INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

A long-held tenet considers that electrical activity is a crucial component in brain development and synaptogenesis (Katz and Shatz, 1996; Catalano and Shatz, 1998; Ikonomidou et al., 1999). Action potentials propagated along the axon evoke the release of neurotransmitters at the synaptic terminal and thereby trigger neuronal signaling. Voltage-gated sodium channels (NaChs) are responsible for the inward flow of sodium current across the neuronal membrane during the rising phase of an action potential (Armstrong and Hille, 1998). NaChs are heteromeric membrane proteins that consist of one  subunit that determines the permeation and gating properties of the channel, and two different  subunits that modulate channel kinetics and density (Catterall, 1995; Hartshorne et al., 1985). Four different  subunit genes encoding tetrodotoxin-sensitive NaChs have been identified in mammalian neurons: _I, _II, _III, and _VIII, also denoted SCN1A, SCN2A, SCN3A, SCN8A, in keeping with the systematic mammalian gene symbols (cf. Plummer and Meissler, 1999). NaCh_II is the most abundant in rostral regions of the central nervous system (CNS) (cortex, hippocampus, striatum, and midbrain) (Beckh et al., 1989; Felts et al., 1997; Ahmed et al., 1992; Schaller et al., 1995; Smith et al., 1998). The molecular heterogeneity of NaCh  subtypes implies functional differences. Indeed, the mouse NaCh SCN8A and its rat ortholog NaCh6 are abundant in the cerebellum, and SCN8A determines a persistent sodium conductance that plays a key role in the firing properties of Purkinje neurons (Schaller et al., 1995; Smith et al., 1998). And a loss-of-function point mutation of SCN1B, the gene encoding a NaCh 1 subunit, was identified in human generalized epilepsy with febrile seizures (Wallace et al., 1998). In addition, neural activity serves a crucial role during the development of the brain in the control of neuronal survival (Ikonomidou et al., 1999) and in synaptogenesis (Katz and Shatz, 1996; Catalano and Shatz, 1998). It is well recognized, especially in the visual system, that blockade of NaChs by tetrodotoxin alters drastically the fine patterning of neuronal connectivity (Katz and Shatz, 1996; Catalano and Shatz, 1998). And blockade of glutamate stimulation through NMDA receptors in the developing brain results in massive apoptotic neurodegeneration (Ikonomidou et al., 1999). To investigate the role in vivo of NaChs in neuronal apoptosis and synaptogenesis, and to develop an animal model to identify the function of specific NaChs, we used gene targeting (Capecchi, 1989) to generate knockout mice deficient in the brain sodium channel _II subunit (NaCh_II or SCN2A). Mice homozygous for the mutation (NaCh_II^/) die perinatally with severe hypoxia and extensive neuronal cell death, presumably resulting from the brainstem deficiency of the NaCh_II gene product. A preliminary account of this research was presented elsewhere (Planells-Cases et al., 1999).

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Generation of NaCh_II-deficient mice

All procedures involving mice conformed with institutional and National Institutes of Health guidelines. The targeting vector, Naneo, contains ~9.5 kb of mouse genomic DNA encompassing exon 1 of NaCh_II, which was interrupted with the neo^r gene (Fig. 1 A). Homologous recombination between the target locus and the targeting construct leads to a modified gene that now contains the positively selectable gene and truncates the NaCh_II protein after the seventh amino acid. To construct Naneo, 2 × 10⁵  bacteriophages containing genomic DNA from an OLA129 mouse strain library were screened using a human brain NaCh_II cDNA probe, (nt 248 to 605 in the cDNA sequence) (Ahmed et al., 1992). Two clones were isolated; restriction analysis, in combination with Southern blot and sequencing, identified two overlapping inserts, both containing the entire exon 1 of the NaCh_II gene. Na164/pGEM5Z was constructed in two steps. An 8-kb NcoI fragment, isolated from bacteriophage clone 2, was subcloned in pGEM5Z (Promega) and named Na16. In addition to exon 1, Na16 contained ~7.5 kb of homologous flanking intron sequence in the 5' end and ~500 nt at the 3' end. No other exons were identified by Southern blot analysis using as probes different cDNA fragments. To increase the extent of homology at the 3' end, an additional fragment of 2.4 kb, Na64, was subcloned at the EcoRV site of Na16 to generate Na164. Overlapping sequences between the 8- and 2.4-kb fragments were sequenced and confirmed. The clone Na164 was the basis for the targeting construct. A 325-nt KpnI-EcoRV fragment that contains all except the first seven amino acids of exon 1 was excised from Na164 and replaced by an EcoRV-HindIII insert containing the neo^r gene from pHA178 (a gift from Hein TeRiele). To identify targeting events, three different probes were designed (Fig. 1): the internal probe, a ~500-nt NcoI-EcoRV fragment flanking exon 1 in the 3' end; the external probe, a ~1-kb BglII-SacI fragment that is adjacent but not contained in the targeting construct; and a probe generated from the neo^r gene. J1 and R1 ES cells were transfected with Naneo after linearization with ApaI, thus generating ~7.5 kb of homology in the 5' UTR, followed by seven amino acids of exon 1, and ~1.9 kb of homology in the 3' UTR, surrounding neo^r (Fig. 1 A). Electroporated cells were subjected to positive-negative selection for 8 days in 150 µg/ml G418 (GIBCO-BRL). Positive colonies were expanded and analyzed by Southern blot. Thereafter, genomic DNA was isolated, analyzed by digestion with BglII and EcoRI, separately, and sequentially labeled with the internal, external, and neo^r probes (Fig. 1 B). Two targeted ES cell clones, R1-34 and J1-87, were injected into C57BL/6 mouse blastocysts. About six to eight blastocysts were implanted per pseudopregnant female. From J1-87 implants, no chimeric males were born and the single >90% chimeric female was sterile. From R1-34 implants, one wild-type female, two 30% chimeric females, and four >90% chimeric males were born. Chimeric males were mated with C57BL/6J females to set up a breeding colony and test for germline transmission of the mutant allele. Heterozygous F1 animals have been intercrossed through more than five generations. Genotyping of tail DNA with polymerase chain reaction (PCR) required two primers flanking the neo^r gene insertion. The sense primer was (5'TGCGAGGAGCTAAACAGTGATTAAAG3'), and the antisense primer was (5'GGCTCCATTCCCTTATCAGACCTACCC3').

View larger version (40K): [in this window] [in a new window]   FIGURE 1   Targeted disruption of the NaChII locus. (A) Recombination design between the chromosomal locus (top) and the targeting vector Naneo (bottom). The Neor gene replaced most of exon 1. The replaced fragment was 700 nt larger than the excised exon 1. Internal, external, and neo probes are indicated as shaded rectangles; scale bars are shown under each clone. (B) Southern blot analysis of wild-type, J1-87, and R1-34 ES cell clones digested with EcoRI and BglII. The same blot was sequentially hybridized with internal, external, and neo probes. (C) PCR analysis of genomic DNA from NaChII+/+, NaCh+/, and NaChII/ littermates. The 0.45-kb band corresponds to the wild-type allele, and the 1.3-kb band indicates a NaChII null allele. (D) Southern blot analysis of the same tail genomic DNA digested with BglII and labeled with internal probe.

RNase protection assay

Levels of specific mRNAs were determined using solution hybridization RNase protection assays (RPAs) (Rockenstein et al., 1995). Poly A+ RNA was extracted from adult mouse brain, and reverse transcriptase-polymerase chain reaction (RT-PCR) was used to generate a 239-nt murine NaCh_II riboprobe with primers derived from the rodent sequence (GenBank accession number M22254, nt 4965-5204). The actin riboprobe was nt 480-559 (GenBank accession number X03672). Amplification products were subcloned into pCRII (Invitrogen), and T7 or SP6 RNA polymerase was used to generate ³²P-labeled antisense riboprobes from 100 ng of linearized plasmids. Total RNA was isolated from snap-frozen tissues with the TRI reagent and stored in Formazol buffer (Molecular Research Center) at 20°C. Statistical analysis of differences between the three groups of mice was performed utilizing the one-factor analysis of variance test.

Western blot analysis

Brains were homogenized, fractionated into cytosolic and particulate components, and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on 7% gels (Masliah et al., 1998b). Gels were blotted onto nitrocellulose membranes, and blots were incubated overnight with subunit-specific antipeptide antibodies (Alomone) (Westenbroek et al., 1992) (0.16 µg/ml NaCh_I or 0.38 µg/ml NaCh_II), followed by ¹²⁵I-protein A. Blots were analyzed with a PhosphorImager; the specific ~200-kDa signals, corresponding to NaCh_I and NaCh_II, were quantified using the ImageQuant software.

Hippocampal cultures

Newborns were placed under ice to induce hypothermia and then decapitated. Brains were removed and kept on ice in a divalent cation-free balanced saline solution (HBS): 144 mM NaCl, 3 mM KCl, 10 mM HEPES, pH 7.3. Hippocampi were dissected in ice-cold HBS solution and then incubated in Dulbecco's modified minimum essential medium (DMEM) (BioWhittaker) supplemented with 0.1% trypsin for 15 min at 37°C. Hippocampi were then washed twice in HBS and transferred to "complete medium," i.e., DMEM supplemented with Ham's F12 (BioWhittaker) and heat-inactivated bovine calf serum (Hyclone) (8:1:1 v/v), 24 U/ml penicillin, and 24 µg/ml streptomycin (Sigma). Tissue was triturated by repeated passings (10) through a fire-polished Pasteur pipette. Cells were counted in the presence of 0.04% Trypan blue and plated (~10⁴ viable cells/cm²) on poly-L-lysine-coated coverslips. Cultures were maintained on complete medium at 37°C in a humidified atmosphere containing 6% CO₂.

Electrophysiology

Neurons cultured for 5-9 days in vitro were used. The extracellular recording solution was 50 mM NaCl, 110 mM tetraethylammonium chloride, 2 mM BaCl₂, 0.3 mM CdCl₂, and 10 mM HEPES (pH 7.4). Whole-cell recordings (Hamill et al., 1981) were obtained at 23 ± 2°C, using pipettes pulled from Corning 7052 capillaries (Garner) on a P-97 puller (Sutter). Electrodes were filled with an internal solution containing 117 mM CsCl, 9 mM EGTA, 9 mM HEPES, 5 mM NaCl, 1.8 mM MgCl₂, 14 mM Tris-creatinePO₄, 4 mM MgATP, and 0.3 mM Tris-GTP (pH 7.4) (Raman et al., 1997). The pipette tip resistance was 5.5-7 M. Access resistance (~20 M) was compensated for by ~75%, and capacitative transients were minimized through the analog compensation circuitry of the patch-clamp amplifier (List EPC-7), which was also used to estimate cell capacity. Data were acquired and analyzed using an ITC-16 interface (Instrutech) and Pulse/PulseFit acquisition and analysis software (Heka). Currents were leak subtracted using the P/4 protocol. Processed data are reported as mean ± SE; n denotes the number of experiments. Statistical significance was assessed using Student's t-test.

Saxitoxin binding, autoradiography, and image analysis

Procedures were performed as described (Xia and Haddad, 1994). ³H-Saxitoxin (³H-STX) (24.0 Ci/mmol) and unlabeled STX were from Amersham and Calbiochem, respectively. Newborns were snap-frozen in isopentane cooled with liquid nitrogen. Frozen tissue blocks were mounted with OLR media and sectioned with a Leica cryostat. Frozen sections were incubated with ³H-STX in STX binding buffer in the 10-24 nM concentration range. Nonspecific binding was estimated by the addition of excess cold STX. One set of sections was used for autoradiography and another was scraped and quantified in a scintillation counter.

Tissue processing and histological analysis

A total of 150 newborns, ~50 each of NaCh_II^+/+, NaCh_II^+/, and NaCh_II^/ mice, were analyzed. Mice were placed in Bouin's fixative for 24 h, then divided sagitally and postfixed for 48 h. Specimens were washed, dehydrated, paraffin-embedded, serially sectioned at 7 µm with a Leica microtome, and stained with hematoxylin and eosin (H&E) or cresyl violet. Subsets of paraffin sections were used for immunocytochemistry and terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) staining.

Immunocytochemistry

Brain NaCh_II or NaCh_I subunits were localized by utilizing antibodies generated against peptides corresponding to residues 467-485 or 465-481 of the rodent proteins (Alomone) (Westenbroek et al., 1992), using 10-µm-thick paraffin sections (Masliah et al., 1998b). Blind-coded paraffin sections were first washed in sodium phosphate buffer, treated with H₂O₂, blocked with secondary antibody species-specific serum, and incubated overnight at 4°C with NaCh_II or NaCh_I antibodies. Sections were then washed and incubated with biotinylated goat anti-rabbit IgG, followed by avidin D horseradish peroxidase (ABC Elite, Vector Laboratories), and reacted with diaminobenzidine tetrahydrochloride containing 0.001% H₂O₂.

TUNEL assay and EM analysis of apoptosis

Cells undergoing DNA fragmentation were identified using a modified version (Masliah et al., 1998a) of the TUNEL method (Ben-Sasson et al., 1995). TdT-positive neurons were determined by image analysis and expressed as a mean per unit area. For EM, brains were fixed with 2% glutaraldehyde/0.1% osmium tetroxide in 0.1 M sodium cacodylate and embedded in epoxy. Blocks were sectioned with a Leica Ultracut E ultramicrotome, placed on copper grids, stained by the Sato lead method, and evaluated with a Zeiss EM10 electron microscope (Masliah et al., 1996).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Targeted disruption of the NaCh_II gene

The NaCh_II gene was inactivated by homologous recombination deleting most of exon 1, encoding from amino acid 1 to amino acid 89, and substituting this sequence by the neomycin resistance gene (neo^r) (Fig. 1 A). Two independently targeted ES cell lines, J1-87 and R1-34, were obtained, injected into C57BL/6J blastocysts, and transferred to pseudopregnant females. Three chimeric males from the R1-34 line gave germline transmission of the mutated NaCh_II gene. Southern blot analysis of wild type (WT) and the two successfully targeted ES cell clones is shown in Fig. 1 B. Confirmed chimeras were then crossed to C57BL/6J mice to generate heterozygous NaCh_II mutants, and heterozygous mice were then crossed to generate the homozygous NaCh_II^/ mice. Tail DNA from offspring mice were genotyped by PCR amplification, using two oligonucleotide primers flanking the neo^r insertion site. The WT locus amplified a band of 450 bp, whereas the targeted locus amplified a band of ~1.3 kb (Fig. 1 C). Southern blot analysis of the same tail genomic DNA digested with BglII, using an internal probe, confirmed the genotype (Fig. 1 D). Genotyping of more than 640 newborn pups derived from interbreeding of heterozygous mice indicated that the NaCh_II deficiency in the NaCh_II^/ mice resulted in perinatal lethality, despite a Mendelian distribution that suggests that the NaCh_II gene is not embryonic lethal (Table 1).

NaCh_II expression is drastically reduced in the brains of NaCh_II^/ mice

We investigated the expression of the NaCh_II. RNase protection assays revealed that the steady-state level of the NaCh_II transcript present in whole brain preparations of NaCh_II^/ and NaCh_II^+/ mice was reduced by 85% and ~50% compared to that detected in brains of NaCh_II^+/+ littermates (Fig. 2 A). In contrast, RNase protection assays specific for the NaCh_I transcript showed no detectable differences between NaCh_II^/ and NaCh_II^+/ mice.

View larger version (48K): [in this window] [in a new window]   FIGURE 2   Analysis of NaChII mRNA levels in brains of NaChII/ mice. (A) Representative RPA autoradiograph. The lane identified as Probe shows signals of undigested radiolabeled probes (indicated on the left; no RNase added). The other lanes contained the same riboprobes and brain RNA samples from newborn littermates and adult NaChII+/+ and NaChII+/ mice, digested with RNase. Labels and arrows on right indicate the expected size for specific protected mRNA fragments after digestion of noncomplementary sequences (vector-derived) and transcripts. Actin signals were used as a control for RNA content/loading. (B) PhosphorImager quantitation of RPAs conducted on hippocampus, cortex, and brainstem from NaChII+/+, NaChII+/, and NaChII/ littermates. Columns and error bars represent means ± SE (n = 3). (C) Western blot analysis of NaChI and NaChII immunoreactivity from brain membranes of NaChII+/+, NaChII+/, and NaChII/ littermates. Lanes labeled "rat" indicate positive control preparations obtained from rodent brains. Numbers indicate Mr standards. Anti-NaChI and anti-NaChII indicate blots probed with the corresponding antibody. Nonspecific background and cross-reactivity of anti-NaChII to NaChIII may account for the dim bands present in the NaChII/ samples.

To determine the relative levels of NaCh_II mRNA in different brain regions, brains from NaCh_II^/, NaCh_II^+/, and NaCh_II^+/+ mice were dissected into cortex, hippocampus, and brainstem. This analysis showed that the brain region that expressed the highest NaCh_II transcript level was the brainstem, followed by hippocampus and neocortex. There was a drastic reduction of NaCh_II transcript level in NaCh_II^/ mice in all three brain regions and a reduction in the NaCh_II^+/ mice with respect to wild type, ranging from ~36% in brainstem and ~48% in cortex to ~53% in hippocampus. The most striking feature appeared to be the ratio of transcript level between wild-type mice and the homozygous mutant (), which followed the sequence brainstem (  6) > cortex (  5) > hippocampus (  4) (Fig. 2 B). Statistical analysis of differences between the three groups of mice showed that differences between NaCh_II^+/+ and NaCh_II^+/, between NaCh_II^+/+ and NaCh_II^/, and between NaCh_II^+/ and NaCh_II^/ were significant at the p < 0.05 level (poshoc Fisher). These results suggest that the expression of NaCh_II RNA is dosage sensitive and most drastically compromised in the brainstem of the NaCh_II^/ neonates.

To verify that the reduction of NaCh_II mRNA expression in the NaCh_II^/ mice was associated with a concomitant reduction in protein expression, levels of NaCh_II immunoreactivity were determined by protein immunoblot analysis (Westenbroek et al., 1992) (Fig. 2 C). In brain membranes from NaCh_II^+/+ and NaCh_II^+/ mice NaCh_II immunoreactivity was discernible as an intense band at a M_r of ~200 kDa, whereas in membranes from NaCh_II^/ mice NaCh_II immunoreactivity was barely detectable. In contrast, NaCh_I protein was detected at approximately the same levels in membranes from NaCh_II^+/+, NaCh_II^+/, and NaCh_II^/ mice. Together, these results are consistent with the RNase protection data and show that NaCh_II protein expression is markedly reduced in brains of NaCh_II^/ mice.

Sodium channel currents in hippocampal neurons of NaCh_II^/ mice are sharply attenuated

We investigated the biophysical properties of NaChs, using whole-cell recordings from hippocampal neurons in culture (Hamill et al., 1981). Fig. 3 A (left) shows families of sodium currents in response to a series of 20-ms depolarizing voltage steps, from a holding potential of 120 mV, for NaCh_II^+/+ (top), NaCh_II^+/ (middle), and NaCh_II^/ (bottom) mice. The time course of the current transients is characteristic of voltage-gated NaChs (Armstrong and Hille, 1998): currents increase to a peak and then decline with a sigmoidal early activation and a subsequent exponential decay. The peak current-voltage (I-V) relationship is shown in the corresponding right panels. The most conspicuous feature was the drastic reduction of I in cells from the NaCh_II^/ mice. On average, the maximum conductance (G_max) density, determined from the positive resistance region of the I-V curve, was 0.33 ± 0.05 (n = 15), 0.64 ± 0.06 (n = 45), and 1.12 ± 0.13 nS/pF (n = 26) for cells from NaCh_II^/, NaCh_II^+/, and NaCh_II^+/+ mice, respectively (Fig. 3 B). The differences between the three groups were statistically significant (p < 0.001). This pattern of expression of sodium currents follows gene dosage and is consistent with the findings of the RNase protection assay and protein immunoblots, all indicative of a severe deficiency of the NaCh_II gene product. The residual sodium currents in cells from the NaCh_II^/ mice is consistent with the occurrence of other NaChs, the expression of which is not altered by the targeted disruption of the NaCh_II gene.

View larger version (29K): [in this window] [in a new window]   FIGURE 3   Sodium channel currents in hippocampal pyramidal neurons from newborn NaChII+/+, NaChII+/, and NaChII/ littermates. (A) Family of sodium currents elicited by 20-ms voltage steps from a holding potential of 120 mV. Test potentials ranged from 90 mV to 5 mV in 5-mV increments. Time and current scale bars apply to all three left panels. Right panels show the peak currents plotted versus test potential for the records shown at left. (B) Maximum sodium conductance (Gmax) density evoked by depolarizing potentials for neurons from NaChII+/+ (n = 26), NaChII+/ (n = 45), and NaChII/ (n = 15) littermates. Columns and error bars represent means ± SE. (C) Peak conductance-voltage (Gpeak-V) relationships calculated according to Gpeak = Ipeak/(V  VNa+), where I is the peak current amplitude, V is the applied voltage, and VNa+ is the Nernst equilibrium potential. The solid line depicts the least-squares fit to data points and is given by a Boltzmann function, normalized G = 1/{1 + exp((V  V1/2)/a)}, where V1/2 is the half-activation potential at which the slope a is depicted. The V1/2 and a values for the fitted curves are 40.7 mV (n = 12) and 7.7 for NaChII+/+ (n = 25), 35.0 mV and 7.6 for NaChII+/ (n = 45), and 33.5 mV and 6.8 for NaChII/ (n = 14). Each point represents the mean ± SE. (D) Steady-state voltage dependence of inactivation, determined using a series of 200-ms prepulses, ranging from 100 mV to 10 mV, followed by a 10-ms test pulse to 0 mV. Peak currents were normalized with respect to the maximum and plotted as a function of the prepulse potential. The solid line plots the least-squares fit to the data points given by a Boltzmann function, Normalized current = 1/{1 + exp((V  Vh)/a)}, where V is the prepulse potential and Vh is the half-inactivation potential at which the slope a is depicted. The Vh and a values for the fitted curves are 53.6 mV and 8.1 for NaChII+/+ (n = 15), 56.5 mV and 7.9 for NaChII+/ (n = 20), and 54.9 mV and 9.0 for NaChII/ (n = 9). Each point represents the mean ± SE.

The currents were characterized by examining the peak sodium conductance-voltage relation (G-V) (Fig. 3 C). The G-V curve is a sigmoidal function of voltage, which is taken as an indication of the relative number of channels that are closed versus open as a function of voltage. The solid lines depict the least-squares fit of the data to the Boltzmann function reported in Fig. 3; two parameters were extracted from the fit: V_1/2, the half-activation potential at which the slope a is depicted. For the NaCh_II^+/+, G increased with depolarizing voltages to a maximum at a V  10 mV, exhibiting a V_1/2 at V = 40.7 mV and an e-fold change in G per 7.7 mV. The corresponding parameters for the NaCh_II^+/ and NaCh_II^/ were V_1/2 = 35.0 mV and 33.5 mV, with an e-fold change in G per 7.6 mV and 6.8 mV, respectively. Statistical analysis of the data shows that the V_1/2 values are 40.3 ± 2.2 mV (n = 25), 34.6 ± 1.2 mV (n = 46), and 32.8 ± 1.8 mV (n = 14) for NaCh_II^+/+, NaCh_II^+/, and NaCh_II^/ neurons, respectively. The positive shift of 7.5 mV in the NaCh_II^/ mice with respect to the NaCh_II^+/+ littermates is statistically significant (p = 0.025), whereas the difference in V_1/2 values between NaCh_II^+/ and NaCh_II^/ neurons is not (p = 0.47). Thus the remaining ensemble of NaChs in hippocampi of NaCh_II^/ mice activates at more positive voltage.

The steady-state voltage dependence of inactivation was studied by using a pulse protocol in which the test pulse to 0 mV was preceded by 200-ms depolarizing prepulses to different potentials (Fig. 3 D). As the potential of the prepulse is increased, the current flow during the test pulse decreases sigmoidally because of the voltage-dependent inactivation of channels during the prepulse. For the NaCh_II^+/+, the midpoint of the steady-state inactivation (V_h), extracted from a fit of the data points to a Boltzmann function, was 53.6 mV, with an e-fold decrease in the steady-state state current per 8.0 mV. The corresponding parameters for the NaCh_II^+/ and NaCh_II^/ were V_h = 56.5 mV and 54.9 mV, with an e-fold change in G per 7.9 mV and 9.0 mV. Statistical analysis of the data shows that the V_h values are 52.3 ± 2.3 mV (n = 15), 58.1 ± 1.6 mV (n = 20), and 54.7 ± 1.4 mV (n = 9) for NaCh_II^+/+, NaCh_II^+/, and NaCh_II^/ neurons, respectively. The difference in V_h values between the three groups is not statistically significant (p > 0.1). Together, these results show that the targeted disruption of NaCh_II reduces the NaCh currents in hippocampal neurons with minor changes in the equilibrium voltage-dependent properties.

The density of saxitoxin-binding sites is markedly reduced in the rostral brain and barely detectable in the brainstem of NaCh_II^/ mice

To determine the density of NaChs in brains from NaCh_II^/ mice we used saxitoxin (STX), a specific ligand of voltage-gated NaCh  subunits (Xia and Haddad, 1994; Couraud et al., 1986). Autoradiographic analysis of brain sections exposed to ³H-STX showed that the STX binding density was markedly reduced in the rostral brain and in the brainstem of the NaCh_II^/ mice (Fig. 4, C and D). This finding is to be contrasted with the high STX binding density in the brainstem of NaCh_II^+/+ littermates (Fig. 4, A and D). The slight increment in STX binding affinity displayed by NaCh_II^/ mice suggests that the remaining aggregate of NaChs was not altered (Fig. 4 D).

View larger version (94K): [in this window] [in a new window]   FIGURE 4   Marked reduction in STX-binding density in the brainstem of NaChII/ mice. Autoradiographic images from brains of NaChII+/+ (A) and NaChII/ (C) newborn littermates are pseudocolored to represent the relative STX binding density, from highest (red) to lowest (blue). Autoradiograms were obtained at 15 nM 3H-STX. (B) Comparable brain section stained with H&E, indicating selected regions: OB, olfactory bulb; NCTX, neocortex; BG, basal ganglia; HIP, hippocampus; BS, brainstem. (A) Autoradiograms of NaChII+/+ brains show the most intense binding in the brainstem, followed by the hippocampus and basal ganglia; the olfactory bulb and neocortex display lower binding. (C) Autoradiograms of NaChII/ brains show very weak STX binding in the brainstem and modest binding in the basal ganglia and neocortex. Magnification for A-C: 10×. Scale bar, 3 mm. (D) Quantitation of STX binding density and affinity in brains of NaChII+/+ and NaChII/ newborn littermates. STX binding affinity is expressed as 1/Kd. Statistical significance was assessed using the unpaired, two-tailed Student t-test; the asterisk indicates that the differences are statistically significant with p < 0.05. Columns and error bars represent means ± SE (n = 3).

It is known that SCN1A (Smith et al., 1998), SCN2A (Noda et al., 1986; Suzuki et al., 1988; Smith et al., 1998), SCN3A (Suzuki et al., 1988), and SCN8A (Smith et al., 1998)  subunits bind tetrodotoxin and STX with comparable affinities. It is therefore unlikely that the modest increase in STX binding affinity detected in brains from NaCh_II^/ newborn mice arises from a higher affinity of the remaining aggregate of NaChs. Subtle differences in affinity, discerned by using this type of global measurement (Xia and Haddad, 1994), may reflect the relative abundance of the remaining NaCh in the newborn brain of NaCh_II^/ mice. It is known that the different NaCh  subunit mRNAs display distinct temporal expression patterns in the CNS (Beckh et al., 1989). SCN2A is the most abundant species postnatally and remains at high levels through adulthood (Beckh et al., 1989). SCN3A is expressed predominantly at fetal and early postnatal stages, whereas SCN1A predominates at late postnatal stages (Beckh et al., 1989). No comparable information is available for SCN8A; however, the development and function of spinal motoneurons depends on the postnatal induction of SCN8A expression (Garcia et al., 1998). Based on the information at hand, it is conceivable that, in the absence of the most abundant SCN2A (Beckh et al., 1989), the preponderant species in newborn brains is SCN3A, and that the modest increase in STX binding affinity measured in NaCh_II^/ newborn brains is nothing more than a reflection of this enrichment relative to SCN1A and SCN8A.

Perinatal lethality of NaCh_II^/ mice

Mice heterozygous for the mutation (NaCh_II^+/) appeared to be normally developed and did not display gross morphological anomalies. In contrast, NaCh_II^/ pups died within 1-2 days of birth. During this early postnatal period, the pups appeared to quickly dehydrate and lose weight (~16% lower body weight than NaCh_II^+/ littermates, measured concurrently within 12 h of birth); they were pallid, dyspneic or gasping, and cyanotic (Fig. 5 A). The lower abdominal region was dark, with reduced milk content in the stomach. NaCh_II^/ neonates were hypoxic and presented a drastic reduction of the hemoglobin percentage saturation (59 ± 5%, n = 9), contrasted with NaCh_II^+/+ (86 ± 4%, n = 7) and NaCh_II^+/ littermates (83 ± 4%, n = 7). No seizure activity was detected in NaCh_II^/ neonates; pups move and react to gentle touch without an apparent difference with respect to NaCh_II^+/ littermates.

View larger version (89K): [in this window] [in a new window]   FIGURE 5   Anatomical analysis of NaChII/ mice. (A) Gross examination of the NaChII+/+, NaChII+/, and NaChII/ mice shows normal organogenesis; NaChII/ mice appear smaller in size. Scale bar, 10 mm. (B-D) Neurohistological examination of H&E-stained forebrain (top panels; magnification 20×; scale bar, 2 mm) and hippocampus (bottom panels, magnification 40×; scale bar, 0.5 mm) of NaChII+/+ (B), NaChII+/ (C), and NaChII/ (D) mice shows normal development. (H-K) Analysis of patterns of NaChII (H and I) and NaChI (J and K) immunoreactivity in the CNS revealed intense NaChII immunoreactivity in the brainstem (H) of NaChII+/+ mice, contrasted to the lack of neuronal staining in the corresponding brain region of the NaChII/ mice (I). No differences in NaChI immunoreactivity between the brainstems of NaChII+/+ (I and J) and NaChII/ (K) mice were detected. Sections were obtained from mice ~12 h after birth. Magnification, 263×. Scale bar, 20 µm.

Despite the severe deficiency of NaCh_II, the identification of residual NaCh_II mRNA and protein (Fig. 2) suggests that the targeted disruption of the NaCh_II gene may be incomplete. Arguably, NaCh_II^/ mice do not express sufficient levels of NaCh_II to allow survival to adulthood. In contrast, the NaCh_II^+/ mice breed normally and appear to be functional. Clearly, levels of NaCh_II somewhere between these two boundaries are required for survival.

Absence of neuroanatomical alterations in NaCh_II^/ mice

Inspection of brains from NaCh_II^/ newborn mice revealed no neuroanatomical abnormalities. Microscopic analysis of the cortical (Fig. 5, B-D), hippocampal (Fig. 5, E-G), cerebellar, and brainstem regions showed these to be well developed, comparable to those of the NaCh_II^+/+ and NaCh_II^+/ littermates, exhibiting overall intact organization and well-defined neuronal cytoarchitectures. Immunocytochemical analysis of neonate brains from NaCh_II^+/+ and NaCh_II^+/ mice revealed anti-NaCh_II immunoreactivity associated with the neuronal cell bodies. Pyramidal neurons in the neocortex and hippocampus showed mild immunoreactivity, whereas neurons within the brainstem and dorsal root ganglion presented intense anti-NaCh_II immunoreactivity (Fig. 5 H). In the NaCh_II^/ mice there was a marked reduction of NaCh_II immunoreactivity throughout the nervous system, as shown in a representative section of brainstem (Fig. 5 I). The NaCh_I-specific antibody reacted mildly with neuronal cell bodies throughout the CNS of NaCh_II^+/+, NaCh_II^+/, and NaCh_II^/ mice. No differences were discerned, as displayed in representative sections of brainstem (Fig. 5, J and K) for NaCh_II^+/+ and NaCh_II^/ mice, respectively.

Increased neuronal apoptosis and death in neurons of NaCh_II^/ mice

Given that organogenesis and brain development appeared to be normal in the NaCh_II^/ mice, and considering that neonates die within 24-48 h of birth with a severe hypoxic phenotype, we hypothesized that abnormal neuronal function compromised by the NaCh_II deficit might lead to neuronal cell death, especially in brain areas critically involved in respiratory regulation. To test this model, analysis of neuronal cell death was undertaken, using the TUNEL method (Ben-Sasson et al., 1995). A most conspicuous phenotype of the NaCh_II^/ mice is the marked increment in neuronal apoptosis in the brainstem (Fig. 6, D and F). On average, 4.5-fold and 3-fold increments in the number of TUNEL-positive cells per section were identified in the brainstem (Fig. 6, D and F) and neocortex (Fig. 6, B and E) from NaCh_II^/ mice. This is consistent with a global change of the brain, presumably associated with the hypoxia identified in NaCh_II^/ neonates.

View larger version (82K): [in this window] [in a new window]   FIGURE 6   Increased neuronal apoptosis in the brains of NaChII/ mice: in situ DNA fragmentation assay for apoptosis. In the wild-type neocortex (A and E) and brainstem (C and F), occasional labeling was discerned. In the neocortex (B and E) and brainstem (D and F) of NaChII/ mice, extensive positive staining for apoptosis was observed (arrows). Magnification, 263×. Scale bar, 150 µm. Statistical significance was assessed using the unpaired, two-tailed Student's t-test (p < 0.05, n = 3).

To determine whether the positive TUNEL reaction was due to augmentation of apoptosis or to necrotic changes secondary to hypoxia, tissues were prepared for ultrastructural examination. This analysis revealed that neurons from the NaCh_II^/ mice presented a marked increase in chromatin condensation and segregation (Fig. 7, B and D-F), two prominent features of apoptotic neurons (Portera-Cailliau et al., 1997). Mitochondria appeared to be markedly swollen, with cristae dilatation, in neurons from brains of NaCh_II^/ mice (Fig. 7 D) when compared with NaCh_II^+/+ littermates (Fig. 7 C). Complete condensation of chromatin and apoptotic bodies was clearly discerned in neurons from the NaCh_II^/ neonates (Fig. 7, E and F). In addition, the neuropil and neuronal cell bodies displayed extensive vacuolation. The areas most affected were the neocortex and the brainstem. Thus neurons from NaCh_II^/ neonates display the ultrastructural hallmarks of apoptotic cell death.

View larger version (210K): [in this window] [in a new window]   FIGURE 7   Ultrastructural analysis of neuronal alterations in the brains of NaChII/ mice. (A) In NaChII+/+ mice, neurons displayed a typical appearance. (B) In NaChII/ mice, neurons exhibited extensive vacuolation of the cytoplasm (arrows) and the neuritic processes (n), accompanied by chromatin condensation. (C) In NaChII+/+ mice, mitochondria displayed a normal appearance. (D) In NaChII/ mice, mitochondria were swollen and their cristae were dilated. In addition, neurons displayed several levels of chromatin condensation, progressing from partial segmentation (E) to complete condensation (F). Magnification: A and B, 5000×; scale bar, 10 µm. C and D, 18,000×; scale bar, 10 µm. E and F, 10,000×; scale bar, 5 µm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

We have generated knockout mice deficient in brain NaCh_II that exhibit perinatal lethality with massive neuronal apoptosis. A most striking feature emerging from the biophysical analysis of NaCh currents from hippocampal neurons is the drastic reduction in the expression of sodium currents in the NaCh_II^/ mice. This, in and of itself, would explain the severe decrement in excitability apparent in the NaCh_II^/ mice. The electrophysiological analysis further uncovered the occurrence of a statistically significant depolarizing shift of 7.5 mV of the half-activation potential in the NaCh_II^/ mice with respect to the NaCh_II^+/+ littermates, contrasted with the invariance of the midpoint of the steady-state inactivation. Based solely on these characteristics, neurons from the NaCh_II^/ mice would be expected to display decreased electrical excitability and a higher threshold for activation of action potentials compared with wild-type neurons. Such properties would produce a significant decrease in the probability of the channels that open in response to a given depolarization compared to neurons from wild-type animals, which would express the full complement of the NaChs. It is worth noting that these properties would be expected to lessen the window of excitability, namely the net conductance that would occur in the voltage region where the activation and inactivation curves overlap. This is significant, as it has been documented that small changes in this "window current" are associated with channel dysfunctions or channelopathies. A case in point is hyperkalemic periodic paralysis, in which a mutation (T704M) of the human muscle NaCh  subunit (SCN4A) that is associated with the phenotype, when introduced into the recombinant channel and heterologously expressed, exhibits shifts in the midpoints of the steady-state activation and inactivation curves. The augmented overlap of the steady-state activation and inactivation curves would be expected to produce such a window current, which in turn would account for the persistent depolarizing inward sodium current across the muscle cells in patients with this disease (Yang et al., 1994).

The differences described between NaCh_II^+/+ and NaCh_II^/ neurons are likely to arise from the properties of the remaining NaChs, SCN1A, SCN3A, and SCN8A. The properties of heterologously expressed SCN1A (Smith et al., 1998) and SCN3A (Suzuki et al., 1988) appear to be fundamentally similar to those of SCN2A (Stuhmer et al., 1987, 1989; McCormick et al., 1999). It is documented that SCN8A channels inactivate more rapidly than SCN1A or SCN2A when expressed in amphibian oocytes (Smith et al., 1998). However, the three NaChs, SCN1A, SCN2A, and SCN8A, displayed similar voltage dependence and inactivation kinetics when coexpressed together with the ₁ and ₂ subunits (Smith et al., 1998). Presumably, the targeted deletion of the NaCh_II subunit in the mice reported here did not alter the expression of the ₁ and ₂ subunits. And it is most likely that sodium channels in the CNS are heterooligomers of  and  subunits. It is plausible, therefore, that the findings obtained with neurons from NaCh_II^/ mice arise from the remaining ensemble of NaChs assembled from combinations of SCN1A, SCN3A, and SCN8A  subunits with both ₁ and ₂ subunits.

There are several possible mechanisms by which the elimination of a single constituent from a multicomponent combination may give rise to the properties remaining in the system. First, the component missing in the NaCh_II^/ mice would accordingly be interpreted to exhibit a more negative midpoint of the steady-state activation than the residual components, SCN1A, SCN3A, and SCN8A. Evidence at hand suggests that this may be the case (Smith et al., 1998). Second, the possibility arises that deletion of the NaCh_II subunit induces compensatory up- or down-regulation of SCN1A, SCN3A, and SCN8A, by eliminating the competition of the remaining  subunits for the available complement of ₁ and ₂ subunits. Third, the absence of SCN2A may affect the interaction of the SCN1A, SCN3A, and SCN8A  subunits with the  subunits and thereby modulate the extent of surface expression of functional channels (Isom et al., 1995; Patton et al., 1994). Fourth, there may be accessory proteins other than both  subunits that may modulate the properties of the residual channels. For example, SCN2A channels display persistent currents when coexpressed with the G-protein subunits (Ma et al., 1997). It would be interesting to explore other possible mechanisms that may involve changes in protein trafficking or even gene expression arising from the interplay of NaCh  and  subunits with other protein components involved in synaptic activity, as has been suggested for calcium channels (Sutton et al., 1999).

The present study shows that although expression of NaCh_II is redundant for embryonic development, it is essential for postnatal survival. There are at least two plausible explanations for this finding. One view is that decreased neuronal activity associated with NaCh deficits activates programmed cell death within vital regions of the brain such as the brainstem and, in turn, leads to respiratory depression, hypoxia, and perinatal death. Alternatively, decreased neural activity within the brainstem may produce respiratory depression, hypoxia, oxidative stress, and secondary cell death. A sharp boundary between these two possible mechanisms may not occur, and the two may converge and contribute to neuronal demise and perinatal death. We tend to favor the first hypothesis, based on the following evidence. It is well documented that hypoxia alone is capable of inducing apoptosis in neuronal cultures and that the extent of damage is influenced by the duration and the severity of the hypoxic insult (Banasiak and Haddad, 1998). Furthermore, in this study we demonstrate that the highest level of expression for NaCh_II is within the mouse brainstem neurons, in agreement with previous studies on rat brain (Xia and Haddad, 1994). The brainstem, the most primitive region of the brain, contains the respiratory rhythmogenic neurons and cardiovascular control centers, among others primarily involved in autonomic functions (St-John, 1998; Reckling and Feldman, 1998). The basic neuronal circuits that generate respiratory rhythm in the brainstem have been identified; however, the mechanisms of respiratory rhythmogenesis are still poorly understood (Reckling and Feldman, 1998; Gray et al., 1999). A site in the rostral ventrolateral medulla known as the pre-Bötzinger complex is considered to be necessary and sufficient to generate respiratory rhythm (Reckling and Feldman, 1998; Gray et al., 1999). And numerous distinct types of neurons have been identified in this complex; however, their functional phenotype, e.g., rhythm-generating or pattern-forming, remains to be defined. Despite the paucity of information, electrophysiological analysis of the properties of different neurons in this complex has disclosed the occurrence of neurons with intrinsic oscillatory bursting properties, associated to some extent with the activity of voltage-gated NaChs (Smith et al., 1992; Onimaru and Homma, 1992; Reckling et al., 1996). The abundance of NaCh_II in the brainstem makes it a plausible candidate, endowing a given class of rhythmogenic neurons with their bursting properties. The underlying mechanisms that determine that neurons with