Abnormal neuronal patterning occurs during early postnatal brain development of Scn1b-null mice and precedes hyperexcitability

Abstract

Voltage-gated Na⁺ channel (VGSC) β1 subunits, encoded by SCN1B, are multifunctional channel modulators and cell adhesion molecules (CAMs). Mutations in SCN1B are associated with the genetic epilepsy with febrile seizures plus (GEFS+) spectrum disorders in humans, and Scn1b-null mice display severe spontaneous seizures and ataxia from postnatal day (P)10. The goal of this study was to determine changes in neuronal pathfinding during early postnatal brain development of Scn1b-null mice to test the hypothesis that these CAM-mediated roles of Scn1b may contribute to the development of hyperexcitability. c-Fos, a protein induced in response to seizure activity, was up-regulated in the Scn1b-null brain at P16 but not at P5. Consistent with this, epileptiform activity was observed in hippocampal and cortical slices prepared from the P16 but not from the P5–P7 Scn1b-null brain. On the basis of these results, we investigated neuronal pathfinding at P5. We observed disrupted fasciculation of parallel fibers in the P5 null cerebellum. Further, P5 null mice showed reduced neuron density in the dentate gyrus granule cell layer, increased proliferation of granule cell precursors in the hilus, and defective axonal extension and misorientation of somata and processes of inhibitory neurons in the dentate gyrus and CA1. Thus, Scn1b is critical for neuronal proliferation, migration, and pathfinding during the critical postnatal period of brain development. We propose that defective neuronal proliferation, migration, and pathfinding in response to Scn1b deletion may contribute to the development of hyperexcitability.

Neuronal voltage-gated Na⁺ channels (VGSCs) are composed of one pore-forming α and two β subunits (1). β1 (encoded by SCN1B) modulates channel gating and cell surface expression (2). In addition, β1 is an Ig superfamily cell adhesion molecule (CAM) that participates in cell–cell and cell–matrix adhesion (3). The SCN1B splice variant β1B is a secreted CAM that is expressed predominantly during embryonic brain development (4). SCN1B mutations are associated with the spectrum of genetic epilepsy with febrile seizures plus (GEFS+) (OMIM 604233) epilepsies (5, 6). At the cellular level, SCN1B GEFS+ mutations cause a range of defects, including altered channel availability, disrupted adhesion (7), exclusion of β1 from the axon initial segment (AIS) of pyramidal neurons (8), reduced cell surface expression of β1, and intracellular retention of β1B (4, 9). In addition, a GEFS+ mutation in SCN1A decreases modulation of Na_v1.1 by β1 (10). Together, these results suggest a causal relationship between VGSC α–β1 interactions, cell–cell adhesion, and epilepsy.

Scn1b-null mice display severe spontaneous seizures and ataxia from approximately postnatal day (P)10 (11). Scn1b deletion reduces resurgent Na⁺ current (I_Na) in P14–P16 cerebellar granule neurons (CGNs) (12). In contrast, β1 has no effect on I_Na in dissociated P14–P16 hippocampal neurons (9). However, CA3 action potentials have higher peak voltage and amplitude, suggesting that β1/β1B may regulate VGSCs on neuronal processes rather than on the soma. VGSC expression and localization are altered in Scn1b-null hippocampus (11) and cerebellar AIS (13). P14 Scn1b-null mice exhibit aberrant neuronal pathfinding and defasciculation in the corticospinal tract and cerebellum (12). In zebrafish, scn1bb morphants show abnormal spinal neuron axons and olfactory nerve defasciculation (14). β1-mediated neurite outgrowth and β1-mediated VGSC gating reciprocally modulate each other (13). Together, these results suggest that Scn1b is required for the establishment of normal electrical activity in the brain.

Our aim here was to assess the effects of Scn1b on neuronal pathfinding at developmental time points earlier than P14. We show that Scn1b deletion alters the proliferation and localization of dentate granule neurons (DGCs) and results in axonal fasciculation and pathfinding abnormalities in the hippocampus and cerebellum at P5, before c-fos activation. Spontaneous epileptic activity in brain slices is observed at P16 but not at P5–P7. We propose that Scn1b deletion results in defective neuronal proliferation and pathfinding, which, in turn, may lead to altered synaptic function, neuronal excitability, seizures, and ataxia.

Results

Scn1b-null mice are ataxic and display spontaneous behavioral seizures beginning at ∼P10 (11). Fasciculation and pathfinding of corticospinal axons and CGNs is disrupted in null mice when assayed at P14 (12). In addition, the excitability of null CGNs is impaired as a result of reduced Na_v1.6 AIS expression (13). Similar to ref. 15, we hypothesized that defective neuronal migration in Scn1b-null mice may contribute to the development of abnormal synaptic circuits and altered neuronal excitability rather than result from it. Thus, we investigated neuronal patterning in Scn1b-null brains at P5, before the onset of behavioral seizures and ataxia.

Scn1b-Null Mice Do Not Exhibit c-fos Activation at P5.

Since establishing this mouse model in 2004, we have not observed spontaneous behavioral seizures in Scn1b-null mice before P10 (11). To confirm the absence of seizure activity in P5 null mice, we studied the expression of c-Fos protein in hippocampus and cortex. Seizure activity induces expression of c-fos within hours (16). In P16 null mice, there was strong labeling of c-Fos colocalizing with DAPI in the upper and lower dentate granule cell layer (GCL) of the DG in the hilus and the CA1 pyramidal layer (Fig. 1A). We observed no c-Fos immunoreactivity in the hippocampus of P5 WT mice (Fig. 1B) and only a few c-Fos–positive (+) cells in the hippocampus of P16 WT mice (Fig. S1). Importantly, there was no c-Fos expression in P5 Scn1b-null mice (Fig. 1C). In the cortex of P16 null mice, strong c-Fos+ nuclei were present in multiple layers (Fig. S2 A and B). In P16 WT mice, few c-Fos+ nuclei were observed (Fig. S2 C and D). No c-Fos+ nuclei were detected in either null or WT cortex at P5 (Fig. S2 E–H). These data confirm our observations that seizure activity in Scn1b-null mice, which causes c-Fos up-regulation in hippocampus and cortex by P16, has not yet started at P5. To confirm that c-Fos up-regulation consequent to seizure activity could be detected at P5, we induced seizures in P5 and adult WT mice using pentylenetetrazol (PTZ) and assessed c-Fos expression 4 h later, a time point at which c-Fos expression reaches maximal levels (17). PTZ induced strong expression of c-Fos in both P5 and adult cortex (Fig. S3 A–D) compared with saline controls (Fig. S3 E–H).

Fig. 1.

c-Fos is expressed in Scn1b-null hippocampus at P16 but not at P5. Coronal hippocampal sections from (A) P16 null, (B) P5 WT, or (C) P5 null mice were stained with anti-cFos (green), and nuclei were counterstained with DAPI (blue). Right column of A contains high-magnification images of regions highlighted by white boxes in Left column.

Scn1b Regulates Neuronal Migration in P5 Cerebellum.

Scn1b-null mice have defective cerebellar microorganization at P14 (12). Results showing that P5 Scn1b-null CGNs display defective extension and fasciculation of parallel fibers and their descent to the IGL is disrupted are presented in Figs. S4 and S5.

Scn1b-Null Brains Show Epileptiform Activity at P16 but Not at P5.

We performed extracellular field potential recordings to detect epileptiform activity in P16 null or WT neocortical and hippocampal slices. Epileptiform activity, defined by the observation of multiple population spikes in response to a single submaximal stimulation, was consistently observed in null (Fig. 2 B and D) but not WT slices (Fig. 2 A and C) and was more consistently observed in null neocortex, than hippocampus (Fig. 2F). Within the hippocampus, CA1 and CA3 presented more epileptiform activity than did the DG. In addition, spontaneous bursting and giant depolarization potentials were often observed in CA3 (Fig. 2E). These results suggest that the Scn1b deletion induces region-selective changes in synaptic transmission and neuronal excitability.

Fig. 2.

Epileptiform activities in P16 Scn1b-null neocortical and hippocampal slices. (A and B) Representative fEPSP in layer II/III evoked by single-pulse stimulation of layer IV of a neocortical slice prepared from a WT (A) or null (B) mouse. Following the stimulation artifact (S) are two downward components. The first is a population spike (PS) corresponding to compound action potentials activated by antidromic stimulation of axons of layer II/III neurons (Anti-PS). The second is a glutamatergic fEPSP that is mediated by orthodromic stimulation of presynaptic fibers. (C and D) Representative PSs recorded in CA1 by single-pulse stimulation of the Schaffer collaterals from a (C) WT or (D) null mouse. (E) Representative spontaneous bursting activities (Upper) recorded in null CA3. One bursting response is shown in an expanded scale (Lower). Slices prepared from null mice (B, D, and E) showed multiple spikes or epileptiform activity. Each representative trace (A–D) is the average of 12–15 subsequent traces recorded in a given session under the same conditions. Epileptiform responses in different brain regions are summarized in F with n’s given in parentheses. Each trace is representative of 6–10 individual experiments.

Because epileptogenesis is often associated with changes in short-term synaptic plasticity (18), we examined paired-pulse stimulation (PPS) evoked synaptic responses in neocortical and hippocampal slices from each genotype. Although PPS induced similar paired-pulse facilitation (PPF) responses in the CA1, CA3, and DG of both WT and nulls (Fig. 3; Table S1), the ratio of population spikes (PSs) PS2/PS1 in CA1 and CA3 of null slices at an interstimulus interval (ISI) of 60 ms was larger than those of WT (P < 0.05). In addition, the forms of short-term synaptic plasticity in the neocortex were significantly different between P16 WT and null mice. In neocortical layer II/III of WT, PPS induced paired-pulse depression (PPD) (Fig. 3G). The mean ratio of field excitatory postsynaptic potential (fEPSP)2/fEPSP1 at a 60-ms ISI was 0.96 ± 0.31 (n = 8). In contrast, in the same regions of null slices, PPS induced PPF (Fig. 3H). The average fEPSP2/fEPSP1 ratio at a 60-ms ISI here was 1.21 ± 0.39 (n = 10, P < 0.05). Similar changes in the fEPSP2/fEPSP1 ratio were also observed at a 20-ms ISI (Table S1). Thus, Scn1b deletion appears to promote PPF rather than PPD. These data suggest that Scn1b plays an important role in the development of neuronal synaptic circuits and excitability in the CNS.

Fig. 3.

Representative paired-pulse evoked field potentials recorded in CA1, CA3, and dentate gyrus (DG) of hippocampal slices or layer II/III neurons of cortical slices prepared from P16 WT (Left) or Scn1b-null (Right) mice. PS responses in CA1 (A and B), CA3 (C and D), or DG (E and F) in hippocampal slices or fEPSPs in layer II/III of cortical slices (G and H) were evoked under similar conditions as in Fig. 2 but by two stimulus pulses with equal intensity and interpulse intervals (IPIs) varying at 20–200 ms. The second stimulus-evoked fEPSP (fEPSP2) was smaller than the first stimulus-evoked fEPSP (fEPSP1), although the two Anti-PSs were virtually identical WT neocortex (G). fEPSP2 was larger than fEPSP1 in null neocortex, although the two Anti-PSs were similar (H). Each trace is representative of 4–10 individual experiments.

To further confirm our observations of field potential recordings, we examined spontaneous excitatory postsynaptic currents (sEPSCs) in individual cortical layer II/III pyramidal cells using whole-cell patch-clamp recording. No significant differences in sEPSC frequency or amplitude were observed between genotypes in P5 cortical neurons (Fig. 4 A and B). sEPSC frequencies were 0.17 ± 0.02 Hz (n = 5) for P5 WT vs. 0.23 ± 0.06 Hz (n = 8) for P5 null (P > 0.05). sEPSC amplitudes were 8.87 ± 2.16 pA (n = 5) for P5 WT vs. 8.71 ± 1.52 pA (n = 8) for P5 null (P > 0.05). sEPSC frequencies in slices from P16 null mice (3.93 ± 1.58 Hz, n = 7) tended to be higher than those of P16 WT (1.25 ± 0.45 Hz, n = 5), although this difference was not significant (P = 0.268; Fig. 4 C and D). sEPSC amplitudes for P16 WT (10.26 ± 2.34 pA, n = 5) and P16 null slices (9.50 ± 1.26 pA, n = 7) were also similar, suggesting that Scn1b may not have a direct effect on postsynaptic glutamate receptors. Importantly, P16 null slices, but not WT slices, generated bursting activity and spontaneous giant slow inward currents (Fig. 4 E and F). Further, direct examination of the repetitive firing patterns of cortical and hippocampal neurons showed increased firing frequency in P16 null slices compared with P16 WT slices at lower level ranges of suprathreshold current injections, suggesting a reduced threshold for initiation of action potentials in null slices. In contrast, this difference was not observed in P5–P7 WT and null slices (Figs. S6–S8). Thus, these data suggest that Scn1b deletion-induced changes in synaptic transmission and neuronal excitability are age dependent and that P5 mice have not developed epileptiform-like activity. Moreover, our results suggest that the effects of Scn1b deletion on synaptic function may be presynaptic in origin because short-term synaptic plasticity is generally considered to be caused by changes in the probability for transmitter release from presynaptic terminals (for review, see ref. 19).

Fig. 4.

Scn1b-null brains show changes in spontaneous excitatory postsynaptic currents (sEPSCs) in cortical neurons at P16 but not at P5. sEPSCs were recorded from individual layer II/III pyramidal neurons in representative cortical slices from P5 WT (A), P5 null (B), P16 WT (C), or P16 null (D) using the whole-cell patch-clamp recording technique. A representative depiction of multiple spontaneous giant inward currents is shown in E, of which a single giant inward current is shown in an expanded scale (F) indicated by the asterisk (*). Each trace is a representative depiction of three to five individual mouse experiments.

Scn1b Regulates Neuronal Migration, Proliferation, and Pathfinding in P5 Hippocampus.

We studied the location and density of neurons in the hippocampus by comparing Nissl staining of P5 WT and null coronal sections. The gross distribution and layering of DGCs were similar in both (Fig. 5A). No differences were observed between the distribution and layering of WT and null pyramidal neurons in the CA1 or CA3 fields (Fig. 5 B and C). However, the density of neurons in the GCL of the DG was reduced by 22% in nulls (P < 0.001; Fig. 5D, left bars). This reduction in density was accompanied by a 1.2-fold increase in the transverse thickness of the GCL (P < 0.001; Fig. 5E, left bars); however, there was no change in the DGC soma area (P = 0.80; n = 180). There was no change in the density of neurons, or layer thickness, in the pyramidal layer of the CA1 or CA3 fields (Fig. 5 D and E, center and right bars). Thus, Scn1b deletion causes region-specific DGC dispersion, which results in an increase in layer width and a reduction in neuron density in the DG GCL.

Fig. 5.

Neuronal density is reduced and transverse thickness is increased in the DG GCL of P5 null mice. Neurons in (A) DG, (B) CA1, or (C) CA3 of P5 WT (i) or null (ii) mice were visualized using Nissl staining. Insets in A are low-magnification images of same hippocampus. (D) Neuron density in the DG GCL and pyramidal cell layer (CA1; CA3). (E) Thickness of DG GCL, CA1, and CA3 pyramidal cell layer. Data are mean ± SEM; ***P < 0.001 (n = 120 for D; n = 45 for E). Sections were analyzed from three pairs of WT and null littermates.

We studied neurogenesis within the DG in more detail by labeling P5 hippocampal sections with an antibody to the proliferation marker Ki67, which accumulates in the nuclei of interphase cells (20). In both WT and null mice, there was strong labeling of Ki67 in the nuclei of cells in the upper and lower GCL and the hilus (Fig. 6 A, i and ii). The density of Ki67+ nuclei in the hilus increased 1.6-fold in nulls (P < 0.001; Fig. 6B, left bars). In contrast, the density of Ki67+ nuclei in the suprapyramidal and infrapyramidal GCL blades was unchanged between genotypes (Fig. 6B, center and right bars). These data suggest that cellular proliferation is increased in the hilus of Scn1b-null mice but not in the GCL.

Fig. 6.

Proliferation is increased in the hilus of P5 null mice. (A) Proliferating cells in coronal sections from P5 WT (i) or null (ii) mice were stained with an anti-Ki67 (green). Nuclei were counterstained with DAPI (blue). Right columns of i and ii contain high-magnification images of regions highlighted by white boxes in Left columns. G, dentate gyrus granule cell layer; H, hilus. (B) Ki67+ cell density in the hilus. Data are mean ± SEM; ***P < 0.001 (n ≥ 11). Sections were analyzed from three pairs of WT and null littermates.

To localize the distribution of newborn DGCs, we labeled P5 sections with an antibody to Prox1, which is highly expressed in the nuclei of newborn DGCs (21). Consistent with ref. 21, Prox1+ nuclei were mainly restricted to the suprapyramidal and infrapyramidal blades of the GCL of both WT and null hippocampus (Fig. 7 A, i and ii). Prox1+ nuclei were also present in WT and null hilus, but at a much lower density than in the GCL (Fig. 7 A, i and ii). Prox1+ hilar ectopic DGCs arise from progenitors following status epilepticus (SE) (22). Hilar Prox1+ nuclei increased by 1.6-fold in nulls (P < 0.01; Fig. 7B). These data suggest that Prox1+ hilar ectopic DGCs are increased in Scn1b nulls; however, in contrast to previous studies (22), this occurs before seizure onset, as assessed by c-fos activation. As expected, at P16, levels of hilar Prox1+ nuclei in WT mice decreased compared with P5 levels (Fig. S9 A, i). However, the level of hilar Prox1+ nuclei in Scn1b-null mice remained significantly higher than WT at this time point (2.4-fold, P < 0.0001; Fig. S9 A, ii and B).

Fig. 7.

Hilar ectopic DGCs are increased in P5 null mice. (A) DGCs in coronal sections from P5 WT (i) or null (ii) mice were stained with anti-Prox1 (green). Nuclei were counterstained with DAPI (blue). Right columns of i and ii contain high-magnification images of regions highlighted by white boxes in Left columns. G, dentate gyrus granule cell layer; H, hilus. (B) Prox1+ neuron density in the hilus. Data are mean ± SEM; **P < 0.01 (n = 12). Sections were analyzed from three pairs of WT and null littermates.

To determine whether hippocampal pathfinding defects occur in Scn1b-null brains, we labeled neuronal processes in P5 WT and null mice with anti–neurofilament-M (NFM). NFM is highly expressed on cell bodies and axons in the DG ML and CA3 regions of early postnatal rodent hippocampus (23, 24). In WT (Fig. 8 A–D), the highest expression of NFM was observed on the axons of cells in the ML, with considerably fewer NFM+ processes in the hilus and little detectable NFM immunoreactivity on cells in the GCL. There was a clear delineation between the strongly NFM+ ML and very weakly NFM+ GCL (Fig. 8 A–D). There were NFM+ transverse processes in the pyramidal layer of the CA1 field (Fig. 8E). In null hippocampus, we observed a dense population of NFM+ processes in the ML and little NFM immunoreactivity in the GCL (Fig. 8 F–I, K, and M). Importantly however, we observed several aberrations in the null brain. First, there was an increase in NFM+ processes and cell bodies in the hilus (Fig. 8 F–I, K, and M). Second, the delineation between the ML and GCL was disrupted, such that the NFM+ processes were less well fasciculated and crossed from the ML into the GCL (Fig. 8 G, H, K, and M). Third, a number of the NFM+ transverse processes in the pyramidal layer of the CA1 field appeared misoriented, and an increased number of NFM+ cell bodies were observed in this region (Fig. 8 J, L, and N). These defects showed complete penetrance and were observed in all null mice studied compared with their absence in WT littermates.

Fig. 8.

Scn1b deletion results in axonal pathfinding defects in the hippocampus. Neuronal processes in coronal sections from P5 WT (A–E) or null (F–N) mice were stained using anti-NFM (green). (A and F) Low-magnification (10×) images of WT or null hippocampus. (B–D) High-magnification (40×) images of ML, GCL, and hilus (H) of a WT hippocampus. (G–I) Comparative images from a null hippocampus. (E and J) 40× images of pyramidal layer (PY) and Stratum radiatum (SR) in WT and null CA1 fields. (K and M) additional examples of ML, GCL, and H staining from null mice. (L and N) Additional examples of pyramidal layer and SR staining from null mice. Arrows indicate staining of neuronal soma and processes that are misoriented in the H and SR. Three null mice were examined. All showed similar abnormalities compared with WTs.

Altered Microorganization of Hippocampal Inhibitory Neurons in P5 Scn1b-Null Mice.

Haploinsufficiency of Scn1a in GABAergic interneurons is proposed to be a causative factor in Dravet syndrome (DS) (25). Knock-in of a heterozygous loss of function DS mutation in Scn1a impaired action potential burst firing in parvalbumin (PV)+ inhibitory neurons, suggesting that altered excitability and/or connectivity within inhibitory circuits play important roles in epilepsy (26). Functional loss of SCN1B also results in DS (9), showing that it is critical in regulating neuronal excitability, although the involvement of inhibitory neuronal circuitry in this process is unclear. Reduced PPD and increased PPF are often associated with loss or reduction of GABAergic inhibition (27–29). Thus, the enhanced PPF responses shown here (Fig. 3) suggest that inhibitory activity may be impaired in some regions of the Scn1b-null brain. We investigated whether Scn1b deletion affected the migration of PV+ inhibitory neurons at P5. PV+ cells were observed in the DG GCL, the pyramidal layer, and Stratum radiatum of the CA3 field and the pyramidal layer and Stratum oriens of the CA1 field of both WT and null mice (Fig. 9 A–F), consistent with refs. 30, 31. PV+ somata in the WT DG GCL were tightly packed into a single layer (Fig. 9A). In contrast, the layering of these somata in the Scn1b-null DG GCL appeared to spread laterally (Fig. 9B, arrowheads). Thus, the density of PV+ somata touching a single line drawn along the center of the immunoreactive layer was reduced by 12% in nulls (P < 0.01; Fig. 9G). In the WT CA1 pyramidal layer, neuronal processes on PV+ cells projected laterally to the layer (Fig. 9D). In the Scn1b-null CA1 pyramidal layer, the majority of processes followed the same lateral projection (Fig. 9F). However, the processes on some cells deviated from this path (Fig. 9E, arrows). There were no aberrations in the distribution of PV+ cells in the CA3 of Scn1b null mice compared with WT (Fig. 9 C and E); however, further investigation of changes in GABAergic synaptic transmission will be required to confirm these immunohistochemical observations.

Fig. 9.

Microorganization of PV+ cells is disrupted in P5 null mice. Cells in coronal sections from P5 WT (A, C, and D) or null (B, E, and F) mice were stained using anti-PV (green). (A and B) Images of DG. (C and E) Images of CA3. (D and F) Images of CA1. GCL, granule cell layer; H, hilus; PY, pyramidal layer; SR, Stratum radiatum. Arrowheads in B indicate lateral expansion of PV+ cells in null GCL. Arrows in F indicate PV+ processes that are misoriented in the CA1. (G) Density of PV+ cells in the suprapyramidal GCL blade per 100 μm. Data are mean ± SEM; **P < 0.01 (n > 30). Sections were analyzed from three pairs of WT and null littermates.

Discussion

Scn1b Is Essential for Normal Hippocampal Development.

Hippocampal neurogenesis and formation of new synapses is a continual process (32). The birth of DGCs and formation of the suprapyramidal and infrapyramidal blades of the DG GCL coincide with the onset of β1 expression (33, 34). Here, we found that the density of neurons in the P5 DG GCL was reduced in the absence of Scn1b, before c-fos induction. This reduction was accompanied by increased thickness of the GCL, suggesting that Scn1b deletion causes region-specific DGC dispersion. In addition, cell proliferation was increased in the hilus, suggesting that Scn1b may regulate the proliferation of neuronal precursors in this region. In agreement with this, the density of Prox1+ hilar ectopic DGCs was increased in null mice. Hilar ectopic DGCs increase following status epilepticus (22). However, our data suggest that in Scn1b-null mice, Prox1+ hilar ectopic DGCs increase before behavioral seizure onset. Thus, Scn1b may play an important role in regulating the hilar progenitor cells and their subsequent migration to the GCL (32).

The possibility that Scn1b regulates DG GCL formation and that Scn1b deletion results in DGC dispersion is further supported by our observation of defective projection of NFM+ axonal processes at the interface between the GCL and the ML in P5 null mice. These mislocalized processes may belong to GABAergic interneurons, as we observed a lateral spreading of PV+ somata in the GCL. In addition, there was misorientation of NFM+ and PV+ processes in the CA1 pyramidal layer that was remarkably similar to the misorientation of CGN axons that we observed previously in the cerebellar ML of P14 null mice (12).

Together, our results suggest that Scn1b regulates neuronal migration in the DG and CA1 fields at P5. In addition, Scn1b regulates the proliferation of precursors in the dentate hilus. Interestingly, the latter is in contrast to the role of Scn1b in the cerebellum, where it does not regulate proliferation of CGN precursors in the EGL (12). Thus, Scn1b appears to play region- and/or neuron-specific roles in regulating migration and proliferation. We propose that β1/β1B regulate the development of these regions via cell–cell adhesion. Importantly, abnormal neuronal migration in the absence of β1/β1B-mediated cell adhesive interactions may lead to aberrant synaptic connections resulting in neuronal hyperexcitability. Scn1b-null mice may thus represent an example of a seizure model in which defective neuronal migration contributes to the development of hyperexcitability rather than results from it (15).

Critical Role for SCN1B in Epilepsy.

Mutations in SCN1B are associated with GEFS+ spectrum disorders (9). Seizure activity may in part be caused by impaired regulation of VGSC α subunit–dependent excitability by these mutant β subunits (7, 8, 10) or altered α subunit expression (11). Alternatively, as proposed here, seizure activity may be caused by impaired cell adhesive interactions that result in aberrant neuronal patterning. Scn1b helps to regulate neuronal fasciculation and migration in the cerebellum and hippocampus of mice and in zebrafish spinal neurons and olfactory nerve (refs. 12 and 14 and the present study). Disruption of the functional reciprocity between Scn1b-mediated migration and Scn1b-mediated VGSC gating may be an important factor in excitability-related disorders (13).

Aberrant migration of GABAergic interneurons in the P5 null hippocampus suggests that β1/β1B may be critical in regulating the excitability of hippocampal neuronal circuitry. Haploinsufficiency of Scn1a in hippocampal inhibitory interneurons is proposed to be a causative factor in DS, by reducing negative feedback regulation of hippocampal circuitry (25, 26). Given that Na_v1.1 expression and activity are regulated by Scn1b (10, 11, 13), our results suggest that there may be a causal relationship between VGSC α–β interactions, adhesion, and spontaneous seizure activity. Thus, mutations in Scn1b in inhibitory neurons may bias the brain toward hyperexcitability. We conclude that inherited epilepsies associated with mutations in SCN1B may result, at least in part, from abnormal electrical activity arising from aberrant neuronal proliferation, migration, pathfinding, and/or fasciculation in brain.

Materials and Methods

Animals.

Scn1b WT and null littermate mice were generated and maintained in accordance with the guidelines of the University of Michigan Committee on the Use and Care of Animals (11). Mice were bred from Scn1b heterozygous animals that had been repeatedly backcrossed to C57BL/6 mice for at least 18 generations, creating a congenic strain. Additional details regarding seizure induction are given in SI Materials and Methods.

Histochemistry.

Procedures for tissue preparation, antibody labeling, Nissl staining, and immunohistochemistry were described previously (13, 35). Additional details are given in SI Materials and Methods.

Electrophysiological Slice Recording.

Hippocampal and cortical slices were prepared, and electrophysiological recordings were performed and analyzed as described previously (36). Additional details are provided in SI Materials and Methods.

Data Analysis.

Data are presented as mean and SEM, unless stated otherwise. Pairwise statistical significance was determined with Student’s two-tailed unpaired t test or Mann-Whitney test as appropriate. Results were considered significant at P < 0.05.