Abstract
The β subunits of voltage-gated Na channels (Scnxb) regulate the gating of pore-forming α subunits, as well as their trafficking and localization. In heterologous expression systems, β1, β2, and β3 subunits influence inactivation and persistent current in different ways. To test how the β4 protein regulates Na channel gating, we transfected β4 into HEK (human embryonic kidney) cells stably expressing NaV1.1. Unlike a free peptide with a sequence from the β4 cytoplasmic domain, the full-length β4 protein did not block open channels. Instead, β4 expression favored open states by shifting activation curves negative, decreasing the slope of the inactivation curve, and increasing the percentage of noninactivating current. Consequently, persistent current tripled in amplitude. Expression of β1 or chimeric subunits including the β1 extracellular domain, however, favored inactivation. Coexpressing NaV1.1 and β4 with β1 produced tiny persistent currents, indicating that β1 overcomes the effects of β4 in heterotrimeric channels. In contrast, β1C121W, which contains an extracellular epilepsy-associated mutation, did not counteract the destabilization of inactivation by β4 and also required unusually large depolarizations for channel opening. In cultured hippocampal neurons transfected with β4, persistent current was slightly but significantly increased. Moreover, in β4-expressing neurons from Scn1b and Scn1b/Scn2b null mice, entry into inactivated states was slowed. These data suggest that β1 and β4 have antagonistic roles, the former favoring inactivation, and the latter favoring activation. Because increased Na channel availability may facilitate action potential firing, these results suggest a mechanism for seizure susceptibility of both mice and humans with disrupted β1 subunits.
Introduction
Ionic current through tetrodotoxin (TTX)-sensitive, voltage-gated Na channels generates the upstroke of most neuronal action potentials. In voltage-clamped neurons, step depolarizations evoke rapidly activating and inactivating “transient” Na currents that are grossly similar across cells. Nevertheless, neurons differ in the amplitude of “persistent” Na current remaining after transient currents inactivate (Crill, 1996; Magistretti et al., 1999) and in the expression of “resurgent” current evoked by repolarization from positive potentials (Raman and Bean, 1997; Do and Bean, 2003; Afshari et al., 2004). By increasing channel opening and availability at subthreshold voltages, persistent and resurgent currents generally increase firing rates (Stafstrom et al., 1984; Khaliq et al., 2003). In fact, drugs such as phenytoin and carbamazepine, which stabilize inactivation and minimize persistent current (Chao and Alzheimer, 1995; Kuo et al., 1997; Lampl et al., 1998), are used clinically to control seizures.
“Window” current, which flows at voltages at which activation occurs but inactivation is submaximal, contributes to persistent current (Attwell et al., 1979). At more positive voltages, persistent current flows through Na channels that occasionally fail to inactivate (Alzheimer et al., 1993; Brown et al., 1994) and/or have an incomplete equilibrium occupancy of inactivated states (Taddese and Bean, 2002). Resurgent current, in contrast, results from an open-channel blocking protein that binds channels at positive voltages. After repolarization, the blocker unbinds, permitting current to flow briefly (Raman and Bean, 2001).
The molecular basis for Na current diversity depends partly on subunit composition. In central neurons, the pore-forming α subunits that carry Na current include NaV1.1, NaV1.2, and NaV1.6 (Felts et al., 1997; Smith et al., 1998). These associate with β subunits, which regulate both trafficking and gating. β1 accelerates fast inactivation (Isom et al., 1992, 1995a; Chen and Cannon, 1995); β3 increases persistent current (Qu et al., 2001); β2 can do both (Isom et al., 1995b; Qu et al., 2001). The β4 subunit (Yu et al., 2003) is indirectly implicated in resurgent current, because its cytoplasmic domain mimics the action of the endogenous blocking protein (Grieco et al., 2005), but this behavior has not been replicated in heterologous expression systems (Chen et al., 2008).
The varied effects of β subunits on gating raise the question of whether interactions among subunits generate Na channel complexes with distinct properties. To explore this possibility, we studied how β4 modulates Na currents when expressed either alone or with other β subunits. In both expression systems and hippocampal neurons, β4 overexpression favored channel activation and increased persistent current. This effect was antagonized, and inactivation was accelerated by coexpression of β1, but not β1C121W, a mutant subunit linked to generalized epilepsy with febrile seizures plus (GEFS+) (Wallace et al., 1998, 2002). Activation was facilitated, and inactivation rates were slowed in β4-expressing neurons of Scn1b null mice. These data suggest that β1 has a dominant role in reducing Na channel activity, raising the possibility that disruption of β1 in inherited epilepsies may slow inactivation rates in some neurons, thereby contributing to the excessive firing associated with seizure disorders.
Materials and Methods
HEK-293 cells.
HEK-293 cells stably transfected with NaV1.1 (HEK–NaV1.1 cells) were obtained from GlaxoSmithKline under a Materials Transfer Agreement. These cells provide a non-neuronal mammalian cell line containing neuronal Na channels, in which proteins associated with neuronal Na channels can be easily transfected and studied. Like other cells of adrenal origin, HEK–NaV1.1 cells express the splice variant β1B (Moran et al., 2000); however, Western blots indicate no detectable expression of β1A (hereafter called β1), β2, β3, or β4 in these cells (data not shown; R. Rusconi and L. L. Isom, unpublished observation). One day before recording or transfection, cells were plated on poly-l-lysine-coated coverslips at ∼5000–10,000 cells per coverslip.
CA3 hippocampal cultures.
Cultures were prepared according to a protocol modified from Tovar and Westbrook (2002). All chemicals, except as noted, were from Sigma-Aldrich. Neurons and glia were prepared by dissecting the CA3 region of the hippocampus from C57BL/6 mice, aged postnatal day 0 (P0), in cold D1 dissection saline (140 mm NaCl, 5 mm KCl, 0.1 mm Na2HPO4, 2.2 mm KH2PO4, 5 mm HEPES, 4 mm sucrose, 30 mm glucose, 10 μl/ml penicillin/streptomycin, and 0.001% phenol red). Tissue was incubated in 20 U/ml papain (Worthington Biochemicals), 1.7 mm cysteine, 100 μm CaCl2, and 50 μm EDTA in the D1 solution (pH 7.3 with NaOH) for 40 min at 32°C. The tissue was washed in neuronal medium [5% heat-inactivated fetal calf serum (HyClone), 20 mm glucose, and 0.5 mm Glutamax (Invitrogen) in Minimum Essential Medium (Invitrogen)] with 2.5 mg/ml each of bovine serum albumin and trypsin inhibitor. The tissue was then triturated with polished Pasteur pipettes to release individual cells. To make glial beds, cells were plated at a concentration of 100,000 ± 25,000 cells per coverslip onto poly-l-lysine-coated coverslips coated with collagen (Cohesion) and were allowed to grow to confluence. One week later, surviving neurons were killed by excitotoxicity with (in mm) 0.2 glutamate, 2 CaCl2, 165 NaCl, 5 KCl, and 5 HEPES, and freshly isolated cells (neurons and glia) were plated onto the glial beds at the same concentration as above. To prevent further glial cell proliferation in neuronal cultures, 5 μm cytosine arabinoside was added after cells were plated. Neurons were transfected at 2–3 d in vitro, and recordings were made at 6–15 d after transfection.
β4 clones, chimeras, and β subunit mutations.
A PCR strategy was used to generate β4 cDNA. Two sets of oligonucleotide primers were designed from the rat Scn4b mRNA sequence in GenBank, a set of external primers (forward, 5′-CTGTGCACACTGTCCTATCCAAGC-3′; reverse, 5′-CACATCTCAGACAGGACTCGGCATC-3′) and a set of internal primers (forward, 5′-CAACTCGAGCGCTCCGGAGAGAACAGGAC-3′; reverse, 5′-ATCGAATTCACCATCAGAAAGTGAGGCTC-3′). The external primer set was used to perform reverse transcription-PCR from rat brain total RNA. Then, using an aliquot of the first PCR product as a template, a second round of PCR was performed using the set of internal primers. With this design, a PCR product of ∼700 bp was amplified. This PCR product was digested with Xho I and EcoRI and subcloned into pcDNA 3.1/Zeo(−). The sequence of the cDNA insert was found to be identical to the GenBank sequence. To facilitate stable mammalian cell expression with sodium channel α subunits, the β4 cDNA was subsequently subcloned into pcDNA3.1/Hygro(−).
PCR strategies were used to generate the β1/4 and β2/4 subunit chimeras. For generation of β1/4, cDNA fragments encoding the extracellular and transmembrane domain of β1 and the intracellular domain of β4 were amplified from plasmids containing the rat Scn1b or rat Scn4b cDNAs, respectively. The β1 domain was generated with the following primer set: β1-forward, 5′-ATCTCCTGTCGCCGCGCTCT-3′; β1/4-reverse, 5′-AGTGATGAGCTTCTTGTAGCAGTA-3′. The β4 domain was generated with the following primer set: β1/4-forward, 5′-TACTGCTACAAGAAGCTCATCACT-3′; β4-reverse, 5′-CAGGGCCTCACACTTTTGTGG-3′. The resulting PCR fragments were analyzed by agarose gel electrophoresis and purified. β1/4-reverse and β1/4-forward contained complementary, overlapping sequences such that the purified PCR fragments could be denatured, annealed, and used as a template to generate the final chimera using β1-forward and β4-reverse as PCR primers.
β2/4, containing the extracellular and transmembrane domains of β2 and the intracellular domain of β4, was generated using a similar strategy, substituting rat Scn2b cDNA and the following set of primers for amplification of β2: β2-forward, 5′-ATGCACAGGGATGCCTGG-3′; β2/4-reverse, 5′-CAGCAACAGCACACATTT-3′. The β4 domain was generated with the following primer set: β2/4-forward, 5′-AAATGTGTGCTGTTGCTG-3′; β4-reverse, as described above. The β2/4 chimera was then generated from the purified fragments by PCR using β2-forward and β4-reverse primers.
The resulting chimeras were cloned into pcDNA3.1/Hygro(−) (β1/4) or pcDNA3.1/Zeo(−) (β2/4) for use in mammalian cell expression. The integrity of each construct was confirmed by DNA sequencing. The amino acid sequence of β1/4 is as follows (underlined residues indicate β4 sequence): MGTLLALVVG AVLVSSAWGG CVEVDSETEA VYGMTFKILC ISCKRRSETT AETFTEWTFR GKGTEEFVKI LRYENEVLQL EEDERFEGRV VWNGSRGTKD LQDLSIFITN VTYNHSGDYE CHVYRLLFFD NYEHNTSVVK KIHLEVVDKA NRDASIVSEI MMYVLIVVLT IWLVAEMVYC YKKLITFILK KTREKKKEC1 VSSGNDNTE NGLPGSKAEE KPPTKV. The amino acid sequence of β2/β4 is as follows (underlined residues indicate β4 sequence): MHRDAWLPRP AFSLTGLSLF FSLVPSGRSM EVTVPTTLSV LNGSDTRLPC TFNSCYTV KQFSLNWTYQ ECSNCSEEMF LQFRMKIINL KLERFGDRVE FSGNPSKYDV SVTLKNVQLE DEGIYNCYIT NPPDRHRGHG KIYLQVLLEV PPERDSTVAV IVGASVGGFL AVVILVLMV KCVLLLKKLI TFILKKTREK KKECLVSSSG NDNTENGLPG SLAEEKPPTK V.
Transfection and identification of transfected cells.
HEK–NaV1.1 cells and CA3 pyramidal neurons were transfected with Lipofectamine (Invitrogen) according to Dalby et al. (2004). The cDNA for the β subunit(s) of interest and cDNA for green fluorescent protein (GFP; Clontech) was added to cells in a 5:1 concentration ratio, so that GFP-labeled cells would be highly likely to be transfected with the β subunit. Lipofectamine (4 μl) was added to 50 μl of Opti-MEM (Invitrogen) and incubated for 30 min at room temperature. Separately, 1–2 μg of cDNA was added to 50 μl of Opti-MEM and incubated for 5 min. The cDNA and Lipofectamine solutions were then mixed for 20 min at room temperature. This mixture was added to the cells for 3 h, after which the medium was exchanged. Recordings were made at least 24 h after transfection. GFP-labeled cells were identified with an X-Cite 120 Fluorescence Illumination System (Exfo) on a Nikon Eclipse TE2000-U microscope. Control data from nontransfected cells and cells transfected with GFP alone were indistinguishable and were therefore pooled.
Scn1b and Scn2b null mice.
Scn1b (β1) and Scn2b (β2) null mice were generated as described by Chen et al. (2002, 2004). Congenic strains were created by repeated backcrossing of Scn1b+/− or Scn2b+/− mice to C57BL/6 (Silva et al., 1997) for >10 generations. Double null mice (Scn1b/Scn2b null) were obtained by mating N10 Scn1b+/− with N10 Scn2b−/− mice to generate Scn1b+/−/Scn2b−/− mice. Breeding these mice produced offspring that included the double null mutants. Experiments were done blind to genotype. Genotypes of all mice from which neurons were isolated for electrophysiological recordings were confirmed by PCR (Chen et al., 2002, 2004).
Purkinje cell dissociation.
Cerebellar Purkinje cells were acutely dissociated from Scn1b and Scn2b null and littermate control mice (P14–P19) as described previously (Raman et al., 1997; Grieco et al., 2005). Briefly, mice were anesthetized with halothane, and the superficial layers of the cerebellum were removed and minced in ice-cold dissociation solution containing 82 mm Na2SO4, 30 mm K2SO4, 5 mm MgCl2, 10 mm HEPES, 10 mm glucose, and 0.001% phenol red (pH 7.4 with NaOH). The tissue was incubated in dissociation solution (3 mg/ml protease XXIII, pH 7.4) for 7 min at 31°C with 100% oxygen blowing over the surface, washed, and microdissected in dissociation solution with 1 mg/ml each of bovine serum albumin and trypsin inhibitor, pH 7.4. The tissue was then transferred to Tyrode's solution (in mm: 150 NaCl, 4 KCl, 2 CaCl2, 2 MgCl2, 10 HEPES, and 10 glucose, pH 7.4, with NaOH). Individual neurons were released by trituration with polished Pasteur pipettes. Cells settled in the recording dish for 1 h, and recordings were made 1–6 h after trituration.
Recording and analysis.
Borosilicate pipettes (1.8–3 MΩ; A-M Systems) were coated with Sylgard or wrapped with Parafilm and filled with intracellular solution containing (in mm) 108 CsCH3SO3, 1.8 NaCl, 1.8 MgCl2, 9 HEPES, 1.8 EGTA, 48 sucrose, 4.5 tetraethylammonium (TEA)-Cl, 14 Tris-creatine PO4, 4 MgATP, and 0.3 TrisGTP (pH 7.4 with CsOH). For experiments on Purkinje cells, the same solution was used, but without TEA and with 9 NaCl. As indicated, the β4 peptide (200 μm; Open Biosystems), which consists of amino acids 154–167 from the cytoplasmic tail of the β4 protein (Grieco et al., 2005), was added to the intracellular solution. Whole-cell voltage-clamp recordings were made with an Axopatch 200B amplifier (Molecular Devices). Series resistance was compensated >85%. Data were obtained with pClamp 9.0 (Molecular Devices). Solutions were exchanged by positioning cells in front of a pair of gravity-driven flow pipes. For HEK–NaV1.1 and CA3 cells, the first pipe contained either Tyrode's solution with (in mm) 10 TEA-Cl, 0.3 CdCl2 or 150 NaCl, 10 TEA-Cl, 0.3 CdCl2, 2 BaCl2, 10 HEPES, and 10 glucose. The second pipe contained the control solution, but with 900 nm TTX (Alomone Labs) to block Na currents. For Purkinje cells, pipes contained the same solutions, except that NaCl was reduced to 50 mm and replaced with equimolar TEA-Cl to achieve better voltage clamp of the very large somatic Na currents. Currents were not leak subtracted. To isolate TTX-sensitive Na current, raw records obtained in TTX were subtracted from those recorded in control solutions (without TTX). Cells that showed evidence that leak currents changed between recordings in control and TTX were discarded.
Data were analyzed with IGOR-Pro (Wavemetrics) and are reported as mean ± SEM. Conductance–voltage plots were measured by dividing peak currents in each cell by the driving force to yield conductances, which were normalized and fit with a Boltzmann function, G/Gmax = 1/(1 + exp(−(V − V1/2)/k)), where G is conductance, Gmax is the maximal conductance, V1/2 is the half-maximal voltage of activation, and k is the slope factor. Voltage control in each cell was assessed by verifying that conductance–voltage plots were continuous and that the rise time of currents evoked by step depolarizations decreased gradually. Cells without these attributes were excluded from the analysis. Steady-state inactivation (availability) curves were normalized to peak current and fit with a modified Boltzmann of the form I = Fss+ (1 − Fss)/(1 + exp((V − V1/2)/k)), where I is the normalized current at 0 mV, V1/2 is the half-maximal voltage of inactivation, k is the slope factor, and Fss is the fraction of steady-state or noninactivating current (reported as a percentage). The percentage of persistent or resurgent current was calculated by dividing the current amplitude by the peak transient current at 0 mV. If conductance was not maximal at 0 mV, the current at 0 mV was corrected by dividing by the fractional conductance at 0 mV. Statistical significance was assessed with Student's two-tailed t tests, the Mann–Whitney U test, or two-way repeated-measures ANOVAs, with cell condition and voltage as variables (SPSS); main effects are reported. Cells from which data were not obtained at all voltages were excluded from ANOVAs but were included in plots of mean data. In Figure 1C, ANOVAs included measurements only at voltages from −40 to 0 mV because data was not gathered at +10 and +20 mV in all cells. For all statistical tests, p values are reported, and significance is taken as 0.05. Capacitative artifacts have been digitally reduced in most figures.
Coimmunoprecipitation.
HEK-293T cells, cultured as described previously (Rusconi et al., 2007), were transfected with 10 μg of hNav1.1-pCDM8 (Rusconi et al., 2007) plus 4 μg of rat β1-V5-pcDNA3.1-Hygro− (β1 cDNA containing a C-terminal V5 epitope tag) or 4 μg of rat C121Wβ1-V5-pcDNA3.1/Hygro(−) (β1C121W cDNA containing a C-terminal V5 epitope tag) or rat β4-pcDNA3.1/Hygro(−) using the calcium phosphate method (Rusconi et al., 2007) and allowed to grow for 36 h. Protein A- or protein G-Sepharose beads (Sigma) were prepared by washing three times with PBS at 4°C, followed by resuspension in 250 μl of dilution buffer [DB; 60 mm Tris/HCl, pH 7.5, 180 mm NaCl, 1.25% Triton X-100, and 6 mm EDTA, pH 8, containing Complete Mini protease inhibitor tablets (Roche) at two times the manufacturer's recommended concentration] at 4°C. The beads were then incubated overnight at 4°C with 4 μg of polyclonal (rabbit) pan Na+ channel antibody (Sigma) or with 4 μg of monoclonal (mouse) anti-V5 antibody (AbD; Serotec), or with 4 μg of rabbit or mouse IgG as a negative control. The supernatant was removed, and the beads were resuspended in DB containing 2% bovine serum albumin followed by incubation for 3 h at 4°C to saturate nonspecific binding sites. The transfected cells were detached from the culture dishes using 50 mm Tris and 10 mm EGTA, pH 8, and centrifuged at 5000 rpm in a microfuge for 10 min at 4°C. The cell pellet was resuspended in DB for cell lysis. After 30 min of lysis on ice, a 10 min centrifugation at 10,000 rpm in a microfuge was performed to remove insoluble material. The resulting supernatant was added to the beads and incubated by rotating end-over-end for 3 h at 4°C. The beads were then washed three times with washing buffer (50 mm Tris, pH 7.5, 150 mm NaCl, 0.1% Triton X-100, 0.02% SDS, and 5 mm EDTA, pH 8, containing Complete Mini protease inhibitor tablets at two times the manufacturer's recommended concentration), followed by one wash with the same buffer but without Triton X-100. Samples were then separated by SDS-PAGE on 8.5% polyacrylamide gels and transferred to nitrocellulose for Western blot analysis as described previously (Brackenbury et al., 2008). Immunoblots were probed with anti-V5 or anti-β4 antibody (Wong et al., 2005), as indicated, and detected with Westfemto Chemiluminescent reagent (Pierce).
Results
β4 expression increases persistent Na current in HEK–NaV1.1 cells
To test the influence of the Na channel β4 subunit on the properties of Na currents, TTX-sensitive currents were recorded from human embryonic kidney (HEK) cells stably expressing NaV1.1 with and without transfection of β4. Activation curves were obtained by evoking transient Na currents from −110 mV with depolarizing steps in 5 mV increments. Figure 1A (top) illustrates currents and the activation curve from a representative control cell, along with a conductance–voltage curve from a cell transfected with β4. Voltage control was assessed in each cell (see Materials and Methods), and all cells included for analysis are shown in supplemental Figure 1 (available at www.jneurosci.org as supplemental material). Neither the slope factors (k) nor the maximum conductances (Gmax) were significantly changed by the expression of β4 (Fig. 1A, bottom) (control vs β4: k = 4.9 ± 0.3 vs 5.2 ± 0.3 mV, p = 0.7; Gmax = 45 ± 8 vs 38 ± 7 nS, p = 0.4; n = 15, 13). The half-maximal voltage of activation (V1/2), however, was sensitive to β4 expression. In control cells, V1/2 was −16.8 ± 0.9 mV, and β4 expression negatively shifted this value to −21.8 ± 1.0 mV (p = 0.001) (Fig. 1A, bottom). Boltzmann curves with mean fit parameters (Fig. 1A, bottom right) illustrate the average effect of β4 expression on activation. The observation that the properties of Na currents were modified after β4 transfection suggests that β4 was successfully incorporated into channel complexes. A similar leftward shift of the activation curve has been reported when β4 is coexpressed with NaV1.2 in tsA-201 cells (Yu et al., 2003).
Expression of β4 negatively shifts activation and increases the noninactivating component of Na currents in HEK–NaV1.1 cells. A, Top left, Representative TTX-sensitive Na currents (bottom) evoked by the activation voltage protocol (top). Top right, Single representative activation curves, plotted as normalized conductance against voltage for a control cell (circles) and a cell transfected with β4 (triangles). Parameters are as follows: control: V1/2 = −15.6 mV, k = 5.7 mV; β4: V1/2 = −20.6 mV, k = 5.7 mV. Bottom left, Mean parameters of fits to data from control (n = 15) and β4-transfected (n = 13) cells. Left bars, V1/2; middle bars, k; right bars, Gmax. Asterisks in all figures indicate statistical differences from control, unless indicated otherwise. Transfection with β4 significantly hyperpolarized the V1/2 of activation. Bottom right, Boltzmann functions with mean fit parameters. B, Top left, Representative TTX-sensitive Na currents (bottom) evoked by the inactivation voltage protocol (top). Top right, Single representative inactivation curves, plotted as normalized current (availability) at 0 mV against conditioning potential, for a control cell (circles) and a cell transfected with β4(triangles). Parameters are as follows: control: V1/2 = −43.6 mV, k = 6.4 mV; percentage noninactivating is 1.1%; β4: V1/2 = −44.4 mV, k = 7.4 mV; percentage noninactivating is 7.2%. Bottom left, Mean parameters of fits to data from control (n = 15) and β4-transfected (n = 13) cells. Left bars, V1/2; middle bars, k; right bars, percentage of noninactivating current. β4 transfection increased the steady-state component of the inactivation curve. Bottom right, Boltzmann functions with mean fit parameters. C, Top, Voltage protocol and representative traces evoked in the absence and presence of β4 subunit expression. The box indicates the region of the trace in which persistent current was measured. Bottom, Persistent current measured as the mean current in the last 10 ms of each 100 ms step, normalized to the peak transient current at 0 mV and plotted versus voltage. β4-transfected cells (n = 13) had significantly more persistent current than control cells (n = 15). Error bars indicate SEM.
Next, we measured steady-state inactivation (availability) of Na currents by applying 100 ms conditioning steps followed by steps to 0 mV. Peak currents evoked at 0 mV were normalized, plotted as availability versus voltage, and fit with Boltzmann functions modified to incorporate a noninactivating component. Data for a representative control cell, along with availability for a β4-transfected cell, are shown in Figure 1B (top). Mean values for fits and Boltzmann curves with mean fit parameters are shown in Figure 1B (bottom). With β4 expression, V1/2 shifted slightly negative, and the slope factor increased slightly (control vs β4: V1/2 = −42.1 ± 0.7 vs −45.7 ± 1.7 mV, p = 0.07; k = 6.4 ± 0.4 vs 8.1 ± 1.0 mV, p = 0.13; n = 15, 13). The most substantial change, however, was that the β4-transfected cells had an unusually large noninactivating component (control vs β4: 1.8 ± 0.3 vs 5.8 ± 1.4%, p = 0.02). These results therefore suggest that β4 expression in HEK–NaV1.1 cells destabilizes fast inactivation and instead favors channel opening at 0 mV.
Given the increase in availability measured at 0 mV, we tested whether β4 influenced the amount of persistent Na current at other voltages by measuring the mean current in the last 10 ms of the 100 ms conditioning steps (Fig. 1C). Although the transient current amplitudes were similar with and without β4 (control vs β4: −4.3 ± 0.8 vs −3.9 ± 0.7 nA at 0 mV; n = 15, 13), the persistent current was increased more than threefold in the presence of β4 (control vs β4: −46 ± 13 vs −143 ± 27 pA at −10 mV). To control for variations in current density across cells, the persistent current amplitude was normalized to the peak amplitude of the transient current at 0 mV in each cell (corrected as necessary to represent maximal conductance; see Materials and Methods), and the percentage of persistent current was plotted against voltage. Consistent with the availability curve, β4-transfected cells showed more persistent current compared with control cells, with the increase being most pronounced at voltages between −20 and 0 mV (Fig. 1C, open triangles) (significant main effect of condition on current; F(1,26) = 6.9; p = 0.014).
β4 expression is not sufficient to produce resurgent Na current in HEK–NaV1.1 cells
Previous work from our group demonstrated that a 14 amino acid sequence from the β4 intracellular domain (the “β4 peptide”) can bind open Na channels of neurons in a voltage-dependent manner. Channels are blocked by the peptide at positive voltages and become unblocked after repolarization, allowing resurgent Na current to flow, both in Purkinje neurons after enzymatic removal of their endogenous open-channel blocking protein, or in CA3 neurons, which lack endogenous block (Grieco et al., 2005). To test whether expression of the full-length β4 protein might replicate this behavior in HEK–NaV1.1 cells, we stepped cells to +60 mV to maximize the possibility of voltage-dependent block and repolarized to potentials between −40 and +20 mV. As expected, in control cells, little current was detectable after repolarization (Fig. 2A). To verify that NaV1.1 channels expressed in HEK cells were capable of undergoing block and unblock in a manner similar to Na channels in their native neuronal environments, we included the β4 peptide (200 μm) in the recording pipette. In the presence of the peptide, repolarization indeed evoked a current with kinetics resembling native resurgent current in neurons, but with a maximal resurgent current between −10 and −20 mV rather than at −30 mV as in neurons (Fig. 2A). This shift is consistent with the fact that the V1/2 of activation is ∼15 mV more depolarized than in neurons (Raman and Bean, 1997). In contrast, with expression of the full-length β4 protein, little, if any, resurgent current was evident. Instead, repolarization evoked a brief tail followed by a large steady-state current (Fig. 2A), consistent with the increased noninactivating component observed with step depolarizations.
Expression of the β4 subunit increases persistent, but not resurgent, Na current. A, Voltage protocol and representative traces for each condition, as labeled. Traces were normalized to the peak current evoked at 0 mV in each cell. B, Top, Mean peak currents evoked during repolarization after subtraction of the steady-state current at the end of each trace. Currents were normalized to the peak transient current evoked in each cell at 0 mV and plotted versus voltage. Control, n = 15; β4 peptide, n = 10; β4 protein, n = 13. Bottom, Mean persistent currents, measured as the mean steady-state current in the last 10 ms of the repolarizing step. Currents were normalized and plotted as in the top panel. Same cells as in the top panel. C, Top, Voltage protocol (top) and representative control traces (middle) for assaying recovery from inactivation after conditioning at +60 mV. The first step to 0 mV in the voltage protocol is the reference step. Bottom, The conditioning current and first test current at higher gain. Mean peak currents evoked by test steps after a conditioning step to +60 mV, normalized to the peak current evoked by the reference step and plotted versus recovery interval, are shown. Recovery was faster in β4 peptide cells (n = 4) compared with either control (n = 5) or β4 protein (n = 5) cells. Error bars indicate SEM.
To quantify these data, the resurgent current was calculated as the difference between the maximal current (after the tail) and the persistent current (at the end of the step). At −10 mV, the current was −26 ± 9 pA in control (n = 15), −166 ± 38 pA with the β4 peptide (n = 10), and −30 ± 5 pA with the β4 protein (n = 13). Because the mean transient current density in the cells with the peptide (−5.3 ± 0.4 nA at 0 mV) was slightly greater than in the control and β4 protein-expressing cells (same cells as in Fig. 1), resurgent current amplitudes were normalized to the peak transient current at 0 mV and plotted as a percentage of resurgent current against voltage (Fig. 2B, top). Cells containing the β4 peptide indeed had more relative resurgent current than control cells, consistent with an effective block and unblock of Na channels by the β4 peptide (significant main effect; F(2,35) = 8.3; p = 0.001; Tukey's post hoc test: β4 peptide vs control, p = 0.001; vs β4 protein, p = 0.07). The current was not significantly different in control and β4-expressing cells, however (Tukey's post hoc test, p = 0.2). In contrast, expression of β4 increased the persistent current flowing after repolarization (control, −48 ± 14 pA; β4 peptide, −185 ± 74 pA; β4 protein, −131 ± 24 pA). When persistent currents in each cell were normalized to the transient current at 0 mV, larger relative currents consistently occurred in the presence of the β4 subunit (Fig. 2B, bottom) (significant main effect; F(2,35) = 3.5; p = 0.04; Tukey's post hoc test: β4 protein vs control, p = 0.03; β4 peptide vs control, p = 0.47). Together, these data demonstrate that expression of the full-length β4 subunit with NaV1.1 is not sufficient to generate resurgent current in HEK–NaV1.1 cells, consistent with the recent report that β4 does not induce a resurgent current in tsA-201 cells expressing NaV1.2 (Chen et al., 2008). Instead, like the results obtained with step depolarizations, the results suggest that coexpression of β4 with NaV1.1 destabilizes inactivated states at voltages between −30 and +10 mV, permitting a higher occupancy of open states than does expression of NaV1.1 alone.
At strongly hyperpolarized potentials, however, expression of β4 did not modify the transition from inactivated to closed states. Figure 2C illustrates recovery at −110 mV from inactivation induced by a short step to +60 mV in control cells, cells containing the β4 peptide, and cells expressing the β4 protein. Recovery in cells expressing the β4 protein was indistinguishable from control (double-exponential fit parameters to mean data from control, β4 protein, β4 peptide: τfast = 1.4, 1.5, 1.2 ms; τslow = 16.6, 12.9, 9.1 ms; %fast = 44, 42, 47%; n = 5, 5, 4). The similarity of recovery times between the control and the β4 protein condition supports the idea that depolarizing steps favor the same fast inactivated states in both cases and that β4 reduces the stability of inactivation only at voltages in the vicinity of 0 mV. In contrast, channels exposed to the β4 peptide recovered more rapidly than in control conditions, consistent with the idea that block by the peptide prevents fast inactivation at positive potentials and permits reopening and deactivation at negative potentials (Raman and Bean, 2001; Grieco et al., 2005).
β1 coexpression counteracts the effect of β4 on inactivation
Expression of β4 increased persistent sodium current beyond 5% of the transient sodium current, a value that is considerably greater than anything observed in central neurons, in which persistent currents have been reported to range from 0.7 to 4% of transient currents (Cummins et al., 1994; Parri and Crunelli, 1998; Magistretti and Alonso, 1999; Maurice et al., 2001; Taddese and Bean, 2002). It therefore seemed likely that, in neurons, other factors might limit the ability of β4 to enhance persistent current. Given the widespread expression of β1 in the nervous system, we considered the possibility that β1, which accelerates inactivation in heterologous expression systems (Isom et al., 1992; Chen and Cannon, 1995), might modify the influence of β4. To test this idea, we first transfected HEK–NaV1.1 cells with the Na channel β1 subunit (n = 9) and measured currents evoked by 100 ms step depolarizations (Fig. 3A, top). With β1 transfection, the amplitudes of persistent currents were similar to control (Fig. 3B, left). As in other heterologous expression systems, β1 expression accelerated fast inactivation, reducing the decay time constant at 0 mV from 0.72 ± 0.03 ms in control (n = 26) to 0.63 ± 0.03 ms (n = 9; p = 0.005) (Fig. 3C, left). Thus, β1 and β4 have contrasting effects on the macroscopic properties of fast inactivation.
Coexpression of wild-type β1 subunit, but not the GEFS+ mutant subunit β1C121W, prevents the β4-mediated destabilization of inactivation. A, Voltage protocol to elicit persistent current and representative traces evoked with β subunit transfection as labeled. The box indicates the region of the trace analyzed in B. B, Persistent currents measured as in Figure 1C, plotted versus voltage. Mean persistent currents for control (dashed line) and β4-transfected (dotted line) are replotted for comparison. Left, β1-transfected (n = 9) cells have small persistent currents, similar to control cells. Right, The increase in persistent current by β4 was prevented by coexpression of β1 (n = 14). Coexpression of β1C121W with β4 (n = 16) prevented the increase in persistent current nearly as well as wild-type β1. C, Top, Representative Na currents evoked by a 5 ms step to 0 mV for five conditions, as labeled. Bottom, Mean time constants from single exponential fits to currents evoked as in A (left bars) and the percentage of current remaining at the end of the 5 ms step (right bars) are shown for each condition for control cells (n = 26) and for cells transfected with β1 (n = 9), β4 (n = 20), β1+β4 (n = 14), and β4+β1C121W (n = 17). Asterisks indicate significant differences from control. Error bars indicate SEM.
This contrast is of interest because it is likely that α, β1, and β4 subunits assemble to form heterotrimeric channels: biochemical studies indicate that the majority of Na channel α subunits in the brain associate with one noncovalently linked β subunit, such as β1, and one covalently linked β subunit, such as β4 (Reber and Catterall, 1987; Yu et al., 2003). Therefore, to test how β1 and β4 subunits interact functionally, we transfected HEK–NaV1.1 cells with both β1 and β4. Coexpression of β1 and β4 (n = 14) produced persistent current amplitudes that overlapped with those of control cells (n = 15) or cells transfected with β1 (n = 9) (Fig. 3A,B, right). Moreover, the decay time constant at 0 mV was significantly faster in cells coexpressing β1 and β4 (Fig. 3C, left) (n = 14; 0.58 ± 0.03 ms; p < 0.001 vs control), similar to β1 expression alone. These changes indicate that, despite the presence of endogenous β1B, expression of β1 has a distinct and specific effect on β4, namely, to inhibit the β4-mediated destabilization of inactivation.
An epilepsy mutation in Scn1b decreases the ability of β1 to reverse the effects of β4
These results raise the possibility that disruptions of β1, such as those that occur in epilepsy and other types of seizure disorders, might alter the regulation of β4. Specifically, a mutation of a cysteine to a tryptophan in the β1 extracellular domain (β1C121W) leads to GEFS+ in humans (Wallace et al., 1998, 2002). To test whether this mutation changes the influence of β1 on β4, we measured Na currents in HEK–NaV1.1 cells, in which β1C121W was coexpressed with β4. β1C121W prevented the β4-induced increase in persistent current as effectively as did β1 (at −10 mV: β4 alone, 6.0 ± 2.0%; β1+β4, 1.9 ± 0.4%; vs β1C121W+β4, 1.9 ± 0.5%; n = 13, 14, 14) (Fig. 3A,B, right). Inspection of the traces, however, indicated that the inactivation time constant was slower when β4 was coexpressed with β1C121W than with wild-type β1, resembling the condition with β4 alone (β1C121W+β4, 0.76 ± 0.06 ms, n = 17; vs β1+β4, p = 0.01). This difference in the rate of entry into inactivated states was even more apparent after examination of the percentage of current remaining at the end of a 5 ms step. This amplitude, which reflects a slower component of inactivation, was relatively small in control or with β1 alone, but relatively large with β4 alone (Fig. 3C). Moreover, when β4 was expressed with β1C121W, the current was nearly twice that with the wild-type β1 (β1+β4 vs β1C121W+β4; 8.7 ± 1.4% vs 15.7 ± 2.9%; n = 14, 17; p = 0.04). Thus, the GEFS+ mutation makes the β1 subunit less effective in counteracting the destabilization of inactivation by β4, raising the possibility that a prolongation of Na currents contributes to the alteration of neuronal firing patterns in carriers of this mutation.
β1 subunits might exert their effects on persistent current either by preventing β4 from associating with α subunits and/or by having a dominant influence on channel gating. To test the likelihood that NaV1.1, β1, and β4 form functional heterotrimeric complexes, we transfected HEK-293T cells with different combinations of subunits and assessed their association by coimmunoprecipitation. Cotransfection of NaV1.1 with V5-tagged β1 indicated that these subunits associated (Fig. 4A) and that this association persisted but was weakened when β4 was also present (Fig. 4B). Transfection of cells with only β1 and β4 revealed a strong interaction between these two subunits even in the absence of NaV1.1 (Fig. 4C), suggesting that one action of β1 may indeed be to sequester β4 and limit its association with α subunits. If so, the macroscopic electrophysiological properties measured in HEK–NaV1.1 cells transfected with both β subunits may result in part from NaV1.1 monomers, thus mimicking the control condition. β4, however, interacted strongly with NaV1.1 alone (Fig. 4D), suggesting that any free β4 would be likely to enter a channel complex. Because the macroscopic currents in cells expressing the three subunits mimic neither the control condition nor the condition with a single β subunit, it seems likely that a non-negligible subset of channels contain NaV1.1, β1, and β4 and that these heterotrimers contribute to the overall electrophysiological phenotype. Consistent with this idea, with all three subunits present, an interaction of β4 with NaV1.1 was evident (data not shown).
Association of Nav1.1, β1, β1C121W, and β4 subunits. Coimmunoprecipitation experiments of Na+ channel α and β subunits were performed on transfected HEK-293T cells. All molecular weight standards are indicated in kilodaltons. A, Nav1.1 associates with β1 and with β1C121W. Cells expressing Nav1.1+β1-V5 or Nav1.1+β1C121W -V5 were immunoprecipitated with anti-pan Na+ channel antibody or rabbit IgG. The immunoblot was probed with anti-V5 antibody to detect β1-V5 or β1C121W -V5. Immunoreactive β1 bands are indicated by the arrow. B, Nav1.1 and β1 or β1C121W associate in the presence of β4. Cells expressing Nav1.1+β1-V5+β4 or Nav1.1+β1C121W-V5+β4 were immunoprecipitated with anti-pan Na+ channel antibody or rabbit IgG. The immunoblot was probed with anti-V5 antibody to detect β1-V5 or β1C121W-V5. Immunoreactive β1 bands are indicated by the arrow. C, β1 or β1C121W associates with β4 in the absence of α subunits. Cells expressing β1-V5+β4 or β1C121W-V5+β4 were immunoprecipitated with anti-V5 antibody or mouse IgG. The immunoblot was probed with anti-β4 antibody. Immunoreactive β4 bands are indicated by the arrow. D, Association of Nav1.1 and β4. Cells expressing Nav1.1+β4 were immunoprecipitated with anti-pan Na+ channel antibody or rabbit IgG. The immunoblot was probed with anti-β4 antibody. Immunoreactive β4 bands are indicated by the arrow. A lane containing rat brain membranes, prepared as by Brackenbury et al. (2008), is included as a positive control to show β4 immunoreactivity. IP, Immunoprecipitation; IB, immunoblot.
Next, we repeated these experiments with β1C121W substituted for the wild-type β1 subunit. The mutant subunit associated with NaV1.1 and, in contrast to wild-type β1, this association remained strong in the presence of β4 (Fig. 4A,B). Conversely, the association of β1C121W and β4 in the absence of the α subunit appeared less robust than with wild-type β1. This result suggests that the β1–β4 interaction is mediated by the extracellular Ig domains (Fig. 4C) and is consistent with previous results showing that β1C121W does not function as a cell adhesion molecule (Meadows et al., 2002). Considered in the context of the electrophysiological experiments, which demonstrated reduced persistent current but slowed inactivation rates relative to control, these results support the idea that heterotrimeric channels comprising α, β1C121W, and β4 do indeed assemble and suggest that wild-type β1 need not prevent β4 association with the α subunit to oppose the effects of β4 on gating.
The extracellular domain of β1 regulates persistent current
Because the C121W mutation is in the extracellular domain, these data suggested that this region of β1 is necessary for the normal regulation of inactivation. We therefore tested whether the suppression of persistent current and promotion of inactivation could be achieved without the intracellular domain of β1 by expressing a “β1/4” chimeric subunit, which consisted of the extracellular and the transmembrane domains of β1 and the intracellular domain of β4. In cells expressing β1/4 (n = 7), both the persistent current amplitudes and the time course of inactivation were indistinguishable from β1-expressing cells (Fig. 5A, top left, B). The simplest interpretation of these results is that the wild-type extracellular domain directly modulates the stability of inactivation. The extracellular domain, however, also contains sites required for interactions with the α subunit, thereby determining the position of β subunits in the channel complex (McCormick et al., 1998). Because the sites of α–β interaction are likely to differ for β1 and β4, an alternative interpretation is that the β1/4 chimera inhibits the channel openings that are favored by β4 simply by wrongly positioning the β4 intracellular domain.
Chimeric β subunits suggest that the extracellular domain regulates persistent current amplitude. A, Persistent currents measured as in Figure 1C, plotted versus voltage. Mean persistent currents for control (dashed line) and β4 transfected (dotted line) are replotted for comparison. The percentage of persistent current is plotted versus voltage for NaV1.1 with expression of the following subunits: top left, either β1 (n = 9) or the β1/4 chimera (n = 7); top right, either β2 (n = 12) or the β2/4 chimera (n = 9); bottom left, β1+β4 (n = 14) or β1/4+β4 (n = 11). B, Time constants from single exponential fits to the decay of transient current at 0 mV (top) and the percentage of current remaining at the end of the 5 ms step (bottom) as in Figure 3C for cells transfected with β1/4 (n = 7), β2 (n = 12), β2/4 (n = 9), and β1/4+β4 (n = 11). Relevant data for control, β1, β4, and β1+β4 are included for comparison. Asterisks indicate significant differences from control. Error bars indicate SEM.
To address this possibility, we coexpressed β1/4 and β4. With both subunits present, heterotrimeric channels are predicted to have two β4 intracellular domains, one in the site normally occupied by the β1 cytoplasmic tail and one in the normal position for β4. Under these conditions, the persistent current remained at control levels and was indistinguishable from coexpression of β1+β4 (n = 11) (Fig. 5A, bottom). These results support the idea that the extracellular domain of β1 primarily governs persistent current in NaV1.1, as it does with other α subunits (Chen and Cannon, 1995; McCormick et al., 1998, 1999). The slow phase of inactivation, reflected by the percentage of current remaining at 5 ms, was also restored to control levels, as it was with β1+β4. The fast inactivation time constant was not consistently reduced to levels achieved by expression of β1+β4 (Fig. 5B), however, leaving the possibility open that intracellular domains also contribute to the regulation of inactivation (Spampanato et al., 2004).
Next, we tested whether the β2 subunit, which resembles β4 both in sequence similarity and in its disulfide linkage to Na channel α subunits (Yu et al., 2003), might also resemble β4 in its influence on Na current. Indeed, expression of β2 in HEK–NaV1.1 cells increased persistent current amplitudes (n = 12), although to a lesser extent than β4 (Fig. 5A, top right). Expression of β2 also increased the percentage of current remaining at 5 ms, while leaving the inactivation rate unaffected relative to control (Fig. 5B). A β2/4 chimera (n = 9), composed of the extracellular and the transmembrane domains of β2 and the intracellular domain of β4, behaved in much the same way as β2 and β4 (Fig. 5A, top right, B). These data indicate that β2, β4, and β2/4, the extracellular domains of which are expected to bind in a similar way to the α subunit, influence inactivation in a qualitatively similar manner.
Together, the data indicate that persistent current amplitudes, as well as the percentage of current remaining at 5 ms, can be either relatively large (occurring with β4, β2, or β2/4) or relatively small (occurring with β1, β1/4, β1+β4, or β1/4+β4, as well as with NaV1.1 alone). For convenience (with no mechanistic implication), the former group will be referred to as the “disulfide-linked” group, and the latter (excluding the control) will be referred to as the “wild-type β1extra” group. The β1C121W+β4 condition presents an anomaly that will be considered separately.
β subunits affect both window current and the percentage of noninactivating current
To explore the basis for the differences in persistent current between the disulfide-linked and wild-type β1extra groups, we examined the activation and inactivation curves. These allow an estimation of the size of the window current between the curves, as well as the percentage of current that does not inactivate even at the most depolarized potentials. We began by analyzing the availability curves recorded in all conditions. As in Figure 1, data from each cell were fitted with Boltzmann functions to obtain values of V1/2, k, and percentage of noninactivating current. The mean values for the noninactivating current fell into two groups. The wild-type β1extra group and control cells had <2.3% current, whereas the disulfide-linked group and β1C121W+β4 had >3.4% current (Fig. 6A). These data suggest that the disulfide-linked subunits actively increase the equilibrium occupancy of the open state, whereas the subunits with the wild-type β1 extracellular domain counteract this effect. With the exception of β1C121W+β4, this grouping parallels the amplitudes of persistent currents measured at negative voltages.
Changes in activation and inactivation parameters increase the window current in HEK–NaV1.1-expressing β4 and β2 but not β1. A, Inactivation parameters were estimated as in Figure 1B, and the noninactivating component is shown for all conditions, as labeled. The n values are as in Figures 3 and 4. B, V1/2 of inactivation for all conditions; it was not correlated with the percentage of noninactivating current (R2 = 0.06). C, The noninactivating component versus the inactivation slope factor, for all conditions, as labeled. The dashed line is the linear fit to the data, with R2 = 0.79. D, Activation parameters were estimated as in Figure 1A, and the activation k is shown for all conditions, as labeled. The n values are as follows: control (α), 15; β4, 13; β2/4, 9; β2, 11; β1/4, 5; β1+β4, 13; β1, 9, β1CW+β4, 13; β1/4+β4, 11. Activation k did not correlate with the percentage of noninactivating current (R2 = 0.11). E, The noninactivating component of the availability curve versus the activation V1/2. The dashed line is the linear fit to the data, with the point corresponding to β1C121W+β4 (open symbol) excluded. R2 = 0.67. F, Conductance and availability curves with the mean activation and inactivation parameters for β1 (thin line) and β4 (thick line) in HEK–NaV1.1 cells, illustrating the larger window and noninactivating current with β4 relative to β1. G, Conductance and availability curves as in F for β1+β4 and β1CW+β4. The inset illustrates the difference in the time course of inactivation between β1+β4 and β1C121W+β4 (same traces as in Fig. 3C). Error bars indicate SEM.
Next, we tested whether β subunits expressed in HEK–NaV1.1 cells modified the window current. In general, the window current may be increased by a negative shift in the activation curve, a positive shift in the inactivation curve, and/or a flattening of the slope of either curve. As shown in Figure 6B, the V1/2 of inactivation was relatively insensitive to β subunit expression. In nearly all conditions, the mean V1/2 fell between −41 and −44 mV. The value for β4 alone was slightly negative to this range (−45.7 ± 1.7 mV) and for β1C121W+β4 was slightly positive to this range (−39.7 ± 1.6 mV). The slope factor of the curve, k, however, was indeed affected by the different β subunits. The four conditions with more noninactivating current had shallower slopes, with k values ≥7.5 mV (disulfide linked and β1C121W+β4), whereas the five conditions with less noninactivating current had steeper slopes, with k values ≤6.5 mV (wild-type β1extra and control) (Fig. 6C). Consequently, the inactivation k was correlated with the noninactivating current (R2 = 0.79). Because the inactivation V1/2 is relatively constant across conditions, shallower slopes widen the voltage range over which window current can flow. For instance, with a V1/2 of −42 mV, shifting the slope factor from 6.25 mV to 7.5 mV doubles the availability at −15 mV. The overlay of availability curves with the mean fit parameters of β1 and β4 illustrates this effect (Fig. 6F). These data therefore suggest that persistent current across a range of potentials may be promoted in the disulfide-linked subunits by both a weaker voltage sensitivity of inactivation and a greater equilibrium stability of the open state.
Across conditions, the complement of β subunits affected the parameters of activation as well. The slope factors covered a relatively wide range of values but were not correlated with the amount of noninactivating current (R2 = 0.11) (Fig. 6D). The steepest slopes occurred with the chimeras (β2/4, β1/4) and the coexpressed subunits (β1+β4, β1/4+β4), suggesting that, at least in some contexts, the intracellular domain of β4 makes channel opening more sensitive to voltage. This effect is particularly noticeable when β1 is compared with β1/4 (p = 0.07) or β2 is compared with β2/4 (p = 0.02). An exception, however, is the moderate k value of β4 alone (5.2 ± 0.3), indicating that the effects of the intracellular and extracellular domains are not altogether independent of their context in a full protein. Excluding the data for β1C121W+β4, the V1/2 for activation was negatively correlated with the percentage of noninactivating current (R2 = 0.67) (Fig. 6E). The V1/2 value was most negative for β4 (−21.8 mV). Also, when β4 was coexpressed with either β1 or β1/4, the V1/2 tended to shift negative relative to the value for β1 or β1/4 alone. A hyperpolarization of V1/2 also occurred when the β4 tail was added to β2 to make the β2/4 chimera. Thus, expressing the β4 intracellular domain positioned correctly (i.e., on a disulfide-linked subunit) promoted channel opening at more negative potentials. The resulting shift in the activation curve is expected to expand the window in which persistent current can flow (Fig. 6F).
Coexpressing β1C121W and β4 provided an exception to the generalizations that pertained to the other subunits. β1C121W and β4 generated channel complexes that resembled the disulfide-linked group in their large noninactivating components and shallow slope of the availability curves, as well as in their large-percentage current remaining after 5 ms. Nevertheless, they generated small persistent currents, as did the wild-type β1extra group. The activation parameters, however, offered a likely explanation for the peculiarities of the mutant subunit. The activation curves had properties at the extreme of the distribution, with the largest k (5.8) and most positive V1/2 (−12.5 mV). The depolarizing shift and flattening of the activation curve are expected to diminish persistent current by narrowing the window in which it flows (Fig. 6G). Thus, in HEK–NaV1.1 cells, the GEFS+ mutation in β1 makes it even more effective than the wild-type subunit at stabilizing closed over open states, such that larger depolarizations are necessary to open the channel. At voltages positive enough to activate the channels, however, the mutation renders the β1C121W subunit unable to counteract the β4-induced favoring of open over inactivated states (Fig. 6G).
Together, these data suggest the following. First, expression of wild-type β1 favors inactivated states, and this effect is dominated but not wholly controlled by the extracellular domain. Second, expression of β4 favors open states, and this effect is dominated but not wholly controlled by the intracellular domain. Third, expression of the β1 GEFS+ mutant with β4 generally weakens the overall voltage sensitivity of gating, so that channels remain closed rather than open at moderately negative voltages and they remain open rather than inactivated at more positive voltages.
Na currents in neurons made to overexpress β4
The results in HEK–NaV1.1 cells raise the question of which effects of β subunits are evident in neuronal environments, where Na channel gating is also influenced by factors such as additional associated proteins and posttranslational modifications. To address this issue, we overexpressed the β4 protein in cultured pyramidal neurons from the CA3 region of the hippocampus (Fig. 7A). We selected these neurons because they normally lack β4 (Yu et al. 2003) but express high levels of β1 and β2. They also express β3, which, like β1, binds noncovalently to α subunits (Oh et al., 1994; Morgan et al., 2000; Whitaker et al., 2000). Na channels in CA3 neurons are therefore expected to comprise α subunits (NaV1.1, NaV1.2, or NaV1.6) with β1+β2 or β3+β2, and these heterotrimers likely interact with other neuronal proteins. We reasoned that transfection of neurons with β4 might allow some fraction of channels to incorporate β4 instead of β2 and/or generate a subset of channels that exist as α+β4 heterodimers.
Overexpression of β4 produces a small but consistent increase in persistent current in cultured CA3 hippocampal neurons. A, Left, Cultured CA3 neuron transfected with GFP. Scale bar, 25 μm. Right, Voltage protocol to generate steady-state availability curves and representative Na currents elicited in a β4-transfected neuron. B, Left, Single availability curves from a control and a β4-transfected neuron. Parameters of fits for control and β4 are as follows: V1/2, −53.7, −54.5; k, 4.5, 6.4; percentage noninactivating, 1.3, 0.9. Right, Boltzmann curves with the mean parameters for control and β4-transfected cells. C, Voltage protocol to evoke resurgent and persistent current and representative responses from a control neuron (thin trace) and a β4-transfected neuron (thick trace). Currents are normalized to the peak transient current evoked in each cell at 0 mV. The inset shows the persistent current at a higher gain. D, Cumulative probability plot of the persistent current amplitude in control (n = 14) and β4-transfected (n = 16) cells. The y value indicates the probability of finding a cell in which the persistent current is less than or equal to the x value. The distribution was significantly shifted to larger persistent currents in β4-transfected cells.
Because the complex morphology of cultured neurons made space clamp of transient currents difficult at the foot of the activation curve, we restricted our analysis to transient currents at 0 mV, where conductance is maximal and less affected by voltage escape, and to small, slow currents evoked by repolarization, where voltage clamp is optimal. We first assayed steady-state inactivation after 100 ms conditioning steps in control and transfected neurons (Fig. 7A). Expression of β4 produced small changes in the V1/2 of inactivation and steady-state components of the availability curve (Fig. 7B). The V1/2 was −51.8 ± 2.1 mV in control neurons (n = 5) and −57.6 ± 1.4 mV with expression of β4 (n = 5; p = 0.12), whereas the noninactivating component was 1.0 ± 0.2% in control neurons and 1.9 ± 0.3% with β4 (p = 0.15). Although neither change was statistically significant, the tendency for a negative shift in inactivation and an increase in the noninactivating component resembles the changes seen with transfection of only β4 into HEK–NaV1.1 cells. Furthermore, expression of β4 produced a significant change in the slope factor, which was 5.0 ± 0.3 mV in control and became 6.1 ± 0.2 mV with transfection of β4 (p = 0.025) (Fig. 7B). Again, the decrease in the steepness of the inactivation curve resembles the effects obtained with transfection of β4 alone into HEK–NaV1.1 cells.
Next, to measure persistent and (if any) resurgent current, cells were held at −90 mV, and a step depolarization to +30 mV was applied, followed by a repolarizing step to −30 mV. Currents were measured relative to the transient current amplitude at 0 mV (Fig. 7C). Control neurons showed little, if any, resurgent current, and expression of β4 did not generate a larger resurgent component (control, β4: 0.9 ± 0.2%, 1.1 ± 0.2%; p = 0.53; n = 14, 16). Persistent currents were also small in all neurons but were consistently larger in β4-overexpressing neurons, doubling from 1.1 ± 0.2% in control to 2.2 ± 0.6% with β4. The cumulative probability plot illustrates that the expression of β4 increased the likelihood of a larger persistent component (Fig. 7D) (Mann–Whitney U test; Z = −2.12; p = 0.034). Like the change in the availability curve, the change in persistent current resembles the changes observed when HEK–NaV1.1 cells were transfected with β4 alone. It is therefore possible that transfection of CA3 cells led to the assembly of α+β4 heterodimers. Alternatively, because β3 subunits have been implicated in increasing persistent currents in expression systems (Qu et al., 2001), it may be that α+β3+β4 channels were also formed and that β3, unlike β1, permits β4-mediated destabilization of inactivation.
Na currents in neurons expressing β4 but lacking β1 and/or β2
The changes in Na currents were smaller in hippocampal neurons than in the HEK–NaV1.1 cells. These differences may result from a population of channel complexes that failed to incorporate β4, the expression of other proteins that modulate the effects of β4, and/or the antagonism of β4 effects by endogenous β1. To explore how the removal of β1 affects channels that normally include β4, we recorded from neurons acutely isolated from Scn1b (β1) null and Scn1b/Scn2b (β1/β2) double null mice as well as from littermate wild-type or Scn2b null controls. For these experiments, we selected cerebellar Purkinje neurons, which normally express high levels of β4, as well as β1 and β2 (Yu et al., 2003), but lack β3 (Morgan et al., 2000). A complicating factor in these experiments is the variety of proteins, in addition to β subunits, that are known to modulate Na channels in real neurons (Abriel and Kass, 2005). These include GTP-binding protein βγ subunits (Ma et al., 1997; Mantegazza et al., 2005; Kahlig et al., 2006), calmodulin (Mori et al., 2000; Deschênes et al., 2002; Herzog et al., 2003; Kim et al., 2004; Young and Caldwell, 2005), FGF-homologous factor (Wittmack et al. 2004; Lou et al. 2005), and the endogenous blocking protein of Purkinje cells (Grieco et al., 2005). Nevertheless, we reasoned that comparing the Na currents in the wild-type and null mice might indicate whether the influence of β1 that was present in HEK–NaV1.1 cells might also be evident in Purkinje neurons.
We began by measuring activation curves and comparing parameters of the Boltzmann fits (Fig. 8A). Wild-type and Scn1b null mice showed no significant differences in the V1/2 or k of activation (control vs Scn1b null: V1/2 = −31.9 ± 1.0 vs −34.0 ± 1.4 mV, p = 0.62; k = 5.7 ± 0.3 vs 6.0 ± 0.4 mV, p = 0.3; n = 8, 6). The k values were also statistically indistinguishable in the Scn2b nulls and the double nulls (6.1 ± 0.2 vs 6.2 ± 0.3 mV; p = 0.9; n = 9, 19). In the double nulls, however, activation was shifted significantly negative (from −32.3 ± 1.2 to −36.1 ± 1.3 mV; p = 0.045). Considering the observation that cells that normally lack β3 do not upregulate β3 after loss of β1 (Lopez-Santiago et al., 2007), the Scn1b nulls should reveal the properties of α+β2 and α+β4 channels, and the double nulls should isolate the properties of α+β4 channels. The minor negative shift in the former and the significant negative shift in the latter illustrate that cells in which β4 is the only available β subunit activate more readily in widely differing contexts, from HEK cells to neurons.
Early persistent currents are increased in Scn1b null and Scn1b Scn2b double-null Purkinje neurons. A, Activation parameters were obtained and plotted as in Figure 1A for Purkinje neurons from Scn1b and Scn2b wild-type, heterozygous, or null mice, as labeled. The n values are as follows: Scn1b wild type and heterozygous, 8; Scn1b nulls, 6; Scn2b nulls, 9; double nulls, 19. Asterisks indicate significant differences relative to littermate controls. B, Inactivation parameters were obtained and plotted as in Figure 1B. C, Left, Representative transient currents evoked by a voltage step to 0 mV for the four conditions. Right, Mean time constants from single exponential fits to current decays at 0 mV. D, Voltage protocols and representative Na currents from Purkinje cells isolated from a Scn1b heterozygous, Scn1b null, Scn1b heterozygous Scn2b null, and Scn1b Scn2b double-null mouse, as labeled. E, The mean early persistent current from the last 5 ms of a 30 ms step (boxed area in D), normalized to the peak transient current at 0 mV and plotted versus voltage. Top, Loss of β1 (n = 11) compared with littermate controls (n = 11). Bottom, Loss of β1 expression on an Scn2b null background (n = 30) compared with littermate controls (n = 11). WT, Wild type; het, heterozygous. Error bars indicate SEM.
Next, we compared inactivation in the presence and absence of β1. Scn1b null Purkinje cells showed a small but significant negative shift in the V1/2 of the availability curve, from −65.1 ± 1.0 mV (wild type) to −68.9 ± 1.0 mV (Scn1b null; p = 0.02) and no effect on k or the percentage of noninactivating current (wild type: n = 9, k = 6.1 ± 0.2; Scn1b null: n = 7, k = 5.8 ± 0.1; p = 0.3; percentage noninactivating: wild type, 0.71 ± 0.18; Scn1b null, 0.63 ± 0.14; p = 0.7) (Fig. 8B). Larger changes were evident in the comparison of double-null Purkinje cells to the Scn2b null littermate control neurons. The percentage of noninactivating current was increased in the double nulls (from 0.92 ± 0.23 to 1.56 ± 0.14%; p = 0.003). Inactivation was also significantly hyperpolarized in the double-null Purkinje cells (from −64.2 ± 1.0 to −70.5 ± 1.0 mV; p < 0.001; n = 9, 22), without a concomitant increase in k (5.9 ± 0.1, 5.8 ± 0.1) (Fig. 8B). This negative shift in V1/2 is greater than that seen in HEK cells and CA3 cultures transfected with β4. In the absence of the β4-induced increase in the slope factor that was present in those cell types, the negative shift in inactivation is likely to suppress the window current at negative voltages. Not surprisingly, therefore, the persistent current measured 90–100 ms after step depolarizations to voltages below 0 mV was not increased in either the single Scn1b or double nulls (p > 0.1 at all potentials between −40 and −10 mV; data not shown).
Despite the lack of change in persistent current, inspecting the families of traces suggested that inactivation was not identical in cells from the four genotypes. For example, the time constant of decay at 0 mV tended to be longer in the double-null cells (Fig. 8C). Because of this apparent slowing of inactivation, we also measured the “early” persistent current as the percentage of current remaining 25–30 ms after the depolarization. This duration is more than fivefold longer than the dominant time constant of inactivation in all genotypes, but inactivation has not yet reached a steady state by this time (Fig. 8D). Plots of this early persistent current normalized to the peak transient current versus voltage illustrated that the absence of β1 increased the amount of early persistent current, both in Scn1b null (n = 11) versus wild-type control (n = 11) (Fig. 8E, top) cells and in double-null (n = 30) versus Scn2b null (n = 11) (Fig. 8E, bottom) cells. As expected from the larger negative shift of the activation curve and the slower rate of inactivation, this effect was greater in the double-null neurons, in which the persistent current at −30 mV was increased by 60%. In fact, persistent currents >3% were seen in 30% of the double-null neurons but in none of the cells from β2 null littermates (Mann–Whitney U test; Z = −2.12; p = 0.034).
Together, the data indicate that the properties of Na channel complexes differ widely in HEK–NaV1.1 cells, cultured CA3 neurons, and isolated Purkinje neurons, even with predicted similarities in β subunit expression. In Purkinje neurons, loss of β1 and β2 hyperpolarizes the V1/2 of inactivation relative to control more than in HEK–NaV1.1 cells (control vs β4 transfected) or even cultured CA3 neurons (control vs β4 transfected) without an increase in the slope factor, producing a small window current. These differences are likely to result from cell-specific factors, such as α-subunit identity, associated proteins, and posttranslational modifications. Nevertheless, in neurons, as well as in HEK–NaV1.1 cells, if β4 is the only β subunit present, channels activate more readily, inactivation proceeds relatively slowly, and the noninactivating component of the availability curve is increased. Moreover, all these characteristics are apparently counteracted by expression of β1.
Discussion
These experiments demonstrate that incorporation of β4 subunits into NaV1.1 channel complexes in HEK cells increases the amplitudes of TTX-sensitive Na current, by activating channels at more negative voltages, increasing the amount of noninactivating current, and flattening the availability curve. β4 expression thereby favors open over closed states at hyperpolarized potentials and open over inactivated states at depolarized potentials. Coexpression of β1, or chimeric subunits including the extracellular and transmembrane domains of β1, suppresses the β4-induced changes, instead stabilizing closed states at hyperpolarized potentials and inactivated states at depolarized potentials. The idea that β1 acts as a brake on channel opening favored by other β subunits may be relevant to disease. For instance, although coexpression of β4 with the GEFS+ mutant subunit β1C121W generates channels that require even stronger depolarizations for activation than with wild-type β1, the resultant channels fail to inactivate as readily as with wild type β1. Thus, the β4–β1–NaV1.1 interaction described here raises the possibility that an unmasking of slowly inactivating or noninactivating Na current may contribute to disorders of hyperexcitability, such as epilepsy.
Experiments in hippocampal and Purkinje cells support the idea that β4 also favors open states in neurons, although the changes were smaller and more variable than in HEK–NaV1.1 cells. This variation is not surprising given the diversity of α subunits and the variety of Na channel-associated proteins present in neurons but not heterologous expression systems (Goldin, 2001). Nevertheless, overexpression of β4 in hippocampal neurons, like in HEK–NaV1.1 cells, consistently flattened inactivation curves and increased persistent currents. Moreover, in Purkinje neurons, which normally express β1, β2, and β4 subunits, some differences between Scn1b null and Scn1b/Scn2b double-null neurons resembled the differences between α+β1+β4 and α+β4 channels in HEK–NaV1.1 cells, namely the shift in activation and the increase in noninactivating current. Unlike in HEK–NaV1.1 cells, however, loss of β1 and β2 from Purkinje neurons did not change the slope of the availability curve and instead significantly shifted it to more hyperpolarized potentials. This discrepancy reinforces the idea that neurons include factors that heterologous expression systems lack, such as additional associated proteins and posttranslational modifications, which contribute to the overall properties of Na currents in the intact brain (Meadows and Isom, 2005).
Resurgent Na current
Although a peptide from the β4 cytoplasmic tail replicates the open-channel block and unblock responsible for resurgent Na current (Grieco et al., 2005), expression of the intact β4 subunit was insufficient to block channels either in HEK–NaV1.1 cells or in cultured neurons. Three explanations may account for this result: (1) Na channel complexes require posttranslational (or other) modification to permit open-channel block; (2) expression of an additional protein(s) may be necessary to allow block by β4; or (3) the endogenous blocking protein in neurons with resurgent kinetics is not the β4 subunit that we expressed but is instead a related protein with a cytoplasmic region that is structurally indistinguishable from the β4 peptide. Regarding the first possibility, in neurons, α subunits are modulated by kinases and phosphatases (Cantrell and Catterall, 2001; Ahn et al., 2007), and even β4 is the target of proteases (Miyazaki et al., 2007). Also noteworthy is that a broad-spectrum phosphatase abolishes the functionality of the blocking protein (Grieco et al., 2002). Regarding the latter two possibilities, because multiple proteins form the Na channel complex (Abriel and Kass, 2005), it is likely to be difficult in a heterologous expression system to replicate the native condition in neurons with resurgent current. Thus, whether the endogenous blocking protein is only structurally related to β4, or is a form of β4 itself, remains an open question.
Persistent Na current
At least two factors contribute to the ability of the β4 subunit to raise the amplitude of persistent current in HEK–NaV1.1 cells: the increase in window current by the modification of the activation and inactivation curves and the increase in open-state occupancy after maximal inactivation. Increases in persistent current have also been observed in NaV1.2-expressing tsA-201 cells when β3 is coexpressed; in those cells, β2 alone does not modulate persistent current but augments the effect of β3 (Qu et al., 2001). Thus, at least in certain cellular environments and with certain α subunits, β2, β3, and β4 all promote persistent currents. In contrast, the subunit that consistently accelerates inactivation and/or decreases persistent current across Na channel α subunits is β1 (Isom et al., 1992; Chen and Cannon, 1995; Smith et al., 1998; Meadows et al., 2002; Valdivia et al., 2002).
Na channels, β subunits, and epilepsy
In several Na channel α subunits, the efficacy of the fast inactivation gate (i.e., the cytoplasmic linker between domains III and IV) (Vassilev et al., 1988, 1989) is influenced by the C terminus. Mutations in the C terminus increase persistent current (Baroudi and Chahine, 2000; Choi et al., 2004) and/or slow the rate of fast inactivation (Wu et al., 2005), probably by disrupting an electrostatic interaction with the III–IV linker (Cormier et al., 2002), and many proteins that modulate Na channels depend on interactions with the C terminus (Abriel and Kass, 2005). Furthermore, wild-type β1 directly interacts with the C terminus of the NaV1.1 channel, and mutations that weaken this interaction generate GEFS+ (Annesi et al., 2003; Spampanato et al., 2004). Together, these results suggest that the C terminus interacts with the III–IV linker in a β1-dependent way to promote inactivation.
In the present study, however, the β1/4 chimera facilitated inactivation and suppressed the effects of β4 in much the same way as the intact β1 subunit, suggesting that the intracellular domain of β1 need not interact directly with either the Na channel C terminus or the III–IV itself. Instead, extracellular (or transmembrane) domains of β1 may act allosterically to stabilize nonconducting states of the channel. This idea is consistent with studies indicating that the extracellular domain of β1 is sufficient to accelerate inactivation of NaV1.2 and NaV1.4 (Chen and Cannon, 1995; McCormick et al., 1998, 1999).
The consequences of disrupting the interaction between β1 and the rest of the channel complex are of clinical interest because several mutations in β1 correlate with GEFS+ (Wallace et al., 1998, 2002; Audenaert et al., 2003; Burgess, 2005; Scheffer et al., 2007). These include β1C121W, which has been electrophysiologically characterized as a loss-of-function or reduction-of-function of β1 (Wallace et al., 1998; Meadows et al., 2002; Tammaro et al., 2002). Our data demonstrate that Na currents resulting from coexpression of β1C121W with β4 in HEK–NaV1.1 cells are distinct from those arising from expression of β4 or β1+β4 alone, suggesting that heterotrimers with emergent properties can indeed assemble. The activation curve suggests that such channels may require larger than normal stimuli to open, but once activated, they inactivate considerably less readily than channels with wild-type β1, a feature that might promote repetitive firing. In fact, other mutations that lead to GEFS+, which are located in voltage-sensing regions of NaV1.1 (Escayg et al., 2000), also destabilize inactivation and increase persistent Na current (Lossin et al., 2002; Kahlig et al., 2006), raising the possibility that distinct mutations converge on a common mechanism to produce a specific disease phenotype.
Another link between reduction of β1 function and epilepsy comes from the observation that Scn1b null mice display spontaneous generalized seizures, although Scn1b null hippocampal neurons show no significant changes in Na current relative to wild type (Chen et al., 2004). Our results, however, raise the possibility that Na channel gating in these mutants might be modified more extensively in brain regions that normally express β4 as well as β1, because changes in Na currents after loss of β1 (and β2) were readily detectable in Purkinje neurons, which express higher levels of β4 than hippocampal neurons (Yu et al., 2003). Nevertheless, because seizures are unlikely to originate in the cerebellum, it seems reasonable to speculate that channel opening may also be upregulated in at least one other brain region in Scn1b null mice. Consistent with this idea, cardiac myocytes (which express NaV1.5 and β4) have more persistent current in Scn1b null than wild-type cells, a condition that correlates with the long-QT syndrome (Lopez-Santiago et al., 2007).
Physiological role of β4
The interaction among β subunits suggests that the expression of β4 is not a direct indicator of whether Na channel opening will be facilitated in any given neuron. Our results do, however, predict that an increased relative expression of β1 may correlate with the stabilization of nonconducting states, whereas an increased relative expression of β4 may correlate with more channel openings. In fact, several brain regions that strongly express β4 (Yu et al., 2003; Oyama et al., 2006) have relatively large persistent currents, including the cerebellum (Llinás and Sugimori, 1980; Raman and Bean, 1997), thalamus (Parri and Crunelli, 1998), cortical pyramidal neurons (Cummins et al., 1994; Maurice et al., 2001), and striatum (Cepeda et al., 1995; Chao and Alzheimer, 1995). Along with our data, these observations support the idea that the complement of β subunits defines the range of persistent current amplitudes that can be produced by the Na channel complex.
Footnotes
-
This work was supported by National Institutes of Health Grants NS39395 (I.M.R.) and MH059980 (L.L.I.), by National Research Service Award F31 NS057866 (T.K.A.), and by a grant from the Partnership for Pediatric Epilepsy (L.L.I.). We thank Dr. Ken Tovar for teaching us culturing methods. We are also grateful to Jason Pugh, Nan Zheng, Jason Bant, and Mark Benton for helpful discussion.
- Correspondence should be addressed to Indira M. Raman, Department of Neurobiology and Physiology, 2205 Tech Drive, Northwestern University, Evanston, IL 60208. i-raman{at}northwestern.edu