogen-activated protein or Rho kinases. FGF-(R) receptor-dependent and -independent models have been proposed to explain CAM-mediated neurite outgrowth (1,30,39). In the FGF-R-dependent model, FGF-R and IGSF CAMs interact in a cis-heterophilic manner following initial IGSF CAM trans-homophilic adhesion. Resulting outgrowth occurs via local increases in N-and L-type Ca 2ϩ channel activity (40). These binding events are thought to involve interacting CAM homology domains on the FGF-R and IGSF CAMs (39). The extracellular domains of ␤1 and ␤2 lack CAM homology domains, as described for L1-CAM, NCAM, and N-cadherin, and thus may act through FGF-R-independent signaling events, for example, lipid raft-associated cascades (30). Alternatively, sodium channel ␤ subunits may act indirectly through CAM homology domain-containing molecules. We have shown that ␤1 and N-cadherin interact in heart (41). Posttranslational modifications such as polysialylation have been shown to be important determinates in NCAM-mediated neurite outgrowth and neuronal fasciculation (4). Sodium channel ␤ subunits are heavily glycosylated (42). Thus, the extent and type of glycosylation in ␤1 and ␤2 during different stages of development or in different cell types may be important in their ability to modulate neurite outgrowth. In summary, we propose that sodium channel ␤ subunits modulate electrical signal transduction and, as CAMs, participate in inter-and intracellular communication. Together, these functions make ␤ subunits critical players in neuronal development.