Journal of Neurophysiology


Ankyrin-G, a modular protein, plays a critical role in clustering voltage-gated sodium channels (Nav channels) in nodes of Ranvier and initial segments of mammalian neurons. However, direct effects of ankyrin-G on electrophysiological properties of Nav channels remain elusive. In this study, we explored whether ankyrin-G has a role in modifying gating properties of the neuronal Nav1.6 channel that is predominantly localized at nodes of Ranvier and initial segments. TsA201 cells transfected with the human Nav1.6 cDNA alone exhibited significant persistent sodium current (Ina-p). On the other hand, Ina-p was barely detected on co-expression with ankyrin-G. Ankyrin-B, another ankyrin, did not show such an effect. Expression of chimeras between the two isoforms of ankyrin suggests that the membrane-binding domain of ankyrin-G is critical for reducing the Ina-p of Nav1.6. These results suggest that ankyrin-G regulates neuronal excitability not only through clustering Nav channels but also by directly modifying their channel gating.


Ankyrins are the multidomain adaptor proteins that link the spectrin/actin network to the cytoplasmic domain of membrane proteins, including ion channels and cell adhesion molecules (Bennett 1992) (Bennett and Gilligan 1993). Ankyrins consist of three major domains: a membrane-binding domain with a similar folding structure and the 24 consecutive repeats (the MB domain), a spectrin-binding domain (the S domain), and a death domain with the C-terminus (the DC domain) (Bennett and Baines 2001). Among three ankyrin isoforms, ankyrin-G is expressed abundantly at the axon initial segment and nodes in myelinated axons and directly interacts with axonal Nav channels (Bennett and Baines 2001; Kordeli et al. 1995). Ankyrin-G is known to play an essential role in rapid saltatory conduction of myelinated axons: in knockout mice of ankyrin-G, Nav channels are absent from axon initial segments and the nodes of Ranvier, and neurons of these mice show severe deficits in action potential firing (Jenkins and Bennett 2001; Zhou et al. 1998).

Recently, in the human Nav1.5 sodium channel, a mutation in the consensus sequence for binding to ankyrin-G has been reported to be associated with Brugada syndrome, a dominantly inherited cardiac arrhythmia (Mohler et al. 2004b). This mutation drastically reduces surface expression of Nav1.5 channel in cardiomyocytes. In addition, heterologous expression of this mutant hNav1.5 channel showed distinct gating behavior compared with the wild-type channel. This suggests that ankyrin-G may have a modulatory effect on gating properties of sodium channel besides clustering or recruitment of Nav channels to the cell membrane.

In the present study, we aimed to explore the regulatory role of ankyrin-G in gating properties of a neuronal sodium channel, Nav1.6, which predominantly expresses at the nodes of Ranvier and the initial segment (Caldwell et al. 2000). cDNA encoding hNav1.6 was transfected into tsA201 cells, which do not express endogenous ankyrin. We show that ankyrin-G negatively regulates persistent sodium current (Ina-p) through the hNav1.6 channel. This raises the possibility that ankyrin-G regulates neuronal excitability not only through clustering Nav channels, but also through modifying gating behaviors of Nav1.6 channels.


cDNA preparation and transfection

Human Nav1.6 cDNA was isolated from a lambda gt11 human spinal cord cDNA library, which contained a 5,940-bp open reading frame (NCBI AB027567). The T4910C mutation was corrected and the cDNA fragment containing the adult type of exon 5 was used to construct a full length cDNA in pUC19. For expression in tsA201 cells, a CMV promoter from the pcDNA3 vector (Invitrogen) was subcloned upstream of the coding region of Nav1.6 in hNav1.6-pUC19 (CMV-hNav1.6-pUC19). Ankyrin-G (270 kDa) and 220-kDa ankyrin-B were expressed as fusion proteins with enhanced green fluorescent protein (GFP) at the N terminus. Ankyrin-B/G chimeras contain GFP at the C terminus (Mohler et al. 2002).

TsA201 cells were transiently transfected with SuperFect or PolyFect (QIAGEN) according to the manufacturer's protocols and cultured for 30–72 h. To express Nav channels consisting only of alpha-subunit protein, cells were transfected with CMV-hNav1.6-pUC19 and pEGFP-C1 (BD Biosciences). For coexpression with rat sodium channel beta1 subunit (Navβ1), Nav1.6 cDNA was cotransfected with the plasmid of pCD8-IRES-β1, containing a pCD8 coding region, in tandem with the Navβ1-coding region, and transfected cells were identified by labeling with anti-CD8 coated beads (DYNAL M-450 CD8).


Electrophysiological recordings of tsA201 cells were performed at room temperature with whole cell patch-clamp techniques using an EPC-9 amplifier (HEKA Electronik, Germany). The extracellular solution consisted of (in mM) 150 NaCl, 2 KCl, 1.5 CaCl2, 1 MgCl2, 10 glucose, and 10 4-(2-hydroxethel)-1-piperazine-ethanesulphanic acid (HEPES), pH 7.4. The internal pipette solution consisted of (in mM) 117 CsCl, 9 ethylene glycol-bis (β-amino-ethyl ether)N,N,N′,N′-tetraacetic acid (EGTA), 9 HEPES, 1.8 MgCl2, 14 Tris-creatine PO4, 4 Mg-ATP, and 0.3 Tris-GTP, pH 7.4. Pipettes of resistance 1.8–3.5 MΩ were used. We noted that Ina-p was highly sensitive to the composition of the internal solution; little Ina-p was observed when CsF instead of CsCl was used, as previously reported (Burbidge et al. 2002). Junction potentials (–2 mV on average) were not compensated. In cells with current amplitude >1 nA, series resistance was compensated ≤80%. No untransfected cell exhibited the peak amplitude of inward sodium currents over 100 pA. TTX subtraction experiments were performed with a fast perfusion system using the DAD12-system (ALA scientific). The current-voltage relationship was analyzed via a step protocol in which cells were depolarized from a holding potential of −80 mV. Each interval was 5 s and leakage currents were subtracted using the P/N method with N values ranging between 4 and 10. Ina-p was measured as steady-state inward currents at 100 ms for most data, and at 20 ms in data used in Fig. 2, after the beginning of depolarization. The voltage dependence of steady-state inactivation was determined via a two-step protocol in which a 300-ms conditioning pulse was applied from a holding potential of –80 mV followed immediately by a test pulse to 10 mV. Analysis was performed using PulseFit software (HEKA) and IgorPro (Wavemetrics). Results are presented as mean ± SE, and statistical significance was determined using an unpaired t-test.

Measurements of Ina-p

%pc was obtained by calculating the ratio of Ina-p against the maximum peak amplitude of transient inward currents in the series of step pulses. Current amplitudes were obtained by leak subtraction using the P over N protocol except in Fig. 2. As seen in figures (e.g., Figs. 1 and 3), inward currents reach steady-state within 20 ms in most ranges of membrane potential. To minimize the error in estimating %pc, we utilized low sodium internal solution. The rationale described in results is based on the fact that Ina-p attains its maximum value around at +10 mV, more positive than the voltage that gives the peak amplitude of the fast inward component. To maximize the inward Ina-p, a solution with low sodium concentration was used in patch pipette throughout this study.

FIG. 1.

Whole cell Nav currents recorded from tsA201 cells transfected either with hNav1.6 or the rNav1.2 α-subunit. A and B: representative traces were elicited by membrane depolarizations for 100 ms ranging from –30 to +50 mV in 10-mV increments from a holding potential of −80 mV. The same trace in A is expanded in B at a different time scale. The peak sodium curents of hNav1.6 and rNav1.2 were 51.2 ± 1.1 pA/pF (n = 30) and 90.0 ± 8.9 pA/pF (n = 4), respectively. C: current-voltage relationship (I-V curves) of hNav1.6 (○) and rNav1.2 (*). Values were normalized by the maximum amplitude. D: voltage dependence of steady-state inactivation and representative traces (insets) for hNav1.6 and rNav1.2. Representative traces with a preconditioning pulse to –110 mV ( · · · ) and –20 mV(—) indicate that the rNav1.2 channel was completely inactivated by a conditioning pulse to –20 mV.


Total RNA of tsA201 was isolated using Trizol reagent (Invitrogen) following the manufacturer's protocol. cDNA was synthesized using Superscript (Invitrogen) primed with oligo-dT. For DNA amplification of human ankyrin-G, -B, -R, PCR primers were designed in two distinct regions: the ankyrin-repeat region and the spectrin binding domain. Sequences were selected such that they could discriminate among three isoform genes. Both sets of primers gave similar results. The following oligonucleotides correspond to the spectrin-binding domain: ank-G-forward (CACAGACAAATATCTTGGG) and ank-G-reverse (TCTAAGAGAATCTGAATCAA) for ankyrin-G; ank-B-forward (GCCTGAGGACCTAAAAGAAC) and ank-B-reverse (CCCCAGCTTAGGCGATAAGC) for ankyrin-B; ank-R-forward (GATGAAGGGGAAGAACTCATC) and ank-R-reverse (TGTCTGAGGTCTCGGTGGCC) for ankyrin-R. Primers corresponding to the ankyrin-repeat domain were: ankG-Repeat-F (GATCATTTAAACTGCGTCCAG) and ankG-Repeat-R (CCATGATGCATTAGTTGTGA) for ankyrin-G; ankB-Repeat-F (GACCACGTGGAATGTGTGAA) and ankB-Repeat-R (GGCTCCGTTCTGCAGCAGAAG) for ankyrin-B; ankR-Repeat-F (GACCACCTCGACTGTGTCCGG) and ankR-Repeat-R (CCCCGCTGCAGGAGGTTCTT) for ankyrin-R.

PCR reaction was performed under the condition: 28 cycles of 94°C for 50 s, 54° for 60°C, 72°C for 60 s. As a positive control, cDNA was synthesized from human adult brain RNA (purchased from BD Bioscience) and amplified with the same sets of primers as in the cDNA from tsA201 cells.


Human Nav1.6 sodium channels expressed in tsA201 cells show robust Ina-p

hNav1.6 plasmid was transfected into tsA201 cells for whole cell patch recording. The full-length hNav1.6 protein by this plasmid was confirmed by the expression of a protein ∼250 kDa in molecular size after transfection into COS1 cells as detected by Western blot (supplementary Fig. 1) with the monoclonal anti-pan sodium channel antibody (courtesy of Dr. James Trimmer). Inward currents recorded from tsA201 cells transfected with hNav1.6 consisted of a rapid, transient component (peak current) and a sustained component that persisted even after pulses of 100 ms (persistent current; Ina-p; Fig. 1) consistent with a previous report (Burbidge et al. 2002). A marked amount of Ina-p was observed in most cells (10.5 ± 1.8% at +10 mV; n = 7; Fig. 1B). The current-voltage relationship (I-V curve; Fig. 1C) for both peak current and the persistent component showed that Ina-p reached the maximum amplitude at a more positive voltage than the transient current; the Ina-p attained its peak at +10 mV, whereas the transient current did so at 0 mV. Such a difference in the peak voltage between the fast transient current and the Ina-p was also noted in Xenopus oocytes (data not shown), consistent with a previous study (Smith et al. 1998).

The steady-state inactivation curve of Nav1.6 currents showed a shoulder around at –50 to –30 mV, and two components of the Boltzmann equation were needed to fit the data (Fig. 1D; the 2 half inactivation voltages, V1/2a, V1/2b, for Nav1.6 are –61.9 ± 1.1 mV and –22.5 ± 2.1 mV with slopes of 9.6 ± 2.2 and 7.8 ± 4.3 (n = 6), respectively). The transient peak current decreased as conditioning pulses became more positive. In contrast, the slow component was more resistant to conditioning pulse up to +10 mV (see nonzero value over +10 mV in Fig. 1D and inset trace). To clarify whether robust Ina-p is a feature unique to Nav1.6, another neuronal sodium channel, rat Nav1.2 (rNav1.2), was also recorded. The peak sodium current of Nav1.2 was 90.0 ± 8.9 pA/pF (n = 4). Ina-p was significantly smaller in Nav1.2-transfected cells than that in Nav1.6-transfected cells [2.5 ± 0.5% (n = 3) at 0 mV for Nav1.2]. In contrast with the biphasic profile in Nav1.6-transfected cells, Nav1.2-transfected cells exhibited a simple profile of the steady-state inactivation curve that could be fitted by a single Boltzmann equation (Fig. 1D; V1/2 –53.8 mV and a slope of 6.0). We also recorded hNav1.1 and hNav1.5 under the same conditions and observed much less Ina-p compared with the hNav1.6 channel (data not shown).

Quantification of Ina-p could be affected by the presence of some endogeneous currents, in particular at high depolarizing levels. To verify the method of estimation of Ina-p with the P/N protocol, the Nav channel currents were compared between the TTX-subtraction method (Fig. 2A) and the P/N protocol in the same cells. Figure 2B indicates a typical example of superimposed traces between the TTX-subtraction method (blue) and P/N protocol (red) from the same cell. The persistent component does not significantly differ between the two protocols over a wide range of membrane potential <20 mV. This was also supported by the plots from accumulated data from nine cells (Fig. 2C). However, the transient inward current with TTX subtraction slightly deviates from that with the P/N protocol at higher depolarizations (Fig. 2B, +20 mV, +30 mV). This is probably because the kinetics of the inward current are faster at higher depolarizations and more readily subject to errors due to fluctuating capacitative currents. Therefore in the following experiments with P/N protocol, we quantified the proportion of Ina-p over a range of membrane potentials, including 0 to +10 mV. The peak amplitude of transient current with P/N protocol tended to be smaller than that with TTX subtraction at higher voltages probably because more uncancelled capacitative currents overlapped with the transient sodium currents that become sharper as the membrane potential is more depolarized. The proportion of Ina-p at each membrane potential was therefore measured by standardizing Ina-p with the maximum peak of the transient inward current in the I-V curve (termed %pc) in later experiments. We also checked the quality of voltage-clamp condition by comparing the I-V curve between the two external solutions, containing 150 and 50 mM sodium ions. Figure 2D shows a typical example of such data. The I-V curve does not significantly change between the two conditions. Similar results were obtained from other four cells. Therefore it is unlikely in our measurements that imperfect voltage-clamp causes error of estimation of %pc.

FIG. 2.

Comparison of inward currents of hNav1.6 between a TTX-sensitive inward current vs. a current measured with P/N protocol. A: example of raw traces before (black) and after (red) application of TTX. Capacitative current and leakage are not corrected. B: representative traces from 1 cell at different membrane potentials. Blue indicates the trace obtained with TTX-subtraction and red indicates that with the P/10 protocol. C: x-y plot of current amplitudes of Ina-p obtained with P/10 and TTX-subtraction from eight cells expressing hNav1.6 cDNA. Data of 0, +10, and +20 mV were from 9 cells. Data of 30 mV was from 8 cells. D: I-V curve of transient inward currents compared under 150 mM sodium and 50 mM sodium. In the lower plot, current amplitude was normalized by the peak value.

Several mammalian Nav channels are known to exhibit the slowly inactivating component in Xenopus oocyte which is normalized by the coexpression of sodium channel β1-subunit (Wallner et al. 1993). To test if inactivation gating of hNav1.6-derived currents depend on the auxiliary subunit, β1-subunit was coexpressed with hNav1.6. Coexpression with β1-subunit did not significantly increase the Nav current densities [51.2 ± 1.1 pA/pF (n = 30) and 87.9 ± 12.2 pA/pF (n = 6) for cells expressing Nav1.6 alone and Nav1.6 plus β1, respectively; Fig. 3]. No significant difference in the %pc was found between transfection of Nav1.6 with the β1-subunit and that of Nav1.6 alone [Fig. 3D; 10.8 ± 0.7% at +10 mV (n = 7) for Nav1.6 alone; 12.8 ± 0.5% (n = 4) for Nav1.6+β1]. The steady-state inactivation curve was shifted to a slightly more negative potential (Fig. 3E). The activation kinetics were compared by measuring the rising time to half of the peak amplitude (t1/2; Fig. 4). The t1/2 (647 ± 37 μs, n = 5) was significantly faster than that of cells without the β1-subunit (912 ± 54 μs, n = 13). A shift in steady-state inactivation and faster rising time verify that the β1-subunit was functionally expressed.

FIG. 3.

Whole cell sodium currents recorded from tsA201 cells expressing the hNav1.6 α-subunit alone or with ankyrin-G (AnkG) or/and β1-subunit. A and B: maximum amplitudes of sodium currents from cells expressing hNav1.6 alone, hNav1.6+AnkG, hNav1.6+β1, and hNav1.6+AnkG+β1 were 51.2 ± 1.1 pA/pF (n = 30), 45.3 ± 0.85 pA/pF (n = 28), 87.9 ± 12.2 pA/pF (n = 6), and 78.3 ± 14.9 pA/pF (n = 4), respectively. The same records in A are shown in a different time scale in B. C: I-V curves obtained from cells transfected with Nav1.6 alone (○), hNav1.6+AnkG (□), hNav1.6+β1 (•), or hNav1.6+β1+AnkG (▪). D: voltage dependence of the proportion of Ina-p (%pc). Without AnkG, %pc was larger (*; P < 0.05, **; P < 0.01). E: steady-state inactivation curve. Current traces during the test stimuli are shown (top). The curve of cells expressing AnkG was fitted with a single Boltzmann equation.

FIG. 4.

Rising time to half the peak amplitude (t1/2) was determined for cells expressing hNav1.6 alone and cells coexpressing Nav1.6 with ankyrin-G and/or the sodium channel β1-subunit. The t1/2 of hNav1.6 alone (○), hNav1.6 plus ankyrin-G (□), hNav1.6 plus β1 (•), and hNav1.6 plus ankyrin-G plus β1 (▪) were 912 ± 54 μs (n = 13), 789 ± 50 μs (n = 13), 647 ± 37 μs (n = 5), and 611 ± 52 μs (n = 4), respectively. *, significant difference.

Coexpression of Ankyrin-G decreases the Ina-p of hNav1.6

Before testing the role of ankyrin-G in regulating Nav channel-gating properties, we first examined whether tsA201 cells express endogenous ankyrin-G or ankyrin-B. RT-PCR was performed with primers each specific to human ankyrin-G and -B (Fig. 5). Both primers gave positive bands from brain cDNA, whereas no signal was obtained for either isoform of ankyrin with cDNA from tsA201 cells. Actin primers as positive control gave signals under the same conditions. The third ankyrin isoform, ankyrin-R, was also tested and found not expressed (data not shown). We therefore concluded that tsA201 cells will provide an appropriate model to test the role of heterologously expressed ankyrin on Nav channels.

FIG. 5.

RT-PCR shows no endogenous expression of ankyrin-G or -B gene in tsA201 cells. Right 2 lanes: no signal from tsA201 cDNA; 3rd and 4th lanes: clear positive signals from brain cDNA. Integritiy of cDNA was checked by the expression of actin (right lane).

Effects of ankyrin-G on hNav1.6 were tested both in the absence or presence of β1-subunit. Coexpression with ankyrin-G did not increase peak inward currents in a statistically significant manner [51.2 ± 1.1 pA/pF (n = 30) for Nav1.6 alone and 45.3 ± 0.85 pA/pF (n = 28) for Nav1.6 plus ankyrin-G]. Peak sodium currents of cells expressing Nav1.6 plus ankyrin-G and β1 [78.3 ± 14.9 pA/pF (n = 4)] were slightly larger than those of cells expressing Nav1.6 alone or Nav1.6 plus ankyrin-G. Importantly, the coexpression with ankyrin-G significantly reduced Ina-p [3.6 ± 0.3% at –10 mV (n = 6) for Nav1.6+ankyrin-G; 3.8 ± 1.2% at 0 mV (n = 3) for Nav1.6+ankyrin-G+β1 vs. 10.8 ± 0.7% at +10 mV (n = 7) for Nav1.6 alone; Fig. 3, B, C, and D]. Such a reduction in Ina-p by ankyrin-G was also reflected in the steady-state inactivation curve, which did not show the shoulder at higher voltage (depolarization ranging from −50 to +10 mV; Fig. 3E). In addition, cells expressing ankyrin-G showed a small noninactivating component at +10 mV and the inactivation curves were fitted by a single Boltzmann equation [V1/[r]2 and slopes were –50.4 ± 2.8 mV and 15.7 ± 3.1, respectively (n = 6), for Nav1.6 plus ankyrin-G and –54.6 ± 1.7 mV and 9.6 ± 0.8, respectively (n = 3), for Nav1.6 plus ankyrin-G and β1]. I-V curves were shifted by ∼8 mV leftward in cells transfected with Nav1.6 and ankyrin-G (Fig. 3C). However, ankyrin-G did not show a significant effect on activation kinetics; the t1/2 of Nav1.6 alone, Nav1.6 plus ankyrin-G, and Nav1.6 plus ankyrin-G plus β1 were 912 ± 54 μs (n = 13), 789 ± 50 μs (n = 13), and 611 ± 52 μs (n = 4), respectively (Fig. 4).

Membrane-binding domain of ankyrin-G is critical for reduction of hNav1.6-derived Ina-p

We tested whether ankyrin-B, another isoform of ankyrin abundantly expressed in mammalian neurons (Kunimoto et al. 1991), also shows the modulation activity of Nav1.6 inactivation. Maximum peak amplitudes of hNav1.6 coexpressed with ankyrin-B were 51.3 ± 5.7 pA/pF (n = 9), which did not show significant difference against that of hNav1.6 alone or hNav1.6 with ankyrin-G. The Ina-p of hNav1.6 did not decrease on coexpression with ankyrin-B (Fig. 6B). The %pc of the cell coexpresssing hNav1.6 with ankyrin-B was as large as that of hNav1.6 alone [12.1 ± 1.1% at 0 mV (n = 4) for hNav1.6+ankyrin-B; 10.8 ± 0.7% at +10 mV (n = 7) for hNav1.6 alone; Fig. 6D]. The steady-state inactivation curve showed a shoulder around –50 to –30 mV, and two components of the Boltzmann equation were needed to fit the data [Fig. 6E; V1/2a, V1/2b –65.6 ± 1.9 mV and –17.1 ± 3.9 mV with slopes of 5.1 ± 1.2 and 7.7 ± 1.6, respectively (n = 5)] as in cells only transfected with hNav1.6 cDNA. I-V curves observed from cells expressing ankyrin-B were similar to those of cells with Nav1.6 alone (Fig. 6C).

FIG. 6.

Distinct effect between ankyrin-G (AnkG) and ankyrin-B (AnkB) on the Ina-p of the hNav1.6 channel. A and B: representative traces are shown with distinct time scale. Maximum current densities of Nav channel currents from cells expressing hNav1.6 alone, hNav1.6+AnkG, and hNav1.6+AnkB were 51.2 ± 1.1 pA/pF (n = 30), 45.3 ± 0.85 pA/pF (n = 28), and 51.3 ± 5.7 pA/pF (n = 9), respectively. C: I-V curves from cells expressing hNav1.6 alone (○), hNav1.6 and AnkB (⧫), and hNav1.6 and AnkG (□). D: %pc at different membrane potentials (*; P < 0.05, **; P < 0.01. E: steady-state inactivation curve. Current traces during the test stimuli are shown (top). Fitting the curve from cells coexpressing with AnkB (▴; n = 5) required 2 components of the Boltzmann equation.

To define a region within ankyrin-G that confers modifying ability for inactivation of hNav1.6, we expressed hNav1.6 with five distinct chimeras between ankyrin-G and B (Mohler et al. 2002) (Fig. 7A). Expression of ankyrin chimera was confirmed by the fluorescent signal of GFP, which was fused at the N-terminus of each chimeric protein. In the cells expressing the protein that contained the MB domain of ankyrin-B, %pc was as the same that of cells expressing hNav1.6 alone (Fig. 6, B and C, Table 1). In contrast, all chimeras that contained the ankyrin-G MB domain showed small %pc (Fig. 7D, Table 1). These findings suggest that the MB domain of the ankyrin-G is critical for the reduction of Ina-p of hNav1.6. This also supports the idea that reduction of Ina-p by ankyrin-G is mediated by direct binding to hNav1.6.

FIG. 7.

Effect of coexpression of ankyrin-B/G chimeras on the Ina-p of Nav1.6. A: scheme of ankyrin-B/G chimeras. B and C: representative traces recorded from tsA201 cells expressing ankyrin-B/G chimeras. Maximum current densities of Nav channel currents were not significantly different among chimeras. D: %pc at distinct membrane potentials.

View this table:

Comparison of proportion of Nav 1.6-derived lna-P from cells transfected with ankyrin-G-B chimeras against cells transfected with ankyrin-B (BBB)

To confirm that physical association of ankyrin-G with hNav1.6 underlies the reduction of Ina-p, we tried experiments of immunocoprecipitation using antibodies against ankyrin-G and hNav1.6. However, we failed to detect this interaction, although pan-sodium channel antibody detected a faint but clear band in the immunoblot of heterologously expressed hNav1.6 (supplementary Fig. 1). Our failure to detect direct binding of hNav1.6 with ankyrin-G in immunocoprecipitation experiment was probably due to the low level of expression of hNav1.6 in our system. The cytoplasmic linker of the sodium channel alpha subunit was known to associate with the membrane domain of ankyrin-G. A target motif in cytoplasmic II-III linker has been identified as a critical region that is sufficient to localize membrane proteins at the axon initial segment by a process involving association with ankyrin-G (Garrido et al. 2003; Lemaillet et al. 2003). To test if this region is required for correction of inactivation gating by ankyrin-G, Nav1.6 lacking the target motif (VPIAVGESD, V1092_D1100, Nav1.6ΔII-III) was constructed and transfected into tsA201 cells. However, Ina-p of this construct was significantly smaller than that of wild-type Nav1.6 [6.6 + 4.6% at 0 mV (n = 6) for Nav1.6ΔII–III; 10.5 + 4.8% at 10 mV (n = 7) for Nav1.6 alone]. This could probably be due to the direct affect of deletion of the 9-amino-acid motif on gating properties. Therefore we were not able to test the direct binding of hNav1.6 to ankyrin-G using this construct.

As alternative approach, GFP-fused ankyrin-G was visualized under confocal microscope following cotransfection with hNav1.6 cDNA. GFP signals were located mainly on the periphery of cells or borders between cells, probably reflecting that ankyrin-G is recruited to cell membrane with hNav1.6 (Fig. 8). This pattern depended on the presence of hNav1.6 because GFP signal was equally distributed in the cytoplasm when GFP-fused ankyrin-G alone was expressed or GFP-fused ankyrin-B was coexpressed with hNav1.6.

FIG. 8.

Localization of ankyrin protein analyzed by confocal microscopy. Cells were transfected with hNav1.6 and GFP-fused ankyrin-G. Note that signal is most abundant at the pheriphery of cells (middle), whereas signal was more diffuse when ankyrin-G alone or ankyrin-B was expressed. Supplementary Fig. 1. (in Supplementary data pdf file) Western blot of hNav1.6 heterologously expressed in COS1 cells using the monoclonal antibody, K58/35.4, against the pan-sodium channel antigens.


We examined the potential roles of ankyrin-G, an adaptor protein enriched at the node of Ranvier, on gating properties of hNav1.6 channels in tsA201 cells. We found that ankyrin-G significantly reduces Ina-p of hNav1.6 channels.

Estimation of Ina-p of the Nav1.6 channel

In some cases with large current densities of Nav currents, estimation of %pc may not be accurate due to imperfectly controlled voltage-clamp condition. To verify voltage-clamp condition, we compared the I-V curve under the two conditions, 50 mM sodium solution and 150 mM sodium solution (Fig. 2D). Results showed that the I-V curve does not significantly shift between the two conditions in all cells showing <1 nA of peak inward currents, suggesting that peak currents are measured under well-controlled voltage-clamp conditions. However, in cells showing >1 nA (corresponding to the current density, ∼200 pA/pF), the I-V curve was slightly shifted leftward. In this study, some cells exhibited the current density >200 pA/pF. This potentially causes some error in estimation of %pc. Due to such potential inaccuracy of estimation of %pc, we should be cautious about the absolute value of %pc in our study. However, it is unlikely that this affects significance of the difference of %pc obtained between the two groups which is three times larger in our study because only a small number of cells showed current amplitudes >1 nA.

Cells transfected with hNav1.6 cDNA showed robust Ina-p, consistent with previous expression studies in Xenopus oocytes (Smith et al. 1998) and HEK293 cells (Burbidge et al. 2002). On the other hand, in recent experiments of forced expression of Nav1.6 into cultured DRG neurons, which were engineered to be TTX resistant by single point mutation, Nav1.6-derived currents did not exhibit robust Ina-p (Herzog et al. 2003). This discrepancy could be in part due to different conditions of electrical recording or cell preparation. In particular, Ina-p could be highly sensitive to the composition of the internal solution. In the previous work (Herzog et al. 2003), CsF was utilized in an internal solution. We observed that Ina-p was significantly smaller in hNav1.6-transfected tsA201 cells with CsF in the internal solution instead of CsCl as also reported previously (Burbidge et al. 2002).

On coexpression of ankyrin-G, the steady-state inactivation curve was shifted leftward in accordance with the reduction in Ina-p. This shift did not occur equally over the entire range of membrane potential but was more remarkable at a less negative region. This is consistent with the idea that the persistent component of Nav1.6 currents in the present study may reflect a channel population that fails to inactivate. However, this does not neglect the possibility that some component of Ina-p by Nav1.6 is due to the “window current,” which has been suggested as one of the underlying mechanisms of Ina-p in CNS neurons (Taddese and Bean 2002), or due to a modal change in the transition into a bursting mode of channel function (Patton et al. 1994).

We should note that some characteristics of Ina-p observed in tsA201 cells are slightly different from those reported in native neurons. In the experiments on heterologous expression, the peak voltage of Ina-p was shifted to a more positive region of the membrane potential, compared with that of the transient current (Smith et al. 1998) (Fig. 1). In some native neurons (Kay et al. 1998; Magistretti and Alonso 1999), however, Ina-p is largest at a more negative membrane potential than the voltage that gives the maximum transient current. In native neurons, Nav channels usually consist of multiple molecular species of the alpha subunits (Chahine et al. 2005). It is possible that the underlying mechanisms for Ina-p could be diverse among Nav channel isoforms, and the behaviors of Ina-p of the Nav1.6 channel in heterologous expression may reflect a restricted population of Nav channels underlying Ina-p observed in native neurons.

Modulation mechanisms of Nav1.6-derived Ina-p by ankyrin-G

In this study, we failed to provide direct evidence that hNav1.6 directly binds to ankyrin-G in our experimental system despite several trials of immuno-coprecipitation experiments. Ankyrin is known to interact with spectrins that associate with the actin cytoskeleton (Bennett and Baines 2001). In several reports (Shcherbatko et al. 1999), the inactivation kinetics of sodium channels significantly changes when membrane associated cytoskeleton is disrupted. This suggests that inactivation gating of sodium channels depends on membrane-associated cytoskeleton. In our recording under the perforated-patch configuration without applying negative pressure (data not shown), we observed robust Ina-p in cells only expressing Nav1.6 and reduced Ina-p in cells coexpressing ankyrin-G. It is unlikely that ankyrin-G modulates sodium channel gating through indirect mechanisms such as global perturbation of the cytoskeleton or change of cellular morphology. In fact, ankyrin-B, which has actin-recruiting activity, did not exhibit Ina-p reduction. Experiments using chimeras between ankyrin-G and -B suggest that the MB-domain confers isoform-specific inhibition of Ina-p, supporting the idea that direct binding of ankyrin proteins to Nav1.6 mediates modulation of Ina-p. Ankyrin specificities have been described for interactions between IP3 receptors and ankyrin-B in cardiac muscle cells (Mohler et al. 2004a). Ankryin G did not rescue mislocalization of IP3 receptors in cardiac muscle cells in ankyrin-B knockout mice (Mohler et al. 2002). From transfection experiments of various chimeras between ankyrin-G and ankyrin-B, the specificity of the interaction with IP3 receptors between ankyrin-B and ankyrin-G has been assigned to the C-terminal regulatory domain but not to the N-terminal binding domain (Mohler et al. 2002). We do not know the reason for such discrepancy of the specificity of ankyrin isoforms between our study and studies on cardiac cells.

One possible biophysical mechanism of suppression of Ina-p by ankyrin-G may be that two populations of Nav1.6 channels with distinct inactivation properties coexist and ankyrin-G may reduce the number of channels that fail to inactivate. Alternatively, ankyrin-G may uniformly affect the whole population of sodium channels and allow channels to more readily adopt a mode in which channels faithfully inactivate by the fast inactivation mechanism. In either case, the amplitude of peak sodium current might be slightly reduced compared with cells expressing only hNav1.6. However, we were not able to detect any significant effect of ankyrin-G on the current density. It is possible that a minor decrease of the current amplitude by ankyrin-G was within the variability of the current density among cells or that slight enhancement of recruitment of Nav1.6 proteins on the cell membrane by ankyrin-G might have balanced the slight reduction of the current amplitude.

It is possible that binding of ankyrin-G changes the local structure of the cytoplasmic domains of hNav1.6 protein, thereby modulating inactivation gating. Recent studies have identified the critical consensus motif, consisting of nine amino acids, in the II-III cytoplasmic loop of Nav channels, that binds to the MB-domain of ankyrin-G and mediates clustering of Nav channels at the node of Ranvier and initial segment (Garrido et al. 2003; Lemaillet et al. 2003). Mutation in the II–III loop has been reported to affect inactivation gating of Nav channels (Kuzmenkin et al. 2003). In addition, ankyrin is known to physically bind to the III–IV linker (Bouzidi et al. 2002) that plays a critical role in inactivation gating. On the other hand, lna-p significantly decreases after long whole cell recordings, and Ina-p is rarely observed when CsF was utilized as the pipette solution. It is therefore possible that Ina-p is sensitive to the cellular metabolic state, for example, through some state of phosphorylation of the cytoplasmic regions of the Nav1.6 channels. Ankyrin-G, a large globular protein, may sterically affect such phosphorylation of Nav1.6 channel proteins. Understanding the mechanisms of modulation of inactivation by ankyrin-G must await further studies such as single-channel recording and biochemical studies.

Possible physiological and pathophysiological implications of “ankyrin-G-dependent gating” of Nav1.6 channels

Functional diversity of Nav channels underlies the complexity of mammalian neuron excitability that include integration, pacemaking, and conduction. Ina-p has been reported to be present in cell soma (Kay et al. 1998) and dendrites (Magistretti et al. 1999). Nav1.6 knockout mice show less Ina-p in accordance with fewer spontaneous firings (Raman et al. 1997), indicating that Nav1.6 is the major contributor of neuronal Ina-p. On the other hand, Nav1.6 is the predominant axonal sodium channel. Robust Ina-p of Nav1.6 has been reported in experiments with heterologous cells, including Xenopus oocytes (Goldin 1999) and HEK293 cells (Burbidge et al. 2002). In the present study, expression of ankyrin-G rendered the Nav1.6 current more completely inactivated such that Nav currents resembled those of the mammalian node (Schwarz et al. 1995). These findings raise the possibility that Ina-p could be the default state property of Nav1.6 channels and that the amplitude of Ina-p is suppressed through interaction with ankyrin-G protein at the node of Ranvier. Ankyrin-dependent inhibition of Ina-p at the node of Ranvier may play a role in ensuring generation of individual action potentials after single action potentials that propagate from the upstream node. On the other hand, in the cell soma, Ina-p based on Nav1.6 escapes inhibition by ankyrin-G probably because ankyrin-G expression is extremely low in the cell soma (Kordeli et al. 1995). Interaction of Nav1.6 channels with ankyrin-G may be one of the molecular bases for the functional diversity of Nav channels in neurons.

If ankyrin-G regulates inactivation gating of in vivo axons, inactivation behaviors of nodal Nav channels may change under some pathological conditions such as demyelination. Abnormal axonal excitations arising spontaneously, and frequently firing, are known to be observed in demyelinating conditions, such as in multiple sclerosis (Smith and Hall 2001) or after nerve injury (Sommer 2003). In multiple sclerosis patients, as well as in experimental autoimmune encephalomyelitis (EAE), a mouse model of multiple sclerosis, a Na-Ca exchanger is colocalized with Nav1.6 along the demyelinated nerve (Craner et al. 2004). Increased sodium influx through Nav1.6 has been proposed to cause an increase in intracellular calcium concentration by activating the Na-Ca exchanger (Craner et al. 2004). Demyelination and nerve injury may downregulate ankyrin-G at the node, leaving a population of Nav1.6 in the membrane. This may lead to disinhibition of Ina-p, increasing sodium influx. This idea could be tested by examination of expression patterns of Nav1.6 and ankyrin-G in such demyelinating nerves.


This work was supported by Grants in Aids for Scientific Research to K. Hayasaka and Y. Okamura.


We thank Dr. J. Trimmer for providing pan-sodium channel antibody, K58/35.4, and help in immunoblot experiment in COS1 cells, Dr. M. Chahine for a critical reading of the manuscript and providing the rat sodium channel β1 subunit, Drs. K. Imoto and M. Noda for providing rat brain Nav1.2 clones, Dr. Alan Goldin for providing mouse SCN8a plasmid, which was used in early phase of this study, Dr. Izumi-Nakaseko for exchanging preliminary results of mouse Nav1.6 channel in HEK293 cells, M. Sasaki for help in RT-PCR, and Dr. T. McCormack for reading the manuscript. Accession number for the full-length human Nav1.6 is NCBI AB027567.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract