Abstract
Voltage-gated Na+ channel β1 subunits are multifunctional, participating in channel modulation and cell adhesion in vitro. We previously demonstrated that β1 promotes neurite outgrowth of cultured cerebellar granule neurons (CGNs) via homophilic adhesion. Both lipid raft-associated kinases and nonraft fibroblast growth factor (FGF) receptors are implicated in cell adhesion molecule-mediated neurite extension. In the present study, we reveal that β1-mediated neurite outgrowth is abrogated in Fyn and contactin (Cntn) null CGNs. β1 protein levels are unchanged in Fyn null brains, whereas levels are significantly reduced in Cntn null brain lysates. FGF or EGF (epidermal growth factor) receptor kinase inhibitors have no effect on β1-mediated neurite extension. These results suggest that β1-mediated neurite outgrowth occurs through a lipid raft signaling mechanism that requires the presence of both fyn kinase and contactin. In vivo, Scn1b null mice show defective CGN axon extension and fasciculation indicating that β1 plays a role in cerebellar microorganization. In addition, we find that axonal pathfinding and fasciculation are abnormal in corticospinal tracts of Scn1b null mice consistent with the suggestion that β1 may have widespread effects on postnatal neuronal development. These data are the first to demonstrate a cell-adhesive role for β1 in vivo. We conclude that voltage-gated Na+ channel β1 subunits signal via multiple pathways on multiple timescales and play important roles in the postnatal development of the CNS.
- voltage-gated Na+ channel
- neurite outgrowth
- auxiliary subunit
- cell adhesion molecule
- contactin
- fyn kinase
Introduction
Voltage-gated ion channels are multifunctional, regulating electrical excitability, intracellular signaling, transcriptional regulation, scaffolding, and cell adhesion (Dolmetsch, 2003; MacLean et al., 2003, 2005; Hegle et al., 2006; Kaczmarek, 2006; Levitan, 2006). Neuronal voltage-gated Na+ channels (VGSCs) contain one pore-forming α subunit, one noncovalently associated β subunit (β1 or β3), and one covalently associated β subunit (β2 or β4) (Catterall, 2000; Meadows and Isom, 2005). β1 modulates channel kinetics, voltage dependence, and cell surface expression when expressed in vitro (Isom et al., 1992). In vivo, β1 also modulates electrical excitability: Scn1b null mice are ataxic and display spontaneous generalized seizures (Chen et al., 2004). Mutations in SCN1B result in human brain disease, including generalized epilepsy with febrile seizures plus and temporal lobe epilepsy (Wallace et al., 1998, 2002; Meadows et al., 2002; Audenaert et al., 2003; Scheffer et al., 2007). VGSC β subunits are unique among ion channel auxiliary subunits in that they also function as immunoglobulin superfamily cell adhesion molecules (IGSF-CAMs), directing VGSC insertion into the plasma membrane, interacting with other signaling proteins, and participating in adhesion in vitro (Isom et al., 1994; Isom and Catterall, 1996).
Increasing molecular evidence implicates VGSC β subunits as key mediators in cell adhesion. β1 and β2 interact with tenascin-C and tenascin-R influencing cell migration, and participate in homophilic cell adhesion resulting in cellular aggregation and ankyrin recruitment (Srinivasan et al., 1998; Xiao et al., 1999; Malhotra et al., 2000, 2002). Furthermore, β1 interacts heterophilically with N-cadherin, contactin, neurofascin-155, neurofascin-186, NrCAM, and VGSC β2 (Kazarinova-Noyes et al., 2001; Malhotra et al., 2004; McEwen and Isom, 2004; McEwen et al., 2004). Interactions between β1 and contactin, neurofascin-186, or β2 result in increased VGSC surface expression (Kazarinova-Noyes et al., 2001; McEwen and Isom, 2004; McEwen et al., 2004). We showed that β1 promotes neurite outgrowth from acutely dissociated cerebellar granule neurons (CGNs) via trans-homophilic cell adhesive interactions, and that this effect is blocked in neurons from Scn1b null mice (Davis et al., 2004). However, the signaling mechanism has not yet been elucidated.
IGSF-CAMs are known to localize to cholesterol and sphingolipid-rich membrane domains (“lipid rafts”) that are rich in signaling molecules (Olive et al., 1995; Simons and Toomre, 2000; Kasahara et al., 2002; Niethammer et al., 2002; Schafer et al., 2004). β1 contains a putative palmitoylation site (McEwen et al., 2004), a common feature of raft-associated proteins, and colocalizes with known raft proteins on sucrose gradients after Lubrol-WX solubilization (Wong et al., 2005). Regulation of CAM-mediated neurite outgrowth involves signaling through the lipid raft-associated nonreceptor tyrosine kinase fyn (Beggs et al., 1994; Ignelzi et al., 1994; Kolkova et al., 2000). A second nonraft-associated signaling route via the fibroblast growth factor receptor (FGFR) is also known (Niethammer et al., 2002; Sanchez-Heras et al., 2006; Maness and Schachner, 2007). In the case of at least one IGSF-CAM, neural cell adhesion molecule 140 (NCAM-140), neurite outgrowth is proposed to occur via a mechanism that requires both raft and nonraft signaling pathways (Niethammer et al., 2002).
The aims of the present study were twofold: First, to evaluate the signaling mechanism(s) underlying β1-mediated neurite outgrowth; and second, to assess the effect(s) of the Scn1b null mutation on neuronal development in vivo. We demonstrate that β1-mediated neurite outgrowth requires the presence of contactin and fyn kinase, but FGFR- and epidermal growth factor receptor (EGFR)-mediated signaling pathways are not involved. The Scn1b null mutation results in neuronal pathfinding abnormalities in the cerebellum and corticospinal tract (CST). We propose that β1 functions as a CAM in vivo, and that mutations in Scn1b result in defective development of neurons in the CNS, leading to altered excitability.
Materials and Methods
Animals.
Scn1b wild-type and null mice were generated and maintained as described previously, in accordance with the guidelines of the University of Michigan Committee on the Use and Care of Animals (Chen et al., 2004). Mice were bred from Scn1b heterozygous animals that had been repeatedly backcrossed to C57BL/6 mice for at least 15 generations, creating congenic strains. Fyn wild-type (B6;129SF2/J) and Fyn null (B6;129S7-Fyn tm/Sor) mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Fyn heterozygotes were then mated to produce litters containing wild-type, null, and heterozygous genotypes. Cntn wild-type and null mice were derived from a mixed line (129SVJ × C57BL/6 × Black Swiss) (Berglund et al., 1999). Animals used in each individual experiment were littermates.
Dissociation and culture of cerebellar granule neurons.
Dissociation of cerebellar tissue from mice [postnatal day 10 (P10) to P12 for Cntn wild-type and null, P14 for all others] and neurite outgrowth assays were described previously (Davis et al., 2004). In some experiments, CGNs were grown in medium supplemented with one or more of the following agents 2 h after seeding: FGF (20 ng/ml; Sigma, St. Louis, MO), 1-tert-butyl-3-[6-(3,5-dimetoxy-phenyl)-2-(4-diethylamino-butylamino)pyrido[2,3-d]pyrimidin-7-yl]-urea (PD173074) (50 nm; Pfizer Global Research, Groton, CT) (Niethammer et al., 2002), or 4-(3-chloro-anilino)-6,7-dimethoxyquinazoline (AG1478) (2 μm; Calbiochem, La Jolla, CA) (Liu et al., 1999). Neurite length measurements were obtained from a minimum of 30 cells per condition, from at least three independent experiments. Data from individual experiments were combined to provide an overall mean and SEM.
Keratinocyte culture and proliferation assay.
N/TERT human keratinocytes, kindly provided by Dr. J. T. Elder (Departments of Dermatology and Radiation Oncology, University of Michigan, Ann Arbor, MI) were cultured as described previously (Dickson et al., 2000). Briefly, cells were plated at a density of 5 × 103 cells/ml in keratinocyte serum-free medium (Invitrogen, Carlsbad, CA) supplemented with 0.4 mm CaCl2, 30 μg/ml bovine pituitary extract, and 0.1 ng/ml EGF. Twenty-four hours after seeding, AG1478 (2 nm to 20 μm) was added to the growth medium and replaced after an additional 24 h. After treatment for 48 h, cells were resuspended by trypsinization and counted. Cell viability was assessed using trypan blue exclusion. Eight measurements were taken per condition, for two independent treatments. Data were combined to provide an overall mean and SEM.
Sucrose step gradient separation.
C57BL/6 mice (P14–P16) were killed by decapitation and brains were quickly removed into 6 ml of ice-cold MES-buffered saline (24 mm MES, 0.15 m NaCl, pH 6.5) including 1% Triton X-100 with Complete protease inhibitors (Roche, Indianapolis, IN). The brain was homogenized using a Dounce, and then centrifuged at 1300 × g for 5 min at 4°C. The supernatant was then homogenized a second time and allowed to settle for 2–5 min on ice. The homogenate was adjusted to 40% sucrose and then 4 ml of a 30% and 3 ml of a 5% sucrose solution, respectively, were layered on top to form a linear sucrose step gradient. The preparation was centrifuged at 100,000 × g in a Beckman (Fullerton, CA) SW-41Ti rotor for 17–24 h at 4°C. Fractions of 550 μl each were collected and analyzed by SDS-PAGE.
Western blot.
SDS-PAGE and transfer to nitrocellulose were performed as described previously (McEwen et al., 2004; Lopez-Santiago et al., 2006). The following primary antibodies were used: (1) anti-β1 antibody (1:1000) (Wong et al., 2005), (2) anti-α-tubulin antibody (1:5000; Cedarlane Laboratories, Burlington, NC), (3) anti-fyn kinase antibody (1:500; Santa Cruz Biotechnologies, Santa Cruz, CA), (4) anti-transferrin receptor antibody (1:1000; Zymed Laboratories/Invitrogen, Carlsbad, CA).
Densitometric analysis was performed using NIH ImageJ software. Signal density was normalized with respect to anti-α-tubulin antibody as a loading control/reference, for at least four repeat experiments.
Immunocytochemistry.
Immunocytochemical analysis of dissociated CGNs was performed as described previously (Davis et al., 2004). Cells were labeled with anti-β1ex antibody (1:500) (Malhotra et al., 2002) followed by Alexa Fluor 488 anti-rabbit antibody (1:500; Invitrogen), and/or with anti-GAP43 antibody (1:500; Chemicon, Temecula, CA) followed by Alexa Fluor 594 anti-mouse antibody (1:500; Invitrogen).
Immunohistochemistry.
P14 Scn1b null and wild-type littermate mice were anesthetized and perfused, and brains were postfixed and frozen as described previously (Davis et al., 2004). Coronal brain sections were cut to a thickness of 20 μm on a Leica (Nussloch, Germany) CM1850 cryostat, and then mounted on glass slides and stored at −20°C until use. Parallel fibers and descending axons were labeled using a monoclonal anti-TAG-1 antibody (1:100; Developmental Studies Hybridoma Bank, University of Iowa, IA), followed by Alexa Fluor 488 anti-mouse antibody (1:500; Invitrogen). Bromodeoxyuridine (BrdU)-incorporated cells were labeled on cerebellar sections from BrdU-injected mice using an anti-BrdU-Alexa Fluor 488- or 594-conjugated antibody (1:10; Invitrogen), followed by anti-mouse Alexa Fluor 594 secondary antibody (1:500; Invitrogen) to amplify BrdU signal intensity. Cell nuclei were counterstained with 4′,6-diamidino-2-phenylinodole (DAPI) (10 μg/ml; Sigma).
Confocal microscopy.
Samples were viewed using an Olympus (Tokyo, Japan) Fluoview 500 confocal laser-scanning microscope with 10 and 100× objectives. Images (1024 × 1024 pixels) were initially processed with the Olympus Optical Fluoview software, and later exported into NIH ImageJ and/or Adobe Photoshop. For imaging of immunohistochemical sections, projections were created from confocal Z-series spanning 20 μm (interplanar distance, 0.5 μm).
Quantitative analysis of BrdU labeling.
Scn1b null and wild-type littermates (P12) were injected intraperitoneally with 50 mg/kg BrdU and killed 48 h later (Vaillant et al., 2003). Animals used in each individual experiment were littermates. Ten images from the cerebellum were analyzed per mouse, for a total of three wild-type and three Scn1b null mice. Data were combined to provide an overall mean and SEM. The following measurements were made using NIH ImageJ software: (1) Thickness of BrdU-positive proliferating zone. For each image, three separate measurements were made from the pial layer (PL) to the innermost limit of BrdU labeling in the external germinal layer (EGL), using the straight line “Measure” function. (2) Thickness of total EGL. At the same positions used in (1), three measurements were made from the PL to the innermost limit of the EGL, determined using DAPI-labeled nuclei. (3) Number of BrdU-positive cells. For each image, a 60-μm-long section of EGL was delineated using the freeform line “Measure” function, and the number of nuclei positive for both BrdU and DAPI was counted within. (4) Number of DAPI-positive cells. For the same 60-μm-long section of EGL, the number of nuclei positive for DAPI was counted to give an indication of the total number of cells in the EGL.
DiI labeling of corticospinal tracts.
Scn1b wild-type and null mice (P9–P10) were anesthetized by hypothermia as described previously (Danneman and Mandrell, 1997; Coonan et al., 2001). DiI crystals of ∼0.5 mm diameter were implanted into one side of the primary motor cortex, at bregma, just lateral to the midline. The skin was resealed with Vetbond (3M, St. Paul, MN), and the mice were allowed to recuperate for up to 1 h at 30°C. The mice were then returned to their cage and were observed to ensure that no significant motor or behavioral impairments occurred, and to ensure normal feeding. Six days after surgery, mice were anesthetized and perfused, and the brain and spinal cord were removed, postfixed, and frozen as described previously (Davis et al., 2004). Brains and spinal cords were sectioned to a thickness of 50 μm. Images were captured at 10× on a Zeiss Axioplan fluorescent microscope at the University of Michigan Microscopy and Image Analysis Laboratory. Brains were studied from a total of seven wild-type and six Scn1b null mice.
Data analysis.
Quantitative data are presented as mean and SEM, unless stated otherwise. AG1478 dose–response data were fitted using Prism 4 (GraphPad Software, San Diego, CA) to a sigmoidal dose–response equation: y = y min + (y max − y min)/(1 + 10([LogIC50 − x]·B)), where y min and y max are the minimum and maximum response plateaus, respectively; IC50 is the concentration of AG1478 for 50% inhibition of proliferation; x is the logarithm of AG1478 concentration; and B is the Hill slope. Pairwise statistical significance was determined with Student's two-tailed paired/unpaired t tests, as appropriate. Multiple comparisons were made using ANOVA followed by Tukey's post hoc analysis. Results were considered to be significant at p < 0.05 (*).
Results
Many of the cell-adhesive properties of VGSC β1 subunits are similar to those described for L1 family CAMs (Isom, 2002). In addition, we showed that β1 associates with N-cadherin (Malhotra et al., 2004). Thus, we chose to investigate whether the mechanism of β1-mediated neurite outgrowth is similar to that of these CAMs. A number of mechanisms for NCAM-, L1-CAM-, and N-cadherin-mediated neurite outgrowth have been proposed, with the greatest emphasis on a signaling pathway in the growth cone that requires CAM-mediated activation of FGFRs (Walsh and Doherty, 1997). An alternative, FGFR-independent, signaling pathway has been proposed that involves CAM-mediated activation of fyn kinase (Beggs et al., 1994, 1997; Panicker et al., 2003). Niethammer et al. (2002) unified these divergent views on the mechanism of NCAM-mediated neurite outgrowth by showing that distinct signaling pathways occur based on subcellular compartmentalization of NCAM-140 to lipid rafts or nonrafts in the plasma membrane. NCAM-140 in lipid rafts is proposed to act through initiation of the fyn–FAK kinase pathway, whereas NCAM-140 located in nonrafts acts via FGFR signaling. Both pathways are proposed to merge, with the ultimate activation of ERK1/2 (extracellular signal-regulated kinase 1/2) (Perron and Bixby, 1999; Niethammer et al., 2002). An essential feature of this proposed mechanism is that both pathways must be activated for NCAM-mediated neurite outgrowth to occur. With this in mind, we set out to test the hypothesis that β1 acts via a similar, dual pathway signaling mechanism.
β1-Mediated neurite outgrowth is inhibited in Fyn null mice
Fyn, a lipid raft-associated kinase (Wolven et al., 1997), is involved in NCAM-mediated neurite outgrowth (Beggs et al., 1994). To investigate a similar involvement of fyn kinase in the regulation of β1-mediated neurite outgrowth, acutely dissociated CGNs isolated from P14–P16 Fyn wild-type or null mice were grown on monolayers of control β1-expressing 1610 Chinese hamster lung (CHL) cells. β1 expression in the CHL monolayer increased the average neurite length of Fyn wild-type CGNs by 1.5-fold, from 48.7 ± 3.0 to 74.9 ± 4.3 μm (p < 0.01), whereas Fyn null CGNs showed no increase in average neurite length in response to β1 (Fig. 1 A). In addition, β1 increased the neurite length distribution for Fyn wild-type but not null CGNs (Fig. 1 B). This result is similar to that observed when Scn1b null neurons were grown on β1-transfected monolayers (Davis et al., 2004). Densitometric analysis of Western blots demonstrated that β1 protein expression was unchanged in Fyn null mice, suggesting that the lack of response of Fyn null CGNs to the β1-expressing monolayer was attributable to the loss of fyn kinase (p = 0.41) (Fig. 1 C).
β1-Mediated neurite outgrowth is impaired in acutely dissociated CGNs from Fyn null mice. A , Neurite lengths of CGNs isolated from Fyn wild-type and null mice grown on CHL or CHL-β1 monolayers. Data are presented as mean and SEM (n = 250). Significance: **p < 0.01, ANOVA with Tukey's post hoc test. B , Neurite distribution (in percentage) plotted against neurite length for Fyn wild-type and null CGNs grown on CHL or CHL-β1 monolayers. C , Typical Western blot of brain membrane protein prepared from Fyn wild-type and null mice, using anti-β1 antibody. Anti-α-tubulin antibody was used as a control for protein loading. D , Sucrose step gradient separation of protein extracted from wild-type mouse brain tissue. Blots were probed with anti-fyn kinase antibody, anti-β1 antibody, or anti-transferrin receptor (TfR) antibody. The numbers at top are gradient fractions of low-high density. Numbers at left are marker band sizes (kDa). E , Images of an acutely dissociated wild-type CGN labeled with anti-β1ex (green) and GAP-43 (red) antibodies, showing β1 expression at growth cone. The bottom panels are a 3× zoom of growth cone of CGN shown in top panels. Scale bars, 5 μm.
Both β1 and fyn kinase have been shown to localize to detergent-resistant lipid raft membrane fractions (Shenoy-Scaria et al., 1994; Kramer et al., 1999; Wong et al., 2005). To confirm that β1 and fyn kinase were indeed present in mouse brain lipid raft fractions in our hands, we isolated Triton X-100-insoluble microdomains from the brains of P14–P16 C57BL/6 mice on a sucrose step gradient. Western blot analysis revealed that fyn kinase was predominantly in low-density, Triton X-100-insoluble fractions (Fig. 1 D). β1 protein was present in both detergent-resistant fractions with fyn kinase, and in detergent-soluble fractions with the transferrin receptor (Fig. 1 D). Therefore, β1 may be differentially targeted within the same cell or may reside in different subcellular domains, depending on the particular neuronal or glial cell type within the brain. Immunocytochemical labeling of CGNs isolated from wild-type mice revealed that β1 subunits are localized to the growth cones (Fig. 1 E). Together with the Fyn null CGN neurite outgrowth results, these data suggest that the presence of fyn kinase plays a functional role in the mechanism underlying β1-mediated neurite outgrowth in CGNs and that this signaling pathway may be initiated in lipid rafts.
β1-Mediated neurite outgrowth is inhibited in Cntn null mice
Contactin (Cntn) is a lipid raft-associated, glycophosphatidylinositol (GPI)-linked IGSF-CAM that interacts with β1 in vitro (Kazarinova-Noyes et al., 2001) and in vivo (Chen et al., 2004). Contactin and fyn kinase have been shown to coimmunoprecipitate, although likely via indirect interactions (Zisch et al., 1995). To determine whether contactin is involved in the mechanism underlying β1-mediated neurite outgrowth, acutely dissociated CGNs isolated from P10–P12 Cntn wild-type or null mice were grown on monolayers of control or stable, β1-expressing CHL cells, as described previously (Davis et al., 2004). β1 expression in the CHL monolayer increased the average neurite length of Cntn wild-type CGNs by 2.3-fold, from 16.9 ± 1.4 to 38.9 ± 2.9 μm, consistent with previous findings (p < 0.001) (Fig. 2 A, left-hand bars) (Davis et al., 2004). In contrast, Cntn null CGNs grown on β1-transfected monolayers showed no increase in average neurite length compared with control CHL monolayers (Fig. 2 A, right-hand bars).
β1-Mediated neurite outgrowth is impaired in acutely dissociated CGNs from Cntn null mice. A , Neurite lengths of CGNs isolated from Cntn wild-type and null mice grown on CHL or CHL-β1 monolayers. Data are presented as mean and SEM (n = 200). Significance: ***p < 0.001, ANOVA with Tukey's post hoc test. B , Neurite distribution (in percentage) plotted against neurite length for Cntn wild-type and null CGNs grown on CHL or CHL-β1 monolayers. C , Typical Western blot of brain membrane protein prepared from Cntn wild-type and null mice, using an anti-β1 antibody. Anti-α-tubulin antibody was used as a control for protein loading.
The entire neurite length distribution was increased for Cntn wild-type CGNs grown on β1-expressing monolayers, whereas the distribution for Cntn null CGNs remained unchanged (Fig. 2 B). Densitometric analysis of Western blots revealed that the β1 protein level was reduced by 73% in brain homogenates of Cntn null mice (p < 0.001) (Fig. 2 C). The lack of response of Cntn null CGNs to the β1-expressing monolayer thus may result from the observed decrease in the level of β1 expression, or alternatively, contactin may play a novel functional role in the signal transduction mechanism underlying β1-mediated neurite outgrowth in CGNs.
Inhibition of FGFR and EGFR signaling does not affect β1-mediated neurite outgrowth
In hippocampal neurons, FGFR signaling is required for NCAM-dependent neurite outgrowth (Niethammer et al., 2002). To determine whether FGFR signaling is required for β1-mediated neurite outgrowth as well, CGNs were acutely dissociated from P14 C57BL/6 mice and grown on monolayers of control or stable, β1-expressing CHL cells in the presence of the FGFR tyrosine kinase inhibitor PD173074 (50 nm) for 48 h. When CGNs were grown on control CHL cell monolayers, incubation with PD173074 completely abrogated the potentiating effect of 20 ng/ml FGF on neurite outgrowth, thus confirming the activity of PD173074 (Fig. 3 A). In the presence of carrier (DMSO) alone, β1 expression in the CHL monolayer caused an increase in average CGN neurite length of 1.8-fold, from 54.5 ± 3.3 to 95.7 ± 4.7 μm, as before (p < 0.001) (Fig. 3 B, left-hand bars). Addition of PD173074 had no effect on neurite outgrowth, which still increased 1.5-fold, from 62.5 ± 2.9 to 94.2 ± 4.9 μm when β1 was stably expressed by the monolayer cells (p < 0.001) (Fig. 3 B, right-hand bars).
β1-Mediated neurite outgrowth in acutely dissociated CGNs is not affected by inhibition of EGF or FGF signaling. A , Neurite length of wild-type C57BL/6 CGNs grown on CHL monolayers and treated with FGF (20 ng/ml) and/or PD173074 (50 nm) for 48 h (n = 200). B , Neurite length of wild-type C57BL/6 CGNs grown on CHL or CHL-β1 monolayers and treated with/without PD173074 (50 nm) for 48 h (n = 200). C , Dose-dependent inhibition of proliferation of N/TERT keratinocytes by AG1478 (2 nm to 20 μm). The line represents fit to sigmoidal function (n = 16). D , Neurite length of wild-type C57BL/6 CGNs grown on CHL or CHL-β1 monolayers and treated with/without AG1478 (2 μm) for 48 h (n = 250). Data are presented as mean and SEM. Significance: ***p < 0.001, ANOVA with Tukey's post hoc test.
Given that EGF has also been shown to regulate neurite outgrowth (Martinez and Gomes, 2002; Povlsen et al., 2008) and that L1-CAM-mediated cell adhesion activates EGFR tyrosine kinase (Islam et al., 2004), we next tested whether EGFR signaling is involved in β1-dependent neurite outgrowth, using the EGFR inhibitor AG1478. In a control experiment, the activity of AG1478 was confirmed by assessing its dose-dependent effect on the proliferation of N/TERT human keratinocytes. N/TERT proliferation was inhibited by 70% in the presence of 2 μm AG1478 (Fig. 3 C), the maximum concentration at which there was no significant effect on CGN viability (data not shown). CGNs were acutely dissociated from P14 wild-type mice and grown on monolayers of control or stable, β1-expressing CHL cells in the presence of 2 μm AG1478 for 48 h. In the presence of carrier (DMSO) alone, β1 expression in the CHL monolayer caused an increase in average CGN neurite length of 1.6-fold, from 65.3 ± 3.8 to 106.4 ± 5.1 μm (p < 0.001) (Fig. 3 D, left-hand bars). Addition of AG1478 had no effect on neurite outgrowth, which still increased 1.6-fold, from 63.6 ± 3.8 to 102.1 ± 5.5 μm when β1 was expressed by the monolayer cells (p < 0.001) (Fig. 3 D, right-hand bars). These data show that neither FGFR activity nor EGFR activity are required for β1-induced neurite extension in CGNs.
The Scn1b null mutation results in developmental defects in the cerebellum
Scn1b mice are ataxic (Chen et al., 2004), similar to Cntn null mice (Berglund et al., 1999) and consistent with the possibility that defects in cerebellar microorganization may have occurred as a result of the null mutation. The mouse cerebellum undergoes dramatic developmental changes during the first three postnatal weeks (Goldowitz and Hamre, 1998). In the adult, CGNs exhibit a bipolar morphology with T-shaped axons. During postnatal development, CGNs extend single axonal processes (the parallel fibers) into the deep EGL parallel to the pial surface and groups of parallel fibers subsequently become compacted into fascicles. After parallel fiber extension, CGNs attach to Bergmann glial fibers, extend a single process perpendicular to the parallel fibers, and migrate through the molecular layer (ML) to the inner granular layer (IGL) (Ramon y Cajal, 1995; Stottmann and Rivas, 1998; Berglund et al., 1999). Bergmann glia provide migrating CGNs with a substrate that is rich in CAMs, including VGSC β1, thus suggesting that β1 may play a role in outgrowth and migration of CGNs along these fibers via homophilic cell adhesion (Davis et al., 2004). To test whether β1 functions in CGN migration in vivo, we examined the cerebella of P14 Scn1b wild-type and null littermates. Cerebellar sections from Scn1b wild-type and null P14 littermates were labeled with an antibody to the neuronal CAM, TAG-1. TAG-1 expression is transient and restricted to CGNs during the initial stages of axon outgrowth in the EGL between P1 and P21 in rodents (Stottmann and Rivas, 1998). TAG-1 immunoreactivity has been demonstrated to be strongest on CGN cell bodies and parallel fibers and weak on radially migrating CGNs (Yamamoto et al., 1986; Furley et al., 1990; Yamamoto et al., 1990; Bailly et al., 1996). Sections were imaged first at 10×, and then three representative locations were studied at 100× (Fig. 4 A). In both Scn1b wild-type and null mice, there was strong labeling of parallel fibers in the deep EGL (Fig. 4 Ai,ii). In Scn1b wild-type mice, TAG-1-labeled parallel fibers formed tightly compacted fascicles that ran the length of the EGL, close to the interface with the ML (Fig. 4 B–D, left-hand panels), consistent with the previously reported distribution of TAG-1 labeling in wild-type rat cerebellum (Yamasaki et al., 2001). In addition, in Scn1b wild-type mice, TAG-1-labeled CGN axons that descended through the ML were mostly oriented ∼90° to the parallel fibers (Fig. 4 B–D, left-hand panels). In contrast, at equivalent locations in the cerebellum of Scn1b null mice, the parallel fiber fascicles were less compact than those in the wild-type and in places appeared to deviate from the longitudinal axis of the EGL (Fig. 4 B–D, right-hand panels). Although some axons descended normally through the ML, many deviated from this path. Furthermore, the TAG-1 labeling of axons descending through the ML appeared more disparate in the Scn1b null sections, suggesting that some axons may have turned lateral to the tissue section and exited the plane of view (Fig. 4 B–D, right-hand panels). These migration defects were observed in all three Scn1b null mice studied, compared with their respective wild-type littermates.
β1 modulates axonal migration in the postnatal developing cerebellum in vivo. A , Schematic outline of left side of cerebellum in the coronal plane. The black boxes labeled “B–D” show locations of high-magnification images in B–D . CENT6–9, Central lobe, lobules 6–9; COPY, copula pyramidis; PRM, paramedian lobule; sec, secondary fissure; prepyramidal fissure. Inset, Parasagittal diagram indicating rostrocaudal location of coronal section. i , ii , Right-hand plates, Low-magnification images (10×) of TAG-1 immunolabeling (green) on coronal cerebellar sections from Scn1b wild-type and Scn1b null P14 littermates, respectively. Scale bar, 100 μm. B–D , High-magnification (100×) Z-series projections of the same Scn1b wild-type ( i ) and Scn1b null ( ii ) cerebellar sections, at the locations defined in A . Scale bar, 20 μm. The schematic outline in A was adapted with permission (Hof et al., 2000). Three mice of each genotype were examined, with similar results.
To further study the effect of the Scn1b null mutation on the organization of the EGL, we performed BrdU and DAPI double-labeling (Fig. 5 A). The outer proliferating zone (PZ) of the EGL contained the majority of BrdU-positive cells, consistent with BrdU incorporation into DNA of proliferating CGN precursors recently in S-phase (Alcaraz et al., 2006). The thickness of the PZ, defined by the lateral extent of the BrdU-positive cell nuclei, was unchanged between Scn1b wild-type and null mice (Fig. 5 B, left-hand bars). In contrast, the thickness of the total EGL, measured as the lateral extent of the dense population of cell nuclei labeled with DAPI, was increased 1.4-fold in Scn1b null mice, from 18.7 ± 0.8 to 26.2 ± 0.9 μm (p < 0.001) (Fig. 5 B, right-hand bars). We found no change in the number of BrdU-positive cells per 60-μm-long section of EGL between Scn1b wild-type and null mice (Fig. 5 C, left-hand bars). However, the total number of DAPI-labeled cell nuclei per 60-μm-long section of EGL was increased 1.5-fold in Scn1b null mice, from 58.0 ± 2.5 to 85.1 ± 4.8 cells (p < 0.001) (Fig. 5 C, right-hand bars). These results are consistent with more cells remaining in the EGL of Scn1b null mice compared with wild-type, without a change, either in the density of cells within the EGL, or in the absolute number of BrdU-positive, recently divided cells. Thus, the increased EGL thickness in Scn1b null mice is likely attributable to defective migration, rather than to a change in proliferation. Together, these data suggest that, although Scn1b null CGNs appear to be capable of migration, their extension of parallel fibers and subsequent descent from the EGL to the IGL along the Bergman glia fibers may be disrupted and/or delayed.
The external germinal layer is thicker in Scn1b null mice compared with wild type. A , High-magnification (100×) Z-series projections of coronal cerebellar sections from Scn1b wild-type and null P14 littermates, labeled with anti-BrdU antibody (red) and DAPI (blue). Scale bar, 20 μm. B , Thickness (in micrometers) of BrdU-positive zone (left-hand bars), and total EGL (right-hand bars), for Scn1b wild-type and null P14 littermates (n = 90 measurements taken from 3 mice of each genotype). C , Number of BrdU-positive cells (left-hand bars), and total number of cells (right-hand bars) per 60 μm-long section of EGL, for Scn1b wild-type and null P14 littermates (n = 30 measurements taken from 3 mice of each genotype). Data are presented as mean and SEM. Significance: ***p < 0.001, ANOVA with Tukey's post hoc test.
The Scn1b null mutation causes pathfinding defects in the CST
Our results thus far have demonstrated that CGN axonal migration and fasciculation are defective in Scn1b null mice. To determine whether these effects of β1 are limited to the cerebellum or whether β1 plays a more widespread role in axonal pathfinding, we examined the CST of Scn1b wild-type and null mice. The CST was chosen for these experiments because it is the last of the developing major fiber tracts to enter the spinal cord, coinciding with a period of β1 expression (Sutkowski and Catterall, 1990; Stanfield, 1992; Sashihara et al., 1995). In Scn1b wild-type mice, sequential DiI-labeled coronal sections across the pyramidal decussation confirmed normal migration of corticospinal axons from the ventral pyramid of the caudal hindbrain, across the midline, to the dorsal column of the rostral spinal cord (Fig. 6 a–g). In contrast, in Scn1b null mice, we observed significant defasciculation of axons at the pyramidal decussation (Fig. 6 h–n). Some axons began to decussate, but then turned more ventral (Fig. 6 h–l, arrows). In addition, there was mislocalization of axons lateral to the dorsal column of the spinal cord entering into the ventral and dorsal medullary reticular nuclei (Fig. 6 i–k, arrowheads). Furthermore, some axons deviated from the dorsal column after the pyramidal decussation (Fig. 6 m,n, arrowheads). These defects showed complete penetrance, because they were observed in all six Scn1b null mice studied, compared with their absence in seven wild-type littermates (additional examples from Scn1b null mice are in Fig. 6 o–r).
Loss of β1 results in axonal pathfinding abnormalities in the corticospinal tract. a–g , Consecutive coronal sections across the pyramidal decussation in a Scn1b wild-type brain. h–n , Consecutive coronal sections across the pyramidal decussation in a Scn1b null brain. o–r , Example sections from additional Scn1b null mice. h–l , o–r , Arrows, Defasciculation across pyramidal decussation. i–k , o–r , Arrowheads, Mislocalization of axons lateral to dorsal column. m , n , Arrowheads, Axons deviating from dorsal column after pyramidal decussation. v, Ventral pyramid; d, dorsal column. Scale bar, 250 μm. Six Scn1b null mice were examined. All six showed similar CST abnormalities compared with seven wild-type mice.
Together, our results in the cerebellum and in the CST demonstrate a role for β1 in axonal pathfinding and fasciculation in vivo. We propose that β1 subunits may affect the development of late-developing fiber tracts throughout the brain via cell-adhesive interactions.
Discussion
The overall conclusions of this study are that VGSC β1-mediated neurite outgrowth is regulated by a mechanism involving fyn kinase and that β1 functions as a CAM in vivo. The main findings are as follows: (1) β1-mediated CGN neurite outgrowth is completely inhibited in neurons from Fyn null mice without a change in β1 protein level. (2) β1 protein level is significantly reduced, and β1-mediated CGN neurite outgrowth is abrogated in neurons from Cntn null mice. (3) In contrast to L1-family CAMs, the mechanism underlying β1-mediated neurite outgrowth does not involve signaling through EGFR or FGFR. (4) The absence of β1 results in disruption of CGN axonal pathfinding and fasciculation in the postnatal cerebellum in vivo. In addition, axonal pathfinding errors and decreased fasciculation were observed in the Scn1b null CST, suggesting that β1 may play widespread roles in the formation of late-developing neuronal tracts in vivo. The results of this research are novel and significant, both to the basic understanding of the dual roles of VGSC β subunits as channel modulators and as IGSF-CAMs and to the understanding of the role of β1 in human brain disease.
A lipid raft signaling mechanism for β1-mediated neurite outgrowth
Two principal signaling pathways underlying the mechanism of CAM-mediated neurite outgrowth have been identified: (1) activation of fyn kinase in lipid rafts and (2) activation of the FGFR in nonraft domains (Maness and Schachner, 2007; Ditlevsen et al., 2008). Interestingly, NCAM-140-dependent neurite outgrowth is dependent on cosignaling from both compartments (Niethammer et al., 2002). In the present study, we found that β1-mediated neurite outgrowth was completely inhibited in CGNs from Fyn null mice, indicating that fyn kinase is an essential component. β1 protein levels remained unchanged in Fyn null brains, confirming that abrogation of β1-mediated neurite outgrowth in the absence of fyn was not as a result of decreased β1 expression. The notion of fyn kinase as a signaling intermediate in CAM-mediated neurite outgrowth promotion is well established. For example, neurite outgrowth in response to NCAM has been is selectively inhibited in CGNs from Fyn null mice (Beggs et al., 1994).
β1 protein expression levels were reduced by 73% in Cntn null brain. Contactin expression may enhance β1 transcription and/or protein stability by unknown mechanism(s). In agreement with the latter, contactin binds to the β1 Ig loop and enhances steady-state cell surface expression of VGSC α/β subunit complexes (Kazarinova-Noyes et al., 2001; McEwen et al., 2004). Interestingly, Cntn null mice share some of the characteristics of Scn1b null mice, including ataxia, defective CGN axonal guidance, and survival only to P18 (Berglund et al., 1999). It is possible that the reduced β1 expression in Cntn null mice may contribute to this phenotype.
β1-Mediated neurite outgrowth was completely blocked in CGNs from Cntn null mice. The large reduction in β1 protein expression in the Cntn null may account for this abrogation. In addition, contactin may also serve alongside fyn kinase as a component of the signaling mechanism underlying β1-mediated neurite outgrowth. Previous coimmunoprecipitation studies have shown fyn and GPI-linked contactin association, presumably indirectly, via an unknown membrane-spanning molecule (Zisch et al., 1995; Kramer et al., 1999). β1 and fyn coimmunoprecipitated in Triton X-100-solubilized brain membranes in our hands; however, this result was not consistently reproducible, suggesting that these two proteins may associate transiently or with low affinity (data not shown). Binding of contactin ligands result in fyn recruitment (Zisch et al., 1995), and, at least in oligodendrocytes and contactin-transfected CHO cells, the addition of anti-contactin antibodies results in stimulation of fyn kinase (Cervello et al., 1996; Kramer et al., 1999). In oligodendrocytes, contactin-mediated fyn activation occurs specifically in rafts (Kramer et al., 1999). In addition, both β1 and fyn have been shown to interact with protein tyrosine phosphatases in brain (Zeng et al., 1999; Ratcliffe et al., 2000), providing a potential dynamic transduction capability to this signaling complex. Our results suggest that β1 may serve as a key transmembrane signaling molecule linking extracellular contactin-related cell adhesive interactions to intracellular fyn kinase, regulation of tyrosine phosphorylation, and neurite outgrowth.
In agreement with previous findings, we found that β1 and fyn were both present in detergent-resistant membrane fractions (Shenoy-Scaria et al., 1994; Kramer et al., 1999; Wong et al., 2005). Consistent with these observations, β1 contains a putative palmitoylation site juxtaposed to the inner membrane leaflet (McEwen et al., 2004), a modification commonly found in lipid raft-associated proteins, including myelin Po, an IGSF-CAM that is closely related to β1 (McCormick et al., 1998; Pike, 2004). In addition, β1 interacts with multiple raft-associated proteins, including NrCAM, neurofascin-155, and contactin (Kazarinova-Noyes et al., 2001; McEwen and Isom, 2004; Schafer et al., 2004). Finally, β1 is modified by the raft-associated proteins γ-secretase and BACE1 (Vetrivel et al., 2005; Wong et al., 2005). Together, these findings suggest that localization of β1 to lipid rafts may be a key component in the regulation of β1-mediated neurite outgrowth (Davis et al., 2004) and that mutations affecting this localization may result in detrimental neurological effects in vivo.
Pharmacological blockade of either EGFR or FGFR signaling did not affect β1-mediated neurite outgrowth. We therefore conclude that the mechanism underlying β1-mediated neurite outgrowth is lipid raft-associated and requires fyn kinase. In contrast, activation of (nonraft) FGFR- or EGFR-mediated signaling pathway(s) is not involved. Based on our data, we propose that β1 expressed on the monolayer, or on an opposing neuronal or glial cell in brain, initiates a signaling cascade in the CGN via trans homophilic adhesion with neuronal β1 subunits, leading to neurite outgrowth (Fig. 7). The simplest scenario would involve a cis-β1-contactin multiprotein signaling complex providing a novel trans binding site(s) for β1 expressed on adjacent cells, resulting in the subsequent initiation of the fyn kinase pathway and downstream signaling cascades. β1-β1 trans homophilic adhesion would activate contactin via cis-heterophilic adhesion. In support of this, the fibronectin III domain of contactin has previously been shown to interact heterophilically with β1 (Kazarinova-Noyes et al., 2001; McEwen et al., 2004). In addition, in mouse cerebellum, contactin associates with, and transduces signaling through fyn kinase (Olive et al., 1995; Cervello et al., 1996). Alternatively, or in addition, trans heterophilic adhesion may occur between β1 expressed by the adjacent cell and contactin expressed by the neuron, with neuronal β1 acting as the intermediary between contactin and fyn activation. In agreement with this, β1 contains a conserved phosphotyrosine motif (Y181) in its C terminus, which is predicted to be capable of interaction with nonreceptor tyrosine kinases, such as src, fyn, or yes (Malhotra et al., 2004). Substitution of a glutamate residue, mimicking phosphorylation of Y181, abrogates recruitment of ankyrin by β1 in vitro (Malhotra et al., 2002; Nagaraj and Hortsch, 2006). Finally, the model is complicated by the observation that contactin may additionally affect β1-mediated neurite outgrowth indirectly, via regulating levels of surface β1 expression. Future studies will be necessary to elucidate these possibilities, and to evaluate the possible involvement of the VGSC α-subunits in the signaling mechanism.
Schematic representation of a possible signaling mechanism underlying β1-mediated neurite outgrowth in CGNs. β1 from an adjacent cell interacts with a multiprotein complex of β1 and contactin in the CGN, initiating a signaling cascade through fyn leading to neurite outgrowth. Homophilic adhesion between β1 expressed by an adjacent cell and β1 in the CGN, and/or heterophilic adhesion between β1 expressed by an adjacent cell and contactin in the CGN may occur. Ig, Immunoglobulin domain; FN III, fibronectin type III domain. This figure was produced using Science Slides 2006 software.
β1 is essential for normal CNS development
During the first 2–3 weeks of postnatal development, the CNS undergoes a period of neuronal migration and formation of synaptic connections (Singh, 1977; Amaral and Dent, 1981; Callaway and Katz, 1990). VGSC β1 subunits are expressed within this critical developmental period. For example, in rats, β1 mRNA is expressed in the hippocampus and spinal cord as early as P2, and β1 protein is expressed in the forebrain as early as P1 (Sutkowski and Catterall, 1990; Sashihara et al., 1995). In the cerebellum, β1-labeled cells in the EGL and the deep cerebellar nuclei are detected from P1 (Sashihara et al., 1995). Postmitotic CGNs leave the EGL and descend into the IGL during this postnatal period of β1 expression (Altman, 1972). In the present study, we found that β1 protein was localized to the growth cones of CGNs grown in vitro. In addition, in the cerebella of Scn1b null mice, there was disorganization of the parallel fibers as well as abnormal CGN axonal patterning through the ML into the IGL. Similar to the present data, Cntn null mice display misorientation of the parallel fibers with a slight reduction in their compaction within fascicles (Berglund et al., 1999). In contrast, in L1cam or Nrcam null mice, there was a reduction in the thickness of both the EGL and the IGL, without any alteration in the pattern of TAG-1 expression, suggesting that, unlike β1, these adhesion molecules may be important for CGN proliferation and/or cell survival, in addition to migration (Sakurai et al., 2001).
In the CST of Scn1b null mice, we observed reduced fasciculation at the pyramidal decussation, and axons projected laterally from the dorsal column, beyond the pyramidal decussation. As in the cerebellum, these effects of β1 on CST axonal pathfinding appear to be different from those of the L1-family CAMs. For example, in both L1cam and Ncam1 null mice, many axons fail to cross the midline at the pyramidal decussation, but instead project ipsilaterally, resulting in a reduced number of CST axons at the opposite dorsal columns of the spinal cord (Cohen et al., 1997; Rolf et al., 2002). Thus, these L1-family CAMs are proposed to serve as guidance cues for developing axons. In contrast to L1-CAM and NCAM, β1 appears to modulate fasciculation between CST axons rather than play a role in growth cone guidance at the point of decussation, because all of the CST axons were observed to cross the midline at the appropriate point in Scn1b null mice. Together, these findings suggest that, as a CAM, β1 is important for normal development of the CNS.
In conclusion, we have shown that β1 plays a critical role in axon pathfinding and fasciculation during CNS development via a mechanism involving cell–cell adhesion and that is dependent on the presence of the lipid raft-associated kinase fyn and the IGSF-CAM contactin. We propose that, as a CAM, β1 participates in intercellular and intracellular communication at specialized neuronal subcellular domains, including axonal growth cones and points of axonal fasciculation. β1 subunits are multifunctional, participating in signal transduction, first on a millisecond timescale by modulating VGSC activity and, second, on a more prolonged timescale that involves cell adhesion, neurite outgrowth, and neuronal migration.
Footnotes
-
This work was supported by National Institutes of Health (NIH) Grant R01MH059980, a grant from the Wilson Medical Research Foundation, and National Multiple Sclerosis Society Grant RG2882 (L.L.I.); by NIH Grant R01NS38297 and National Multiple Sclerosis Society Grant RG3567 (B.R.); by a University of Michigan Center for Organogenesis Non-Traditional Postdoctoral Fellowship (W.J.B.); and by NIH Grant F31NS43062 (T.H.D.). We acknowledge the expert technical assistance of Audrey Speelman and thank Dr. J. T. Elder for providing keratinocytes and Dr. K. Sue O'Shea for helpful discussions.
- Correspondence should be addressed to Dr. Lori L. Isom, Department of Pharmacology, 1150 West Medical Center Drive, 1301 MSRB III, Ann Arbor, MI 48109-0632. lisom{at}umich.edu
References
- Alcaraz et al., 2006.↵
- Altman, 1972.↵
- Amaral and Dent, 1981.↵
- Audenaert et al., 2003.↵
- Bailly et al., 1996.↵
- Beggs et al., 1994.↵
- Beggs et al., 1997.↵
- Berglund et al., 1999.↵
- Callaway and Katz, 1990.↵
- Catterall, 2000.↵
- Cervello et al., 1996.↵
- Chen et al., 2004.↵
- Cohen et al., 1997.↵
- Coonan et al., 2001.↵
- Danneman and Mandrell, 1997.↵
- Davis et al., 2004.↵
- Dickson et al., 2000.↵
- Ditlevsen et al., 2008.↵
- Dolmetsch, 2003.↵
- Furley et al., 1990.↵
- Goldowitz and Hamre, 1998.↵
- Hegle et al., 2006.↵
- Hof et al., 2000.↵
- Ignelzi et al., 1994.↵
- Islam et al., 2004.↵
- Isom, 2002.↵
- Isom and Catterall, 1996.↵
- Isom et al., 1992.↵
- Isom et al., 1994.↵
- Kaczmarek, 2006.↵
- Kasahara et al., 2002.↵
- Kazarinova-Noyes et al., 2001.↵
- Kolkova et al., 2000.↵
- Kramer et al., 1999.↵
- Levitan, 2006.↵
- Liu et al., 1999.↵
- Lopez-Santiago et al., 2006.↵
- MacLean et al., 2003.↵
- MacLean et al., 2005.↵
- Malhotra et al., 2000.↵
- Malhotra et al., 2002.↵
- Malhotra et al., 2004.↵
- Maness and Schachner, 2007.↵
- Martinez and Gomes, 2002.↵
- McCormick et al., 1998.↵
- McEwen and Isom, 2004.↵
- McEwen et al., 2004.↵
- Meadows and Isom, 2005.↵
- Meadows et al., 2002.↵
- Nagaraj and Hortsch, 2006.↵
- Niethammer et al., 2002.↵
- Olive et al., 1995.↵
- Panicker et al., 2003.↵
- Perron and Bixby, 1999.↵
- Pike, 2004.↵
- Povlsen et al., 2008.↵
- Ramon y Cajal, 1995.↵
- Ratcliffe et al., 2000.↵
- Rolf et al., 2002.↵
- Sakurai et al., 2001.↵
- Sanchez-Heras et al., 2006.↵
- Sashihara et al., 1995.↵
- Schafer et al., 2004.↵
- Scheffer et al., 2007.↵
- Shenoy-Scaria et al., 1994.↵
- Simons and Toomre, 2000.↵
- Singh, 1977.↵
- Srinivasan et al., 1998.↵
- Stanfield, 1992.↵
- Stottmann and Rivas, 1998.↵
- Sutkowski and Catterall, 1990.↵
- Vaillant et al., 2003.↵
- Vetrivel et al., 2005.↵
- Wallace et al., 1998.↵
- Wallace et al., 2002.↵
- Walsh and Doherty, 1997.↵
- Wolven et al., 1997.↵
- Wong et al., 2005.↵
- Xiao et al., 1999.↵
- Yamamoto et al., 1986.↵
- Yamamoto et al., 1990.↵
- Yamasaki et al., 2001.↵
- Zeng et al., 1999.↵
- Zisch et al., 1995.↵
