The C-terminal domain of βIV-spectrin is crucial for KCNQ2 aggregation and excitability at nodes of Ranvier


  • Jérôme J. Devaux

    1. Département Signalisation Neuronale, CRN2M, UMR 6231, CNRS, Université de la Méditerranée-Université Paul Cézanne, IFR Jean Roche, Marseille, France
    Search for more papers by this author

Corresponding author J. J. Devaux: Département Signalisation Neuronale, CRN2M, UMR 6231, CNRS, Faculté de Médecine – Secteur Nord, Université de la Méditerranée–Université Paul Cézanne, CS80011, Bd Pierre Dramard, 13344 Marseille Cedex 15, France.  Email:


The spectrin cytoskeleton has an important function in the targeting of proteins to excitable membrane domains. In axons, βIV-spectrin stabilizes voltage-gated sodium (Nav) channel clusters at nodes of Ranvier and axon initial segments, two regions crucial for the generation and conduction of action potentials. Here, I investigated the physiology of the neuromuscular junction and peripheral nerves in quivering-3J mice, which show a frame-shift base insertion in the Spnb4 gene and lack the C-terminus of βIV-spectrin. The quivering-3J mice show prominent spontaneous and evoked hyperactivities at diaphragm neuromuscular junctions. These neuromyotonic and myokymic discharges were more prominent in adult animals when tremors and ataxia were pronounced. Recordings of sciatic and phrenic nerves showed that the hyperactivities originate in myelinated axons distally from nerve terminals. Axon and myelin structure in the PNS were unaffected in quivering-3J mice. Of interest, KCNQ2 subunit aggregates were undetectable at PNS and CNS nodes, whereas Nav and Kv1.1/Kv1.2 channels were properly concentrated at nodal and juxtaparanodal regions, respectively. The protein level of KCNQ2 subunits was normal in mutant animals, suggesting that KCNQ2 subunit absence stems from clustering or trafficking defects in axons. The quivering-3J nodes also presented high densities of ankyrin-G and CK2α, two cytosolic molecules involved with aggregating Nav and KCNQ2/3 channels in axons. Because βIV-spectrin does not interact with KCNQ2/3 subunits, it is suspected that βIV-spectrin regulates the distribution of KCNQ2/3 subunits in axonal subdomains via regulatory partners. Retigabine, an activator of KCNQ2/3 channels, attenuated the repetitive activities in quivering-3J mice, suggesting that depletion of KCNQ2 subunits at nodes initiates neuromyotonic/myokymic discharges. These findings demonstrate that spectrin cytoskeleton finely regulates ion channel distribution and implicates KCNQ2/3 subunits in axonal excitability and in myokymia aetiology.


compound action potential


compound muscle action potential


casein kinase 2α




disrupted in schizophrenia protein-1


excitatory amino acid transporter 4


enhanced green fluorescent protein


influenza haemagglutinin


voltage-gated potassium channel


voltage-gated sodium channel


pleckstrin homology


schwannomin-interacting protein-1


The nodes of Ranvier play a central role in controlling the rapid propagation of the action potential and nerve excitability. These functions are provided by high densities of voltage-gated sodium (Nav) and potassium (Kv) channels at nodes. In particular, Nav1.6 subunits co-localize with the slow-gating KCNQ2/3 subunits at nodes and axon initial segments (Caldwell et al. 2000; Devaux et al. 2004; Pan et al. 2006). Mutations in SCN8A, the gene encoding Nav1.6, cause important motor deficits in humans and mice (Trudeau et al. 2006), whereas mutations of the KCNQ2 gene in humans are responsible for benign familial neonatal convulsions and myokymia, a peripheral nerve excitability disorder (Dedek et al. 2001; Wuttke et al. 2007). This latter phenotype suggests that KCNQ2 channels are implicated in axon excitability, but direct evidence for this is lacking.

Recent investigations have highlighted the intricate mechanisms regulating the organization of nodes of Ranvier, notably the key function of the cortical cytoskeleton (Susuki & Rasband, 2008). Several isoforms of spectrin and ankyrin are found in mammal myelinated axons. Ankyrin-G and βIV-spectrin are enriched at nodes, whereas ankyrin-B, αII-spectrin and βII-spectrin are found at paranodal regions (Berghs et al. 2000; Ogawa et al. 2006). Ankyrin-G organizes node composition by fixing intrinsic nodal membrane components (Neurofascin-186, NrCAM, Nav channels, Kv3.1b, KCNQ2/3) (Davis et al. 1996; Devaux et al. 2003; Garrido et al. 2003; Lemaillet et al. 2003; Devaux et al. 2004). Alterations in ankyrin-G result in the loss of nodal specialization (Jenkins & Bennett, 2002; Yang et al. 2007).

βIV-spectrin is thought to stabilize the anchorage of nodal components to actin cytoskeleton. However, its elaborated structure and alternative splicing suggest a more complex function. βIV-spectrin comprises an N-terminal calponin homology domain which binds actin, 17 spectrin repeats (the 15th repeat binds ankyrin), a specific domain, and a C-terminal pleckstrin homology (PH) domain. Two alternative splice variants differing at the N-terminus (Σ1 and 6) are found at nodes (Lacas-Gervais et al. 2004). Genetical deletion of βIV-spectrin results in decreased densities of ankyrin-G and Nav channels at nodes (Komada & Soriano, 2002; Yang et al. 2004).

So far only a few studies have addressed the importance of spectrin cytoskeleton in neuron physiology. The quivering-3J mouse (Spnb4qv-3J/qv-3J) presents an autosomal recessive mutation resulting in the C-terminal deletion of βIV-spectrin. In these animals, the granule cells of the dentate gyrus show a decreased action potential threshold (Winkels et al. 2009). Here, I examined whether βIV-spectrin deletion affects nerve and neuromuscular excitability. I found that quivering-3J mice exhibit repetitive muscle contractions of axonal origin. Mutation of βIV-spectrin also resulted in the selective loss of KCNQ2 channels in quivering-3J nodes. These data revealed that mutations in βIV-spectrin may generate channelopathies and peripheral nerve hyperexcitability disorders.


Mutant mice

Spnb4+/qv-3J mice in C57BL/6J background were obtained from The Jackson Laboratory (Bar Harbor, ME, USA) and were maintained by heterozygous crosses. Three-month-old male and female homozygote Spnb4qv-3J/qv-3J mice (here named quivering-3J mice) and their wild-type (WT; Spnb4+/+) littermates were used unless otherwise indicated. The mice were killed with CO2 according to the European Community's guiding principles on the care and use of animals (86/609/CEE) and the experiments complied with The Journal of Physiology's policies and regulations (Drummond, 2009).

Nerve electrophysiology

The sciatic nerves were quickly dissected and transferred into artificial cerebrospinal fluid (ACSF) equilibrated with 95% O2–5% CO2, which contained (in mm): 126 NaCl, 3 KCl, 2 CaCl2, 2 MgSO4, 1.25 NaH2PO4, 26 NaHCO3, and 10 dextrose, pH 7.4–7.5. Recordings of nerve compound action potentials (CAPs) were made at 36°C in a three-compartment recording chamber as previously described (Lonigro & Devaux, 2009). Drugs were applied in the central compartment and measurements were made once the effects had reached a steady state. Nerves were continuously stimulated at a frequency of 0.25 Hz. The delay and duration of the CAPs were calculated at half the maximal amplitude. For recruitment analysis, nerves were stimulated at increasing intensities. For refractory period analysis, two stimuli were applied at different intervals, and the amplitude of the second CAP was measured.

Compound muscle action potential (CMAP) recordings

Diaphragms with rib cage and phrenic nerves were dissected out, and pinned down in a recording chamber as previously described (Zhou et al. 1999). Recordings were made at 36°C in Rees's mammalian solution, which contained (in mm) 110 NaCl, 5 KCl, 2 CaCl2, 1 MgCl2, 5 Bes, 25 NaHCO3, 0.4 l-glutamine, 0.3 glutamic acid, and 10 dextrose, pH 7.4–7.5 complemented with 2 mg l−1 thiamine and 36 μm choline chloride. The nerves were stimulated (40 μs duration) with a bipolar electrode while the nerve CAPs were recorded from the cut end with a suction electrode. The CMAPs were recorded simultaneously with a surface electrode pressed gently on the diaphragm at the same location in all experiments. Drugs were applied in the recording chamber. Electrophysiological signals were amplified, digitized at 500 kHz, and stored on a hard disk. Analysis was performed using pCLAMP 9.2 software (Molecular Devices, Sunnyvale, CA, USA).


Human embryonic kidney (HEK) cells were transiently transfected with KCNQ2 (AF490773), KCNQ3 (AF091247), Σ6 βIV-spectrin (AB055621), and EGFP (to identify transfected cells) using JetPEI (Polyplus-transfection, Illkirch, France). One day after transfection, cells were trypsinated, and voltage-clamp recordings were performed at room temperature (22°C) in the whole-cell configuration the following day using an Axopatch 200A amplifier (Molecular Devices). Only cells with series resistances in the order of 10–20 MΩ were used. Data were acquired using pCLAMP 9.0 software, filtered at 1 kHz and sampled at 50 kHz. Patch pipettes (1.5–2 MΩ) were filled with internal pipette solution, which contained (in mm): 140 KCl, 1 CaCl2, 2 MgCl2, 10 EGTA, 0.4 NaGTP, 2 Na2ATP, and 10 Hepes, pH 7.2. Cells were continuously perfused with external Locke's solution, which contained (in mm): 2 KCl, 154 NaCl, 2 CaCl2, and 10 Hepes, pH 7.4. The data were fitted to a single Boltzmann distribution of the following form:

display math

where V is the test potential, V50 is the half-activation potential, k is the slope, and max is the maximal amplitude for the Boltzmann distribution.


Sciatic nerves were fixed in 2% paraformaldehyde in 0.1 m phosphate-buffered saline (PBS) for 60 min at 4°C, rinsed in PBS, then teased on glass slides and stored at −20°C. CNS tissues were also fixed by transcardiac perfusion, cryoprotected, and cut into 10 μm sections. Frozen sections and teased fibres were permeabilized by immersion in −20°C acetone for 10 min, blocked for 1 h with a solution containing 5% fish skin gelatin and 0.1% Triton X-100 in PBS and incubated overnight at 4°C with various combinations of primary antibodies – rabbit anti-KCNQ2 (1/200; Lonigro & Devaux, 2009), Kv3.1b (1/100; Alomone Labs, Jerusalem, Israel), Kv1.1 (1/100; Alomone Labs), ankyrin-G (1/1000; Bouzidi et al. 2002), or Caspr (1/1000; Menegoz et al. 1997); mouse anti-PanNav channels (K58/35; 1:500; Sigma-Aldrich, St Louis, MO, USA) or Kv1.2 (1/200; UC Davis/NINDS/NIMH NeuroMab Facility, Davis, CA, USA); goat anti-contactin (1/200; R&D Systems, Minneapolis, MN, USA) or CK2α (1/200; Santa Cruz Biotechnology, Santa Cruz, CA, USA). The slides were then washed several times and incubated with the appropriate Alexa conjugated secondary antibodies (1/500; Invitrogen, Paisley, UK). Slides were mounted with Mowiol plus 2% DABCO, and examined using an ApoTome fluorescence microscope (ApoTome, AxioObserver and AxioCam MRm, Carl Zeiss MicroImaging GmbH, Jena, Germany). Digital images were manipulated into figures with Adobe Photoshop and CorelDraw. For quantitative study, teased fibres from four animals were examined for all genotypes, the length of ∼100 individual Caspr-positive paranodes and intercalated nodes were measured in each animal. For quantification of axonal diameter, sciatic nerves were fixed in 2% paraformaldehyde and 2% glutaraldehyde in 0.1 m phosphate buffer (PB) overnight at 4°C and post-fixed in 1% OsO4 in 0.1 m PB for 1 h. Nerves were dehydrated and embedded in epoxy resin. Transverse semi-thin sections were stained with toluidine blue and examined by light microscopy with a 100× objective lens. The diameter of ∼500 axons was measured in two animals for each group.

Immunoblots and immunoprecipitation

Brain and spinal cord membranes were homogenized in ice cold 0.32 m sucrose, 5 mm Tris-HCl, pH 7.4 containing protease inhibitors (2 mm EDTA; 1 μg ml−1 leupeptin and aprotinin; and 0.5 mm phenylmethylsulfonyl fluoride, Sigma-Aldrich) and the homogenates were centrifuged for 10 min at 750 g. The supernatants were sedimented for 60 min at 17,000 g, the resulting pellets were resuspended in 1 mm EDTA, 5 mm Tris-HCl, pH 8.2, plus protease inhibitors, homogenized and placed on ice for 30 min. The lysate membranes were then centrifuged 40 min at 27,000 g and the pellet (P3) was resuspended in 150 mm NaCl, 25 mm Tris-HCl, pH 7.4, plus protease inhibitors and stored at −80°C. Proteins (100 μg) were loaded on a 7.5% SDS-PAGE gel, then transferred onto a nitrocellulose membrane. Membranes were blocked for 1 h with 5% powdered skim milk, 0.5% Tween-20 in PBS and incubated with antibodies against KCNQ2 (1/2000), KCNQ3 (1/2000; Alomone Labs), Kv1.1 (1/1000), Kv1.2 (1/2000), PanNav (1/10000), or Caspr-2 (1/10000; Traka et al. 2003). After several washes, the blots were incubated with the appropriate peroxidase-coupled secondary antibodies (1/5000; Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) for 1 h, washed several times and revealed using BM chemiluminescence kit (Roche, Basel, Switzerland).

For immunoprecipitation, the P3 pellet was solubilized in 1.25% Triton X-100, 0.1 mm KCl, 10 mm Tris-HCl, pH 7.4 plus protein inhibitors, stored on ice for 15 min, then centrifuged at 27,000 g for 60 min. The supernatant (200 μg of protein) was incubated overnight with the following antisera in the presence of 0.1% BSA: rabbit antiserum against KCNQ2 (Lonigro & Devaux, 2009) and mouse antiserum against βIV-spectrin (Berghs et al. 2000). Each sample was then incubated for 30 min at 4°C with 30 μl of Protein G–agarose beads (Invitrogen). After washing three times with 0.1% Triton X-100, 1% BSA plus protease inhibitors, bound proteins were released by boiling in 20 μl of SDS sample buffer for 2 min at 90°C. The released proteins were transferred to a 7.5% SDS-PAGE gel and processed as described above for immunoblotting.

For Triton X-100 extraction, the P3 pellet was solubilized with 1% Triton X-100, 0.1 mm KCl, 10 mm Tris-HCl, pH 7.4 plus protein inhibitors, agitated on ice for 30 min, and centrifuged at 27,000 g for 60 min. Pellets were resuspended in solubilization buffer. Equal amounts of Triton-soluble and Triton-insoluble fractions were solubilized in SDS sample buffer for 2 min at 90°C, and separated on a 7.5% SDS-PAGE gel. Immunoblots were then revealed with rabbit antiserum against KCNQ2.

Cell lysates were also prepared from COS-7 cells transiently transfected for V5-tagged KCNQ2 and myc-tagged Σ6 βIV-spectrin. Cells were solubilized in 1.25% Triton X-100, 0.1 mm KCl, 10 mm Tris-HCl, pH 7.4 plus protein inhibitors, agitated on ice for 15 min, then centrifuged at 27,000 g for 60 min. Immunoprecipitation were performed as described above. The following antisera were used: mouse antiserum against myc (Roche) and rat antiserum against HA (Roche).


Neuromuscular disorders in quivering-3J mice

The quivering-3J mutation results in the deletion of the PH domain of Σ1 and Σ6 splice variants, but does not affect the 15th spectrin repeat that binds ankyrin-G (Parkinson et al. 2001). At an adult age, these mice show pronounced tremors and ataxia (Parkinson et al. 2001; Yang et al. 2004), two phenotypes that may have both CNS and PNS components. To investigate the possibility that the quivering-3J phenotype is linked to neuromuscular dysfunctions, I recorded isolated diaphragm–phrenic nerve preparations from 3-month-old (when tremors and ataxia are important) quivering-3J mice and their WT littermates. The CMAPs evoked in quivering-3J mice by phrenic nerve stimulation consisted of triphasic potentials and appeared normal compared to WT (Fig. 1). Neuromuscular transmission was one to one and there were no signs of transmission defects or fatigability. However, evoked-CMAPs from quivering-3J mice were followed by multiple repetitive muscle activities of varying amplitude with a mean frequency of 89 Hz (3 to 23 spikes within the 100 ms following the initial CMAP; Fig. 1B). The post-spike activities were smaller in amplitude than the initial CMAP (from a few hundreds of microvolts to several millivolts), and reached half of the initial CMAP amplitude in some preparations (Supplemental Fig. 1). These repetitive activities represented action potentials initiated by a small number of the muscle fibres from the same motor unit. The repetitive activities were evoked by nerve stimulation, but not by direct muscle fibre stimulation (Suppl. Fig. 1). To test whether similar activities arise in axons, I recorded the CAPs from phrenic nerves and CMAPs simultaneously on the same diaphragm–phrenic nerve preparations. Phrenic nerve CAPs consisted of a positive monophasic potential that immediately followed the stimulus artifact due to the short length of the nerve segments. I found that repetitive activities occurred in phrenic nerve axons concomitantly with diaphragm muscle contractions (Fig. 1B). The amplitude of the repetitive activities ranged from tens to one hundred microvolts and reflected extracellular recording of action potentials generated by individual axons. These individual axonal activities are prone to activate single motor units or a fraction of the motor unit, and are likely to account for the neuromyotonic/myokymic activities observed in quivering-3J mice.

Figure 1.

Neuromuscular phenotype of quivering-3J mice
A, lower panel, scheme of the recording apparatus. The CMAPs and nerve CAPs were recorded simultaneously from the same preparation of isolated diaphragm–phrenic nerves at position a and b, respectively. A and B, representative CMAPs (a) and CAPs (b) from 3-month-old WT (A; n= 12 animals) and quivering-3J mice (B; n= 19 animals) after stimulating the phrenic nerve. CMAPs consisted of initially positive triphasic potentials, whereas nerve CAPs consisted of positive monophasic potentials that developed closely after the artifact of stimulation (arrows). In quivering-3J mice, nerve stimulation resulted in repetitive muscle and nerve activities of small amplitude (arrowheads) that reflected the firing of individual muscle and nerve fibres. Insets show amplified details of the recordings in boxed areas. C, representative CMAPs recorded prior to nerve stimulation from WT (lower trace; n= 11 animals) and quivering-3J mice (upper trace; n= 17 animals). Spontaneous muscle activities with a mean frequency of 35 Hz were recorded in quivering-3J diaphragms. These were blocked by bath application of d-tubocurarine (20 μm; n= 6 animals). Insets show enlargements of muscle activity recorded before (arrow with a single asterisk) and after bath application of d-tubocurarine (arrow with a double asterisk). D and E, the number of evoked hyperactivities (D) and of spontaneous activities (over a period of 100 s; E) are represented as a function of the age of the animals. Evoked-activities were more sustained in animals older than P40 compared to animals younger than P40 (P < 0.005 by two-tailed Student's unpaired t test for two samples of equal variance). No evoked or spontaneous activities were detected in WT animals. Horizontal lines represent the mean. Spontaneous activities are expressed as log plot.

Tremors and hindlimb clasping are observed as early as 21 day postnatal (P21) in quivering-3J mice and become more sustained in older animals. I thus examined the developmental changes in excitability in the quivering-3J mice. Out of seven animals under P40 tested (from P21 to P40), only three presented hyperexcitabilities (Fig. 1D). In contrast, all animals older than P40 tested (from P40 to P190; n= 12) exhibited sustained repetitive activities (Fig. 1D).

In addition to evoked hyperexcitabilities, quivering-3J mice exhibited spontaneous muscle activities prior to stimulation ranging from tens to several hundred microvolts. Given their small amplitude, these activities reflected the contractions of a small number of muscle fibres. To quantify these, muscle activity was recorded during a period of 100 s. Out of seventeen animals, ten showed spontaneous firing (mean frequency of 35 Hz ranging from 0.02 to 90 Hz; Fig. 1C and E). These myokymic discharges appeared independently of the age of the animals. Worth noting, myokymic discharges occurred in the three animals under P40 showing evoked hyperactivities, revealing that 77% of the mice with evoked hyperactivities show spontaneous firing. Blockade of the neuromuscular transmission using d-tubocurarine (20 μm; n= 6) abolished spontaneous contractions in quivering-3J mice (Fig. 1C), further corroborating the axonal origin of these activities.

I next examined whether hyperactivities could be attenuated by retigabine, an anticonvulsant drug which is a potent activator of the KCNQ2/3 channels expressed in peripheral axons (Main et al. 2000; Devaux et al. 2004). Retigabine (20 μm; n= 4) attenuated both evoked and spontaneous hyperactivities in quivering-3J mice (Suppl. Fig. 1). However, retigabine affected the amplitude and shape of the CMAPs, indicating it also modulates muscle fibre K+ conductances.

Axonal re-excitation in quivering-3J mice

To corroborate the above findings, nerve excitability was evaluated in sciatic nerves from 3-month-old quivering-3J mice. The amplitude and conduction velocity of the quivering-3J CAPs did not differ significantly from WT (Table 1 and Fig. 2A). In addition, the recruitment and refractory period of sciatic nerve axons were unaffected by the quivering-3J mutation (data not shown). These results ruled out the presence of hypomyelination, demyelination, or axonal degeneration in quivering-3J mice. Sciatic nerve CAPs were then examined at high gain amplification to detect single axonal activities. Prominent hyperactivities were recorded at 36°C in quivering-3J animals following supramaximal stimulation of the nerves (Fig. 2). These hyperactivities persisted for several seconds after the stimulus, but were hardly discernable prior to nerve stimulation. WT nerves did not display such hyperactivities. This demonstrated that neuromyotonic/myokymic activities originate in the distal part of the peripheral nerves and occur in the absence of presynaptic terminals. I next examined whether these hyperactivities are also attenuated by retigabine. Retigabine (20 μm; n= 5) abolished evoked hyperactivities in quivering-3J nerves (Fig. 2). It also modestly increased the duration of the CAPs and delayed the recovery phase of the CAPs in both WT and quivering-3J mice.

Table 1.  Electrophysiological characteristics of WT and quivering-3J sciatic nerves
 WT quivering-3J
  1. CVVmax and CVV½: conduction velocity at maximal amplitude and at half the maximal amplitude, respectively. The duration was measured at half the maximal amplitude. *Significantly different; P < 0.01; two-tailed Student's unpaired t test for samples of equal variance. n represents the number of nerves tested.

Amplitude (mV)7.4 ± 2.87.4 ± 1.9
Duration (ms)0.44 ± 0.12*0.28 ± 0.05
CVVmax (m s−1)34.1 ± 10.733.0 ± 6.4
CVV½ (m s−1)61.5 ± 24.950.7 ± 16.4
n 13 (7 animals)7 (4 animals)
Figure 2.

Electrophysiological and morphological characteristics of quivering-3J sciatic nerves
A, representative sciatic nerve CAPs from 3-month-old WT (n= 13 nerves from 7 animals) and quivering-3J mice (n= 7 nerves from 4 animals). CAPs from quivering-3J sciatic nerves had a normal amplitude and conduction velocity compared to WT mice. However, repetitive activities (arrowheads) were revealed at high gain amplification in quivering-3J axons (right panel; WT and quivering-3J traces are superimposed). These reflected action potentials initiated by individual axons. B, bath application of retigabine (20 μm; grey trace) abolished the repetitive activities in quivering-3J sciatic nerves (right panel; n= 5 nerves), but had no effect on WT activity (left panel; n= 6 nerves). In A and B, arrows indicate the artifact of stimulation. Insets show spaced and amplified details of the recordings in boxed areas. C, the distributions of paranode length, node length, and axonal diameter were normal in sciatic nerves from quivering-3J and WT mice (P > 0.05 by Student's unpaired t test and Kolmogorov–Smirnov test). D, transverse epoxy sections of sciatic nerves from WT (left panel) and quivering-3J (right panel) mice stained with toluidine blue. Scale bar: 10 μm.

Morphology and organization of the myelinated fibres in quivering-3J mice

To investigate the possibility that axon morphology or composition is perturbed in these animals, teased fibres were prepared from sciatic nerves of 3-month-old WT and quivering-3J mice and were immunostained for Caspr or contactin to label the paranodes. In addition, transverse semi-thin resin sections were prepared for optical microscopy. Paranode length, node length and axon diameter were normal in quivering-3J sciatic nerves (Fig. 2C and D and Suppl. Table 1). No signs of hypomyelination, demyelination, or degenerating axons were found.

Albeit mutated, βIV-spectrin is still targeted at quivering-3J nodes (Yang et al. 2004). The Nav channel and its anchoring partners, ankyrin-G and CK2α, were also properly clustered at WT and quivering-3J nodes (Fig. 3 and Suppl. Fig. 2). Nav1.6 subunits, but not Nav1.2, were detected at WT and quivering-3J nodes, as previously reported (Yang et al. 2004). Because neuromyotonia and myokymia are generally associated to potassium channel dysfunctioning (Gutmann & Gutmann, 2004; Vincent et al. 2006), I examined the distribution of Kv1.1, Kv1.2 and KCNQ2 channels in quivering-3J nerves. Kv1.1 (Fig. 3) and Kv1.2 (Fig. 4) subunits were appropriately concentrated at juxtaparanodes in quivering-3J mice. In contrast, KCNQ2 channels were not detectable in quivering-3J nodes, albeit they were clustered at high densities in WT nodes (Fig. 3).

Figure 3.

The localization of KCNQ2 channels is affected in quivering-3J PNS nerves
These are teased sciatic nerve fibres from 3-month-old WT and quivering-3J mice double-stained for Nav channels using a monoclonal antibody that recognizes all Nav isoforms (PanNav; A), Kv1.1 (B), KCNQ2 (C), and Caspr (A) or contactin (Con.; B and C) to label paranodes. In both WT and quivering-3J mice, Nav channels are concentrated at nodes of Ranvier and are flanked by Caspr/contactin labelled paranodes (A). Kv1.1 channels are also properly sequestrated at juxtaparanodes in both strains (B). KCNQ2 subunits are concentrated at WT nodes, but are not detectable at quivering-3J nodes (double arrowheads; C). Scale bar: 5 μm.

Figure 4.

The localization of KCNQ2 channels is altered at quivering-3J CNS nodes
These are longitudinal sections of the ventral white matter of the cervical spinal cord (A and B) and horizontal sections of the CA3 region of the hippocampus (C and D) from 3-month-old WT and quivering-3J mice. In WT mice, KCNQ2 channels are enriched at CNS nodes flanked by juxtaparanodal Kv1.2 staining (arrowheads; A), and co-localized with Nav channels at axon initial segments (arrowheads; C). In quivering-3J mice, KCNQ2 channel clusters are undetectable at CNS nodes (double arrowheads; B), but are still found, albeit at lower density, in axon initial segments in the hippocampus (double arrowheads; D). Scale bar: 10 μm.

KCNQ2 channels are absent from PNS and CNS quivering-3J nodes

To confirm the above observation, the distribution of KCNQ2 and Nav channels was examined in six different quivering-3J mice and their WT littermates. In all quivering-3J mice, KCNQ2 channels were undetectable at nodes, whereas Nav channels were properly clustered. Similar results were obtained on fixed and unfixed tissues at all ages tested including P10.

Next, I examined KCNQ2 subunit distribution in CNS axons. In brain, cerebellum and spinal cord, KCNQ2 subunits were undetectable at quivering-3J nodes (Fig. 4A–B and Suppl. Fig. 3). In contrast, Kv3.1b and Nav subunits were properly addressed at nodes (Suppl. Fig. 3), and Kv1.2 channels were found concentrated at juxtaparanodes (Fig. 4). Because KCNQ2 and Nav subunits are also expressed at high densities in axon initial segments in WT animals, I examined whether the distribution of the ion channels is perturbed at these sites. In the hippocampus of mutant animals, clusters of KCNQ2 and Nav channels were still detectable at axon initial segments, but displayed decreased staining intensities (Fig. 4). No aberrant accumulation of KCNQ2 subunits was found in intracellular compartments of quivering-3J neurons. In the spinal cord of mutant animals, the distribution of KCNQ2 channels was more heterogeneous. In a general manner, the density of Nav channels at initial segments was importantly decreased in spinal cord neurons, as in hippocampus and cerebellum. A few neurons exhibited KCNQ2 aggregates at initial segments; however, KCNQ2 subunits were not readily detectable at initial segments in many spinal neurons.

The expression of Kv subunits is normal in quivering-3J mice

To determine whether KCNQ2 loss results from degradation or mis-expression, protein levels were examined in quivering-3J and WT brain by Western blot. The levels of KCNQ2, KCNQ3, Kv1.1, Kv1.2, but also of Caspr2, a cell-adhesion molecule crucial for Kv1.1/Kv1.2 localization (Poliak et al. 2003), were normal in quivering-3J mice (Fig. 5A; n= 3). In contrast, the level of Nav channels were decreased in mutant animals compared to WT (Fig. 5A and Suppl. Fig. 4A), which is in keeping with the decreased Nav staining intensity at CNS nodes and initial segments.

Figure 5.

Protein levels are normal in quivering-3J mice
A, membrane proteins (100 μg) from WT (n= 3) and quivering-3J (n= 3) brains were separated by electrophoresis, and immunoblotted as indicated. The levels of KCNQ2, KCNQ3, Kv1.1, Kv1.2 and Caspr2 were not significantly different in quivering-3J mice. By contrast, Nav channels (PanNav) showed a decreased level in quivering-3J animals. B, brain membranes (100 μg) from WT mice were immunoprecipitated for KCNQ2 and βIV-spectrin (SPNB4), then immunoblotted for KCNQ2, βIV-spectrin, or ankyrin-G (as positive control). A ∼170 kDa isoform of ankyrin-G was pulled down with both KCNQ2 and βIV-spectrin. KCNQ2 was immunoprecipitated by the KCNQ2 antiserum, but not by βIV-spectrin. Conversely, βIV-spectrin was immunoprecipitated by the βIV-spectrin antiserum, but not by KCNQ2. MW markers are shown on the left (in kDa).

βIV-spectrin and KCNQ2 both associate to the cytoskeletal components within the Triton X-100-insoluble fraction of brain membrane (Berghs et al. 2000; Cooper et al. 2000). If βIV-spectrin deletion affects KCNQ2 anchoring, then KCNQ2 solubility ought to be increased in mutant animals. In quivering-3J brain membrane, KCNQ2 subunits were found within the detergent-insoluble fraction, and KCNQ2 solubility was not significantly affected compared to WT (Suppl. Fig. 4). Thus, βIV-spectrin deletion did not seem to modulate the interaction of KCNQ2 with the cytoskeleton or with membrane proteins.

βIV-spectrin does not interact with KCNQ2 channels

The above findings suggest that βIV-spectrin may interact with and cluster KCNQ2 channels at nodes. To test this hypothesis, co-immunoprecipitation experiments were performed in solubilized WT brain membranes. A βIV-spectrin antiserum immunoprecipitated two βIV-spectrin isoforms (∼160 and 270 kDa) and ankyrin-G isoforms (∼190 and 240 kDa), but not KCNQ2 (Fig. 5B). Conversely, the KCNQ2 antiserum immunoprecipitated KCNQ2 and ankyrin-G isoforms, but not βIV-spectrin (Fig. 5B). Co-immunoprecipitation experiments were also performed on solubilized COS-7 cells transfected with myc-tagged Σ6 βIV-spectrin and V5-tagged KCNQ2 subunits (Suppl. Fig. 4B). Myc antibodies immunoprecipitated Σ6 βIV-spectrin in COS-7 cells, but not KCNQ2. Reciprocally, V5 antibodies immunoprecipitated KCNQ2, but not βIV-spectrin (Suppl. Fig. 4B). Altogether, these results indicate that KCNQ2 does not interact directly with βIV-spectrin.

Finally, to test whether βIV-spectrin may modulate KCNQ2/3 channel trafficking to the cell surface, the current density and kinetics of heteromeric KCNQ2/3 channels were measured in presence of Σ6 βIV-spectrin (Suppl. Fig. 5). βIV-spectrin had no major impact on heteromeric KCNQ2/3 current density or voltage dependency (Suppl. Fig. 5).


In the present study, I show that a mutation in βIV-spectrin induces a peripheral nerve hyperexcitability disorder in quivering-3J mice. This phenotype implicated KCNQ2 channels and was attenuated by retigabine. In particular, this study shows that the C-terminal domain of βIV-spectrin is crucial for KCNQ2 channel trafficking or clustering at nodes.

Physiological and pathological implication of KCNQ2/3 channels in axons

Neuromyotonia and myokymia are excitability disorders observed in the child and adult which originate within peripheral nerves and are characterized by continuous muscle movements, fasciculation, or cramps (see for review Kleopa & Barchi, 2002; Gutmann & Gutmann, 2004; Vincent et al. 2006). Neuromyotonia and myokymia are commonly associated with channelopathies. Autoantibodies directed toward Kv1.1, Kv1.2, and Kv1.6 subunits are found in patients with acquired neuromyotonia (Hart et al. 1997; Kleopa et al. 2006) and sera from these patients bind juxtaparanodal regions of myelinated axons. Also, mutations in the KCNA1 gene (which encodes Kv1.1; Kleopa & Barchi, 2002) and in the KCNQ2 gene (Dedek et al. 2001; Wuttke et al. 2007) are associated with inherited neuromyotonia and myokymia, respectively. To date, mutations in the SPNB4 gene have not been associated with inherited neuromyotonia, myokymia, or ataxia. This is the first time that a mutation in a cytoskeleton protein has been demonstrated to induce a peripheral nerve hyperexcitability disorder.

In quivering-3J mice, neuromyotonic/myokymic discharges were recorded within the nerves and correlated with the disappearance of KCNQ2 clusters at nodes. In agreement with a previous report (Komada & Soriano, 2002; Yang et al. 2004), I found that PNS axon shape and composition were unaffected in quivering-3J mice, with the exception of the KCNQ2 depletion. I thus concluded that reduced KCNQ2 channel densities at nodes might underlie the neuromyotonic/myokymic discharges. This conclusion is further supported by the observation that retigabine attenuated the repetitive discharges in quivering-3J mice. This latter observation also indicates that KCNQ2 subunits might persist in axons either with a diffuse distribution or at low densities at nodes.

The neuromyotonic/myokymic discharges were more common in animals older than P40 and appeared to parallel myelin maturation. The distribution of Kv1.1 and Kv1.2 channels is known to depend on myelin maturation (Vabnick et al. 1999). During PNS development, Kv1.1/Kv1.2 subunits are first detected at nodes and paranodes, and then are progressively sequestered at juxtaparanodes (Vabnick et al. 1999). Peripheral nerve hyperactivities in quivering-3J mice thus coincide with Kv1.1/Kv1.2 extrusion from nodes. In keeping, KCNA1-null mice exhibit temperature-sensitive hyperexcitabilities at early developmental stages (P20–P40) which are exacerbated by TEA, a KCNQ2/3 blocking agent (Zhou et al. 1999). The present results further corroborate the synergic function of Kv1 and KCNQ2/3 channels in regulating axonal excitability. The fast-gating Kv1 channels prevent re-excitation in developing (Vabnick et al. 1999) and in small myelinated axons (Zhou et al. 1999; Devaux & Gow, 2008). In contrast, given their activation kinetics (Suppl. Fig. 5), the slow-gating KCNQ2/3 channels are prone to prevent slow membrane potential fluctuations and aberrant repetitive firing in mature axons. The exact reasons why tremors and hindlimb clasping were observed earlier in quivering-3J mice than neuromyotonic/myokymic discharges in vitro are unknown. One possibility is that neuromyotonic/myokymic discharges occur earlier in lower limb nerves than in phrenic nerves. Also, the occurrence of hyperactivities may escalate gradually as the axonal length increases. Of interest, this study indicates that axon initial segments show important alterations in quivering-3J mice, notably a decreased Nav channel density in spinal cord, hippocampus and cerebellum. The decreased Nav channels at axon initial segments is likely to account for the reduced neuronal excitability observed in dentate gyrus of quivering-3J mice (Winkels et al. 2009), and may cause the gait abnormalities and the progressive ataxia observed in the mutant mice. In addition to peripheral nerve hyperexcitability, quivering-3J mice show hearing defects (Kubisch et al. 1999; Kharkovets et al. 2000; Parkinson et al. 2001). It will be of interest to examine KCNQ4 subunit distribution in quivering-3J cochlea, as mutations in the KCNQ4 gene are linked to deafness in humans (Kubisch et al. 1999; Kharkovets et al. 2000; Parkinson et al. 2001).

Possible molecular mechanisms for deprivation of KCNQ2 channels

How does βIV-spectrin control KCNQ2 localization? Up to now the axonal distribution of KCNQ2 was believed to closely depend on ankyrin-G (Devaux et al. 2004; Chung et al. 2006; Pan et al. 2006; Rasmussen et al. 2007). Here, I have challenged this view by showing that βIV-spectrin deletion selectively impairs KCNQ2 aggregation at quivering-3J nodes, without affecting ankyrin-G, CK2α, Nav, or Kv3.1b channels (Fig. 3, and Suppl. Figs 2 and 3). KCNQ2 subunits were still detectable at axon initial segments in the hippocampus, and retigabine showed important effects in quivering-3J mice. Thus, KCNQ2 channels are still targeted into the axonal compartment, albeit at a lower density undetectable by fluorescence microscopy. The levels of KCNQ2/3 subunits were also unchanged in quivering-3J mice, indicating that KCNQ2/3 channel expression and turnover are unaffected. Several studies suggest that βIV-spectrin stabilizes protein aggregation at nodes (Komada & Soriano, 2002; Lacas-Gervais et al. 2004; Yang et al. 2004). The possibility that KCNQ2 depletion stems from a lack of stabilization at quivering-3J nodes is unlikely given that KCNQ2 channels were undetectable at nodes at all ages tested. Exactly how βIV-spectrin regulates KCNQ2 distribution is not yet determined. One possibility is that βIV-spectrin orchestrates directly or indirectly KCNQ2/3 channel clustering at nodes. Alternatively, given the role of spectrin in protein transport (Holleran et al. 2001), it is possible that βIV-spectrin is involved in the trafficking of KCNQ2/3 channels into axonal subdomains. The targeted deletion of βIII-spectrin has recently been shown to induce intracellular accumulation of EAAT4 in cerebellar neurons (Stankewich et al. 2010). There were no signs of intracellular accumulation of KCNQ2 channels in quivering-3J neurons. Neither did βIV-spectrin influence the membrane expression of KCNQ2/3 heteromers. However, one cannot exclude that axonal transport of KCNQ2 channels is selectively affected.

I have shown here that KCNQ2 channels do not co-immunoprecipitate βIV-spectrin both in vivo and in vitro. It is thus unlikely that βIV-spectrin directly orchestrates KCNQ2 clustering. Alternatively, βIV-spectrin may bind partners that regulate KCNQ2/3 channel clustering or trafficking. Two novel cytosolic components, SCHIP-1 and CK2α, have recently been identified at nodes (Brechet et al. 2008; Martin et al. 2008). CK2α distribution is not affected by the quivering-3J mutation (Suppl. Fig. 2). In addition, SCHIP-1 is still detectable at nodes in quivering-3J mice (Brechet et al. 2008; Martin et al. 2008), ruling out that deficiencies in SCHIP-1 or CK2α are responsible for KCNQ2 loss. The quivering-3J mutation affects the specific domain and the PH domain of βIV-spectrin. Both domains play elusive functions. The PH domain may bind phosphoinositides, whereas the specific domain contains ERQES and a proline rich domain that may favour protein interaction. So far, only a few proteins have been found to interact with βIV-spectrin (ICA512, ankyrin-G, DISC-1) (Berghs et al. 2000; Morris et al. 2003). In future work, it will be important to identify the molecular partners of βIV-spectrin and their functions in regulating node composition.

In conclusion, this study demonstrates that defects in spectrin cytoskeleton can cause channelopathies and excitability disorders. It also provides a novel model for testing strategies aiming to improve these disorders.



This work was supported by the Association Française contre les Myopathies (MNM2 2006-12180) and by the National Multiple Sclerosis Society (RG 3839A1/T). I thank Dr Matthew Rasband for the original gift of quivering-3J mice, Drs Laurence Goutebroze, Gisèle Alcaraz, Michele Solimena and Irwin Levitan for generously providing antibodies and constructs. Kv1.2 antibody was obtained from the UC Davis/NINDS/NIMH NeuroMab Facility, supported by NIH grant U24NS050606 and maintained by the Department of Pharmacology, School of Medicine, University of California, Davis, CA 95616, USA.