Rapid protein kinase C-dependent reduction of rat skeletal muscle voltage-gated sodium channels by ciliary neurotrophic factor

Abstract

The ciliary neurotrophic factor (CNTF), known to exert long-term myotrophic effects, has not yet been shown to induce a rapid biological response in skeletal muscles. The present in vitro study gives rise to the possibility that CNTF could affect the sodium channel activity implied in the triggering of muscle fibre contraction. Therefore, we investigated the effects of an external CNTF application on macroscopic sodium current (I_Na) in rat native fast-twitch skeletal muscle (flexor digitorum brevis, FDB) by using a cell-attached patch-clamp technique. The I_Na peak amplitude measured at a depolarizing pulse from −100 to −10 mV is rapidly reduced in a time- and dose-dependent manner by CNTF (0.01–20 ng ml⁻¹). The maximal decrease is 25% after 10 min incubation in 2 ng ml⁻¹ CNTF. There was no alteration in activation or inactivation kinetics, or in activation curves constructed from current–voltage relationships in the presence of CNTF. In contrast, the relative I_Na inhibition induced by CNTF is accompanied by a hyperpolarizing shift in the midpoint of the inactivation curves: −6 and −10 mV for the steady-state fast and slow inactivation, respectively. Furthermore, CNTF induces a 5 mV hyperpolarization of the resting membrane potential of the fibres. The effects of CNTF are similar to those of 1-oleoyl-2-acetyl-sn-glycerol (OAG), a protein kinase C (PKC) activator, when no effect is observed in the presence of chelerythrine, a PKC inhibitor. These results suggest that, in skeletal muscle, CNTF can rapidly decrease sodium currents by altering inactivation gating, probably through an intracellular PKC-dependent mechanism that could lead to decreased membrane excitability. The present study contributes to a better understanding of the physiological role of endogenous CNTF.

Voltage-gated sodium channels, responsible for the generation and conduction of action potentials in excitable tissues such as skeletal muscle, mediate membrane excitability (Hille, 1984). These channels, closed at the resting potential, rapidly open in response to membrane depolarization and then inactivate through two processes. Fast inactivation occurs rapidly after activation (within a few milliseconds) limiting the duration of the action potential and initiating the rapid repolarization of muscle fibre (Hodgkin & Huxley, 1952). Slow inactivation takes effect in response to more prolonged depolarization, limiting the availability of sodium channels on a time scale of seconds to minutes (Ruff et al. 1988). Modifications of sodium channel gating properties can modify sarcolemma excitability and, as a consequence, can alter the contractile properties of skeletal muscle. In particular, defects in fast and/or slow inactivation processes, as observed in inherited skeletal muscle disorders resulting from mutations on the gene encoding for the skeletal muscle sodium channel α-subunit (hyperkalaemic periodic paralysis, paramyotonia congenita, and potassium-aggravated myotonia), lead to sarcolemma hyperexcitability, and then to a muscular weakness (for review, see Cannon, 2000).

Reduced performance and weakness of skeletal muscle are also observed during ageing, after denervation and following prolonged muscle disuse. In parallel with dramatic macroscopic alterations of muscle structure such as the loss of muscle mass, changes in the expression and/or properties of key skeletal muscle protein systems such as sodium channels have also been observed. In particular, during ageing, the number of available sodium channels in skeletal muscle generally increases resulting in enhanced sodium currents (Desaphy et al. 1998). Otherwise, denervated muscle fibres, in addition to the adult isoform of the sodium channel (Na_v1.4 or SkM1) normally present in skeletal muscle, express a juvenile form, namely Na_v1.5 (or SkM2), which is relatively resistant to tetrodotoxin and in which the shallower voltage dependence of slow inactivation could account for fibre hyperexcitability (Pappone, 1980; Kallen et al. 1990; Richmond et al. 1998). Comparatively, the prolonged hindlimb unloading that mainly alters the function of antigravity muscles leading to a slow- to fast-twitch phenotype transition induces an increase in sodium channel density mainly due to a higher expression level of Na_v1.4 (SkM1) isoform (Desaphy et al. 2001).

Among the different therapy strategies developed to limit muscular atrophy and to accelerate muscle performance recovery (for review, see Thompson, 2002), the use of neurotrophic factors was shown to be relevant. Recently, we have demonstrated that subcutaneous administration of ciliary neurotrophic factor (CNTF), a cytokine belonging to the interleukin-6 family, in hindlimb muscles placed in microgravity conditions reduces muscle atrophy and functional alterations observed in slow-twitch fibres (Fraysse et al. 2000). Such long-term potent myotrophic effects of this cytokine were also shown in atrophied skeletal muscles by denervation and during ageing (Helgren et al. 1994; Guillet et al. 1999). CNTF, primarily well known for its ability to sustain the survival of motor neurones in vitro and in vivo (Sendtner et al. 1992b; Sleeman et al. 2000), recognizes a multimeric receptor, composed of two transmembrane signal-transducing proteins, glycoprotein-130 (gp-130) and leukaemia inhibitory factor receptor (LIF-R) and a specific binding subunit known as CNTF receptor alpha (CNTFR-α) (Davis et al. 1991; Ip et al. 1993; Inoue et al. 1996). Compared to LIF-R and gp-130, which are widely expressed throughout the body, the expression of CNTFRα is prevalent in the nervous system and in skeletal muscles (Davis et al. 1991; Helgren et al. 1994; MacLennan et al. 1996). It is generally admitted that CNTF acts through the activation of the JAK/STAT signal transduction pathway, involved in the regulation of gene transcription (Heinrich et al. 1998). The biological activity of CNTF mediated by this cascade of events is a long-lasting process that can take several hours, then accounting for long-term myotrophic effects of cytokine. Nevertheless, other studies have shown that in vitro application of CNTF resulted in an immediate potentiation of transmitter release at developing neuromuscular synapses in Xenopus cell cultures (Stoop & Poo, 1995) and provoked a rapid inhibition of the glutamate-induced increase in [Ca²⁺]_i of hippocampal neurones (Yan et al. 2000). Recently, it was reported that in cortical neurones CNTF rapidly inhibits voltage-activated calcium channels, probably through channel phosphorylation that could involve multiple kinases (PKC, protein kinase B (PKB), mitogen-activated protein kinase (MAPK)) (Holm et al. 2002). Based on these data and the observation of a muscular weakness in CNTF knock-out mice (−/−) (Masu et al. 1993), one could hypothesize that some channel activities involved in muscle fibre contraction, particularly sodium channel activity, can be modulated in the short-term by CNTF.

In the present study we demonstrate that the in vitro application of CNTF induces a dose-dependent inhibition of sodium currents recorded in native dissociated skeletal muscle fibres. The CNTF action is rapid (< 10 min) and is blocked in the presence of PKC inhibitors. The study outlines a novel action mechanism of CNTF in skeletal muscle involving a PKC-dependent signalling pathway and sodium channels mediating membrane excitability.

Methods

Muscle isolation and enzymatic dissociation of muscle fibres

All experiments were authorized by a departmental agreement (no. A29-019-3) and were carried out in accordance with the recommendations of our ethical regional committee and of the European Community (no. 86/609). Male Wistar rats (body weight 250–300 g, age 2–3 months) were killed by stunning and cervical dislocation. Fast-twitch flexor digitorum brevis (FDB) muscle was rapidly excised from the rats and placed in Hepes-buffered physiological solution supplemented with 3.0 mg ml⁻¹ collagenase (type II; Gibco-BRL). The FDB fibres were then incubated for 3–4 h at 37°C for enzymatic dissociation. At the end of this period, dissociated fibres were sampled and rinsed several times with the bath recording solution before being transferred to a 35 mm Petri dish for patch-clamp recordings.

Patch voltage-clamp apparatus

Sodium currents were recorded at room temperature (22 ± 2°C) in the cell-attached configuration of the patch-clamp method (Hamill et al. 1981) with a GeneClamp 500B amplifier and a CV-5-100U headstage (Axon Instruments, Foster City, CA, USA). Pipettes were formed and polished from GC150TF-10 glass (Harvard Apparatus, Ltd, USA) with a DMZ-Universal puller (Zeitz Instruments, Germany). Pipettes had resistances averaging 2 MΩ when filled with the recording pipette solution. Voltage-clamp protocols and data acquisition were performed with WinWCP V3.2.5 (Whole Cell Program, Strathclyde University) through a 12 bit A–D/D–A interface (CED 1401+; Cambridge Electronic Design Ltd, Science Park, Cambridge, UK). Currents were low-pass filtered at 5 kHz and digitized at 35 kHz.

Since sodium channel density is 5- to 10-fold higher on the end-plate border than away from the end-plate (Ruff, 1992), sodium currents were recorded from extra-junctional membrane at a site > 100 μm from the end-plate. This could be visualized with phase contrast under an inverted microscope (Olympus IX 70) and with a Progressive Scan digital camera (XC8500CE, Donpisha) and only fibres with visible end-plates were patched. In order to inhibit potassium currents and to depolarize the sarcolemma, fibres were placed in the bath recording solution containing Cs⁺ ions as the main cation (145 mm). Capacitance currents were almost totally cancelled by the compensation circuit of the amplifier. To further eliminate residual capacitance transient and leak currents, we used the P/4 subtraction procedure (Almers et al. 1983). Briefly, four negative pulses with amplitudes one tenth of the pulse test amplitude are applied to the patch before the test pulse, thus allowing the determination and then the subtraction of the residual leak current.

Pulse protocols

Sodium currents were elicited by depolarizing pulses from the holding potential up to −10 mV, applied at a frequency of 0.3 Hz. The holding potential was usually −100 mV, a value close to physiological values of intact skeletal muscle fibres and at which most of the channels are in a closed state. At this holding potential, direct transitions from closed to inactivated state could occur (Bean et al. 1983; Takahashi & Cannon, 2001) leading a few sodium channels to be non-conductive. Then to evaluate the possibility of a state-dependent CNTF effect we performed some experiments at a holding potential of −130 mV, a value at which the number of channels in an inactivated closed-state is negligible. Recordings were initiated at least 5–10 min after the giga-seal formation when current amplitude reached a steady level. The patches that showed a rundown in the peak current amplitude up to 15% during the first 10–20 min of recordings before CNTF application were discarded from our analysis. The current–voltage relationship was measured by applying to the patch membrane a cycle of 20 ms test pulses from the holding potential of −100 mV to increasing potentials (from −60 to +130 mV in 10 mV increments). The intervals between each test pulse were long enough (3 s) to allow the complete recovery of sodium channels from inactivation. This protocol was repeated 2 or 3 times for each patch to check for sodium current stability. If stable, peak current amplitudes were averaged to obtain a mean value, the patches with non-reliable peak current amplitudes being discarded. The rise and decay time constants of sodium currents, τ_m and τ_h, respectively, were calculated from the fit of the current rise and decay with an exponential function of the following form:

The activation curve was constructed from the current–voltage relationship by converting current to conductance: the current amplitude for each tested potential between −60 and 0 mV was divided by the driving force (V − V_Na) where V is the potential applied to the patch membrane and V_Na is the equilibrium electrochemical potential for sodium ions, estimated to be approximately +70 mV. Activation curves were fitted by the Boltzmann equation G/G_max = 1/{1 + exp [(V − V_a,1/2)/k_a]}, where G_max is the maximal conductance, k_a is the slope factor and V_a,1/2 is the potential at which one-half of the channels are activated.

Sodium channels are known to undergo at least two distinct inactivation processes with different kinetics and voltage dependencies (Vendanthan & Cannon, 1998). The fast inactivation closes channels in a time scale of a few milliseconds, whereas the slow inactivation process has a time constant in the order of seconds. For the study of the steady-state fast inactivation, a first pulse from −140 to +10 mV and of 50 ms duration was applied, followed by a second test pulse of 20 ms in duration at −10 mV. The voltage-clamp protocol used for the study of slow inactivation was almost the same except that the first pulse duration was 2000 ms and there was a recovery time of 50 ms between the first and second pulses. These protocols were repeated 2 or 3 times for each patch and then peak current amplitudes were averaged to obtain the mean value. The steady-state fast inactivation relationships were fitted with the Boltzmann equation I_Na = I_Na,max/{1 + exp[(V − V_h,1/2)/k_h]}, where I_Na is current, I_Na,max is the maximal current, k_h is the slope factor, and V_h,1/2 is the potential for having one-half of the channels inactivated. To describe the voltage dependence of slow inactivation, a non-zero residual current, I_Na,min, was included in the Boltzmann equation:

Measurement of resting membrane potential

The intrinsic resting membrane potential (RMP) of FDB muscle fibres was measured with a single intracellular microelectrode formed from a GC150-F10 glass pipette and filled with a 3 m KCl solution. The intracellular microelectrode resistance averaged 20 MΩ. The RMP values recorded in FDB were −78.5 ± 9.3 mV (n = 71) and decreased to a value close to −10 mV (RMP = −13.7 ± 2.3 mV, n = 40) after 1–2 h of incubation in the CsCl bath recording solution. Thus, the values of potential given here are those held by the patch-clamp amplifier and are not corrected for RMP.

CNTF production and bioactivity

As previously described (Guillet et al. 1999), the rat recombinant ciliary neurotrophic factor (CNTF) used in the present experiments was produced as a glutathione S-transferase-fusion protein using the pGEX-4T2 gene fusion vector from Pharmacia (Uppsala, Sweden), further purified by the GST Purification Module (Pharmacia) and finally stored at −80°C until use. Before each experiment, the CNTF produced in our laboratory was diluted in Ringer physiological solution.

In order to test the biological activity of CNTF, a proliferation assay was performed in vitro on a Ba/F3 cell line stably transfected to express the functional receptor subunits for CNTF (i.e. CNTFR-α, gp-130 and LIF-R) and known to be responsive to CNTF (Kallen et al. 1999). Human recombinant CNTF (hCNTF) and LIF (both products from R&D Systems, Oxon, UK) were used as positive controls when interleukin-2 (IL-2) (from R&D Systems) was used as a negative control. Briefly, BAF GLC cells were grown in RPMI 1640 medium supplemented with 10% fetal calf serum (FCS) and 1 ng ml⁻¹ rat CNTF, and were then seeded in 96-well plates at a concentration of 5 × 10³ cells per well in RPMI 1640 medium containing 5% FCS. Serial dilutions of the different cytokines tested were performed in triplicate. After a 72 h incubation period, 0.5 μCi of [³H]thymidine (³HTdr) was added to each well for the last 4 h of culture and the incorporated radioactivity determined by scintillation counting. The ‘homemade’ CNTF as well as positive controls, hCNTF and LIF, induced a robust proliferation of the cells: the concentration giving a half-maximal effect (EC₅₀) for CNTF was 1.0 ng ml⁻¹ versus 0.6 and 5.5 ng ml⁻¹ for hCNTF and LIF. As expected, no proliferation was observed in the presence of the negative control, IL-2.

Solutions and chemicals

The Ringer physiological solution had the following composition (mm): NaCl, 140; KCl, 5; CaCl₂, 2; MgCl₂, 1; and Hepes, 30. The bath recording solution contained (mm): CsCl, 145; EGTA, 5; MgCl₂, 1; and Hepes, 10. The pipette solution had the same composition as the Ringer solution given above. All chemical compounds were from Merck (Darmstadt, Germany), except Hepes (Sigma, St Louis, MO, USA). The pH was adjusted to 7.4 with trihydroxyaminomethane (Tris) (Sigma).

Commercial recombinant rat CNTF was obtained from R&D Systems (rCNTF_R&D) and was stored as a 10 μg ml⁻¹ stock solution at −80°C. OAG (1-oleoyl-2-acetyl-sn-glycerol) was stored as a 1.25 mm stock solution with DMSO (10%). Bis-indolylmaleimide I was stored as a 55 mm stock solution at −20°C. Chelerythrine was maintained as a 25 mm stock solution with DMSO (0.1%) at −20°C. All chemicals except for rCNTF_R&D were purchased from Sigma. To maintain the characteristics of the seal, different drugs were applied away from the patched cell. This procedure could have induced a delay in observing the initiation of an effect that should to be taken into account when analysing the results.

Data analysis and statistics

Microcal Origin software (Microcal Software, Northampton, MA, USA) was used to analyse experimental data and to perform the curve fittings. Although a single fit to the data averaged from n patches is presented in the figures, fits were performed from each individual patch to obtain mean values and standard error of the mean (s.e.m.) for statistical comparison of the fit parameters between groups, in the absence and in the presence of drugs. Statistical differences were determined by performing Student's t test or a non-parametric Mann-Whitney U test as appropriate. Differences were considered to be significant when P < 0.05.

Results

Effects of CNTF on sodium currents

As illustrated in Fig. 1A, patch membrane depolarizations applied from a holding potential of −100 mV to a test potential of −10 mV enabled the recording of macroscopic-current-like sodium current (I_Na) with rapid onset and total inactivation in < 3 ms. In control conditions, the amplitude of the sodium current (I_Na) peaked at around −500 pA (Table 1), which suggested the presence of about a hundred open sodium channels among a thousand channels in the cell-attached patches, assuming that there are about 10% of channels open at the current peak. The 10 min application of CNTF (1 and 2 ng ml⁻¹) decreased the peak amplitude in a dose-dependent manner, the protein potency being around three times higher at 2 than at 1 ng ml⁻¹ (Fig. 1A and Table 1). The concentration–response relationship for the I_Na reduction produced by a 10 min application of CNTF (Fig. 1B) revealed that the I_Na decrease was significantly detectable at 0.1 ng ml⁻¹ CNTF (3.8 ± 0.9%; n = 5 patches, P < 0.05). The CNTF effect was more pronounced with higher protein concentrations (8.5 ± 0.8% at 1 ng ml⁻¹ CNTF; n = 6 patches, P < 0.05), to reach a maximal level of I_Na reduction at 2 ng ml⁻¹ (25.1 ± 4.1%; P < 0.05, n = 10 patches) (Fig. 1B). Increasing CNTF concentration to 5, 10 or 20 ng ml⁻¹ failed to reduce I_Na amplitude any further (Fig. 1B). The fit of the experimental points of the concentration–response curve allowed the estimation of the CNTF concentration for half-maximal reduction of I_Na (IC₅₀) at 1.1 ng ml⁻¹. CNTF application induced no significant change in the activation and inactivation I_Na kinetics, as determined by the current rise (τ_m) and decay (τ_h) exponential time constants, respectively (Table 1), at any concentration tested.

Figure 1. Dose- and time-dependent CNTF effects on macroscopic sodium currents in cell-attached patches of rat skeletal muscle fibres.

A, traces of macroscopic-current-like sodium current (I_Na) were obtained in two cell-attached patches by depolarizing the membrane patch for 20 ms to −10 mV from the holding potential of −100 mV before and after the 10 min application of 1 and 2 ng ml⁻¹ CNTF. The peak current before and after CNTF application was −0.576 nA and −0.520 nA, respectively, for 1 ng ml⁻¹ CNTF, and −0.618 nA and −0.452 nA, respectively, for 2 ng ml⁻¹ CNTF. B, concentration–response curve for the CNTF effect on sodium current. The amount of peak I_Na amplitude decrease was calculated in conditions of single depolarizing 20 ms pulses from −100 mV to −10 mV with different CNTF concentrations. Each point shows the percentage reduction of I_Na observed in the presence of each CNTF concentration versus I_Na in the absence of the cytokine in the same fibre and is expressed as mean ± s.e.m. for 5–11 fibres. One concentration only was tested in a cell-attached patch. C, time course for development of CNTF effects on sodium current. The amount of peak I_Na amplitude decrease induced by CNTF was plotted against variable cytokine incubation periods. Sodium currents were elicited by depolarizing the membrane patch from −100 to −10 mV in the absence and in the presence of 2 ng ml⁻¹ CNTF. Each point is expressed as mean ± s.e.m. for 5–7 fibres. The curve fit of the experimental points in the concentration– and time–response curves was obtained using a sigmoid function.

Table 1.

Ciliary neurotrophic factor (CNTF) effects on peak amplitude and kinetic parameters of sodium currents in flexor digitorum brevis (FDB) muscle fibres

	Peak amplitude (nA)	τm(ms)	τh(ms)	n
Control	−0.555 ± 0.013	0.113 ± 0.017	0.283 ± 0.029	6
CNTF (1 ng ml−1)	−0.501 ± 0.021*	0.101 ± 0.007	0.267 ± 0.022	6
Control	−0.554 ± 0.083	0.106 ± 0.004	0.259 ± 0.011	10
CNTF (2 ng ml−1)	−0.416 ± 0.055*	0.095 ± 0.005	0.235 ± 0.008	10
Washout	−0.534 ± 0.109	0.103 ± 0.021	0.246 ± 0.064	3

Peak amplitude of sodium current was measured by depolarizing the cell-attached patch membrane from a holding potential of −100 mV to a test potential of −10 mV. τ_m and τ_h are the rise and decay time constants, respectively, and were calculated from the fit of the current rise and decay with an exponential function as described in Methods. Data are expressed as mean ± s.e.m. n is the number of patches. The CNTF peptide (1 and 2 ng ml⁻¹) was applied for at least 10 min. The application of 2 ng ml⁻¹ CNTF was followed in 3 of 10 patches by a 10 min washout in recording solution without peptide.

^**Statistical difference from control value with P < 0.05.

Figure 1C illustrates the time course of the CNTF effect on the I_Na peak amplitude when applied at a concentration of 2 ng ml⁻¹ for 2–25 min. The CNTF-induced I_Na reduction developed quite rapidly, occurring as early as 2 min of incubation. The CNTF-induced I_Na decrease was already detectable at 5 min (6.0 ± 0.1%; n = 6 patches, P < 0.05), and reached a maximal level at 10 min (20.2 ± 5.2%; n = 7, P < 0.05) and remained steady up to 25 min of incubation (22.4 ± 3.0%; n = 5, P < 0.05). After 30 min incubation in the absence of CNTF some patches displayed a slight rundown corresponding to 10–20% of I_Na reduction (data not shown) leading to a more difficult analysis of the results.

It is noteworthy that the CNTF effect was not significantly modified when tested on sodium currents elicited from a hyperpolarized holding potential (−130 mV). In particular, the peak amplitude of sodium current elicited by a depolarizing membrane patch from −130 to −10 mV (I_Na,max = −0.544 ± 0.012 nA, n = 6) was reduced by 23.7 ± 4.5% in the presence of 2 ng ml⁻¹ of CNTF (I_Na,max = −0.423 ± 0.011 nA, n = 6, P < 0.05).

Compared to the CNTF we produced ourselves, a commercial version of CNTF (recombinant rat CNTF from R&D Systems, rCNTF_R&D) produced similar effects when applied at the same concentration. In particular, the 10 min application of 2 ng ml⁻¹ rCNTF_R&D caused a decrease in I_Na peak amplitude by 22.5 ± 7.8% (n = 6 patches, P < 0.05) (Fig. 2A). Furthermore, the CNTF action on sodium channels is reversible: a 10 min washout eliminated the CNTF effect induced by a 10 min incubation (Fig. 2B and Table 1).

Figure 2. Reversibility of CNTF effect and comparison with that of the commercial peptide version on macroscopic sodium currents in cell-attached patches of rat skeletal muscle fibres.

Currents were recorded by depolarizing the membrane patch for 20 ms to −10 mV from the holding potential of −100 mV before and after a 10 min application of CNTF peptide provided by R&D (A) and the version produced in our laboratory as described in Methods. ‘Homemade’ CNTF incubation was followed by a 10 min washout period (B).

We then investigated the hypothesis that the peak I_Na amplitude reduction observed in the presence of the cytokine could result from changes in the voltage dependence of sodium channels.

Effects of CNTF on voltage dependence of sodium channels

Sodium currents were recorded by depolarizing the membrane patch from −60 to +130 mV in 10 mV increments (Fig. 3A) allowing the construction of current–voltage (I–V) relationships (Fig. 3B). In control conditions, the I_Na amplitude peaked at −10 mV and reversed at a potential of around +80 mV (Fig. 3B), a value very close to the equilibrium electrochemical potential for sodium ions, estimated to be approximately +70 mV. At each tested membrane potential, the 10 min application of 2 ng ml⁻¹ CNTF caused an I_Na amplitude decrease but failed to affect the potential of maximal peak current and the reversal potential (Fig. 3A and B). By contrast, in the presence of the cytokine, the conductance estimated as the linear slope from the current–voltage curves was reduced in a dose-dependent manner and to an extent similar to that of the maximal I_Na amplitude decrease (9.9 ± 2.3%, n = 6, P < 0.05, and 25.2 ± 4.5%, n = 10, P < 0.05, with 1 and 2 ng ml⁻¹ of CNTF, respectively) (Table 2). The activation curves constructed from I–V relationships in the presence and in the absence of CNTF are superimposed (Fig. 3C) indicating that the voltage dependence of sodium channel activation is not modified by CNTF application. Indeed, the half-maximal activation potential is unchanged (Table 2) while the maximal conductance is decreased by CNTF (Fig. 3C, dotted line).

Figure 3. CNTF effects on the current–voltage relationship and the activation curve of sodium currents in cell-attached patches of rat skeletal muscle fibres.

A, macroscopic-current-like sodium currents were recorded in the same cell-attached patch before (left traces) and after 10 min application of 2 ng ml⁻¹ CNTF (right traces) by applying a cyclic test-pulse protocol shown in inset. B, current–voltage relationships were constructed before (control) and after the 10 min application of 2 ng ml⁻¹ CNTF from protocol shown in A. C, activation curves were constructed from current–voltage relationships by converting current obtained between −60 and 0 mV to conductance, which was then normalized. Data were fitted to a Boltzmann equation as described in Methods.

Table 2.

CNTF effects on conductance and voltage dependence of activation and steady-state fast inactivation parameters of sodium channels in FDB muscle fibres

		Activation		Steady-state fast inactivation
	Conductance (nS)	Va,1/2 (mV)	ka (mV)	Vh,1/2(mV)	kh(mV)	INa,max(nA)	n
Control	−6.0 ± 0.5	−31.1 ± 2.9	5.6 ± 0.4	−73.3 ± 4.0	−6.2 ± 0.4	−0.565 ± 0.023	6
CNTF (1 ng ml−1)	−5.4 ± 0.5*	−31.6 ± 3.9	5.9 ± 1.0	−77.7 ± 3.6*	−6.4 ± 0.5	−0.529 ± 0.017*	6
Control	−6.8 ± 1.0	−30.2 ± 1.4	6.4 ± 0.6	−76.8 ± 2.4	−7.1 ± 0.3	−0.529 ± 0.083	10
CNTF (2 ng ml−1)	−5.0 ± 0.7*	−31.6 ± 2.4	6.6 ± 0.4	−82.5 ± 2.3*	−7.5 ± 0.3	−0.413 ± 0.075*	10

All parameters were obtained from individual patches and are expressed as mean ± s.e.m. n is the number of patches. Conductance is estimated as the linear slope from current–voltage curves between 0 and +40 mV. Parameters of activation and steady-state fast inactivation are derived from the fit with the Boltzmann equation of normalized current–voltage and steady-state fast inactivation curves, respectively. V_a,1/2: half-maximal activation potential as derived from normalized current–voltage curves between −60 and 0 mV; k_a: activation slope factor; V_h,1/2: half-maximal steady-state fast inactivation potential; k_h: fast inactivation slope factor; I_Na,max: maximal sodium current.

^**Significant difference from control value with P < 0.05.

The CNTF-induced I_Na inhibition could result from a shift in the voltage dependence of sodium channel inactivation to more negative potentials. Sodium channels are known to undergo at least two distinct inactivation processes, a fast and a slow inactivation process, which occur within a few milliseconds and a few seconds, respectively. The former is mainly involved in the current shape and the I_Na,max while the latter regulates the fraction of activable channels, and hence the channel availability (Ruff, 1996a). Although both of these inactivation processes are not exclusive, i.e. channels may be fast and slow inactivated simultaneously, the fast inactivation gate is independent of the state of the slow inactivation gate (Vendanthan & Cannon, 1998). Therefore, we firstly investigated the effect of CNTF on the steady-state fast inactivation by using a double pulse protocol (inset in Fig. 4A). It was previously shown that during long-duration cell-attached patch recordings, a negative shift in the voltage dependence of sodium channel gating could occasionally occur (Desaphy et al. 1998). Therefore, to avoid such a non-specific effect, we took care to measure fast inactivation in the absence and in the presence of CNTF within 30 min following the giga-seal formation. In control conditions, patches showed steady-state fast inactivation curves with half-maximal fast inactivation potentials (V_h,1/2) around −75 mV and with slopes around −6 and −7 mV (Table 2). By comparison, the 10 min application of CNTF at 2 ng ml⁻¹ decreased the maximal I_Na amplitude by 22.5 ± 6.5% (n = 10, P < 0.05) (dotted line in Fig. 4A and Table 2), shifted the steady-state fast inactivation curves to more negative potentials (the shift in half-maximal fast inactivation potential being ΔV_h,1/2 = −5.7 ± 1.4 mV, n = 10 patches, P < 0.05) and slightly increased the slope factor (Fig. 4A, Table 2). In the present experimental conditions, the CNTF-induced shifts in half-maximal fast inactivation potential (ΔV_h,1/2) were more or less pronounced, ranging from −2 to −15 mV, perhaps because of the variability in the control V_h,1/2, ranging from −65 to −90 mV. To test such a possibility we report the shifts in V_h,1/2 as a function of the control values of V_h,1/2. The results illustrated in Fig. 4B show that the data points are correlated by a linear regression (r = −0.684, P < 0.05; n = 10 patches). The variability in the control V_h,1/2 values could also explain the more or less important reductions in the maximal I_Na amplitude for a given concentration of CNTF, ranging from −8 to −45% at 2 ng ml⁻¹ (n = 10). Yet Fig. 4B shows that the CNTF-induced shift in the I_Na,max is also linearly correlated with the control V_h,1/2 values (r = −0.637, P < 0.05; n = 10 patches). The CNTF effects on the voltage dependence of sodium channel fast inactivation were less pronounced with 1 ng ml⁻¹ CNTF, the decrease in maximal I_Na amplitude and in half-maximal fast inactivation potential (ΔV_h,1/2) being −6.3 ± 0.8% (n = 6; P < 0.05) and −4.4 ± 0.9 mV (n = 6, P < 0.05), respectively (Table 2).

Figure 4. CNTF effects on steady-state fast inactivation of sodium currents in cell-attached patches of rat skeletal muscle fibres.

A, normalized steady-state fast inactivation curves obtained from 10 patches were constructed before (□) and after the 10 min application of 2 ng ml⁻¹ CNTF (▪) from a double pulse protocol as shown in inset. The protocol was repeated 2 or 3 times in each patch; peak sodium current amplitudes were measured during the 20 ms test pulse at −10 mV and averaged. These values were then plotted as a function of the potential held for 50 ms and fitted with the Boltzmann equation as described in Methods. The values were then normalized with respect to the maximal amplitude (I_Na,max) obtained from the fit and finally averaged from 10 patches to obtain the mean and s.e.m. The dotted line represents the fit of I_Na values obtained in the presence of CNTF and normalized to the I_Na,max value obtained in control conditions. B, shifts of half-maximal steady-state fast inactivation voltage (ΔV_h,1/2) (left axis) and of maximal I_Na amplitude (ΔI_Na,max) (right axis) were reported as functions of values of half-maximal steady-state inactivation voltage (control V_h,1/2) measured in the same cell before cytokine application.

Sodium channel slow inactivation has also been evaluated between 20 and 30 min after the giga-seal formation, even if no shift in slow inactivation voltage dependence was reported during long-lasting recordings (O'Reilly et al. 1999). Generally, the steady state of slow inactivation is reached when conditioning voltage pulses last more than 10 s (Struyk et al. 2000). Unfortunately, owing to the limited lifetime of cell-attached patches of native fibres, it was difficult in our experimental conditions to use such long conditioning voltage pulses. Therefore, as in previous experiments (Desaphy et al. 2001), we studied slow inactivation at a time point that was different from steady state by using the double pulse protocol with a 2000 ms conditioning voltage pulse followed by a 50 ms fast inactivation recovery at −140 mV (inset in Fig. 5). In control conditions, patches showed slow inactivation curves with half-maximal inactivation potentials (V_s,1/2 = −69.6 ± 2.7 mV, n = 5), slopes (k_s = −12.3 ± 1.0 mV, n = 5) and residual Na⁺ current amplitudes (I_min/I_max = 0.223 ± 0.028, n = 5) that were very similar to those previously reported for other types of skeletal muscle fibres (Desaphy et al. 2001). As shown in Fig. 5, the 10 min application of CNTF (2 ng ml⁻¹) decreased the I_Na,max by 45.7 ± 13.1% (n = 5, P < 0.05) (dotted line) and shifted the slow inactivation curves towards hyperpolarizing potentials (ΔV_s,1/2 = −9.3 ± 2.2 mV; n = 5, P < 0.05) without a significant change in slope or residual Na⁺ current with respect to control conditions. The CNTF-induced shift in the half-maximal slow inactivation potentials (ΔV_s,1/2) was around twofold greater than that observed in steady state fast inactivation curves.

Figure 5. CNTF effect on sodium current slow inactivation curves in cell-attached patches of rat skeletal muscle fibres.

Normalized slow inactivation curves were constructed before (○) and after the 10 min application of 2 ng ml⁻¹ CNTF (•) from a double pulse protocol as shown in inset. Peak sodium current amplitudes were measured during the 20 ms test pulse at −10 mV, normalized to the maximal I_Na amplitude (I_Na,max) as described in Fig. 4, and then plotted as a function of the potential held for the 2 s and fitted with the Boltzmann equation as described in Methods. Each point is expressed as mean ± s.e.m. for 5 patches. The dotted line represents the fit of I_Na values obtained in the presence of CNTF and normalized to the I_Na,max obtained in control conditions.

Involvement of a PKC-dependent mechanism in the effects of CNTF

Taking into account that a 5 min CNTF treatment has been reported to activate PKC in cultured sympathetic neurones (Kalberg et al. 1993) and that in skeletal muscle, the activity of sodium channels is modulated by PKC (Bendahhou et al. 1995), we investigated the possibility of PKC involvement in the CNTF effects on sodium currents by evaluating the effects of a PKC activator (OAG) or inhibitor (chelerythrine) on sodium currents.

As illustrated in Fig. 6A and in Table 3, a 10 min application of 40 μm OAG resulted in a marked decrease in I_Na peak amplitude: −39.1 ± 10.1% (n = 6 patches, P < 0.05) with respect to controls values. As for CNTF, the PKC activator did not alter the activation or inactivation kinetics, or the voltage dependence, of sodium current activation (Table 3). In contrast, the OAG treatment shifted the midpoint of steady-state inactivation by −9.4 ± 3.4 mV (n = 6 patches, P < 0.05) and decreased the I_Na,max by around 30% (Fig. 6C, dotted line and Table 3).

Figure 6. Effects of an activator and an inhibitor of PKC in the presence or absence of CNTF on sodium currents in cell-attached patches of rat skeletal muscle fibres.

A and B, traces of macroscopic-current-like sodium current (I_Na) were obtained by depolarizing the membrane patch for 20 ms to −10 mV from the holding potential of −100 mV before and after 10 min incubation with 40 μm OAG (1-oleoyl-2-acetyl-sn-glycerol) (A) or with 1 μm chelerythrine following by 10 min application of CNTF (2 ng ml⁻¹) (B). C and D, normalized steady-state fast inactivation curves in the presence of OAG (40 μm) (C) or 1 μm chelerythrine following by a 10 min application of CNTF (2 ng ml⁻¹) (D). Inactivation curves were constructed before (○; ▵) and after the 10 min application of drugs (•; ▴) from a double pulse protocol as shown in inset of Fig. 4. The dotted line represents the fit of I_Na values obtained in the presence of OAG and normalized to the I_Na,max obtained in control conditions.

Table 3.

Effects of an activator and of an inhibitor of PKC in the presence or absence of CNTF on the maximal peak amplitude, kinetic parameters and voltage dependence of activation and steady-state fast inactivation parameters of sodium currents in FDB muscle fibres

	Peak INa amplitude (nA)	τm (ms)	τh (ms)	Va,1/2 (mV)	Vh,1/2 (mV)	n
Control	−0.449 ± 0.062	0.101 ± 0.008	0.281 ± 0.015	−30.1 ± 3.4	−75.9 ± 7.6	6
OAG (40 μm)	−0.289 ± 0.083*	0.100 ± 0.003	0.283 ± 0.019	−31.9 ± 4.5	−85.4 ± 9.0*	6
Control	−0.346 ± 0.167	0.105 ± 0.009	0.215 ± 0.030	−38.5 ± 4.4	−88.7 ± 3.1	5
Chelerythine (1 μm)	−0.316 ± 0.156	0.103 ± 0.013	0.213 ± 0.012	−40.5 ± 2.3	n.d.	5
Chelerythine (1 μm) +
CNTF (2 ng ml−1)	−0.342 ± 0.174	0.115 ± 0.027	0.223 ± 0.029	−39.4 ± 4.3	−86.9 ± 2.8	5

All parameters are obtained in individual patches and are expressed as mean ± s.e.m. n is the number of patches. Peak I_Na amplitude, τ_m and τ_h of sodium current were determined as described in Table 1. V_a,1/2: the half-maximal activation potential, as derived from the fit with the Boltzmann equation of normalized current–voltage curves between −60 and 0 mV. V_h,1/2: the half-maximal steady-state fast inactivation potential, as derived from the fit with the Boltzmann equation of normalized steady-state fast inactivation curves. OAG (1-oleoyl-2-acetyl-sn-glycerol) was applied for 10 min. Chelerythrine was applied for 10 min before the 10 min application of CNTF (2 ng ml⁻¹).

^**Significant difference from control value with P < 0.05. n.d., not determined.

When FDB fibres were pre-incubated for 10 min in a solution containing 1 μm chelerythrine, the subsequent 10 min application of 2 ng ml⁻¹ CNTF failed to induce an inhibition of sodium channels (Fig. 6B). Furthermore, in the presence of PKC inhibitor, CNTF failed to significantly shift activation and steady-state fast inactivation curves (Fig. 6D and Table 3). Chelerythrine applied in the presence or absence of CNTF induced no significant change in I_Na kinetics (Table 3). Altogether, these results indicate that CNTF acts on sodium channels probably through a PKC-dependent mechanism.

Effects of CNTF on the resting membrane potential

In our experimental conditions, CNTF shifts the voltage dependence of sodium channel inactivation processes in a hyperpolarizing direction. Fast inactivation helps to terminate the action potential whereas slow inactivation regulates the availability of excitable sodium channels as a function of the membrane potential by changing the channel distribution between the closed-excitable and slow-inactivated states (Ruff, 1996a). Therefore, we investigated whether CNTF by acting on sodium channels modulates the resting potential in skeletal muscle fibres.

In our experiments the resting membrane potential (RMP) measured in FDB fibres was −73.3 ± 0.7 mV (n = 150), a value close to those recorded in other types of fast-twitch fibres (Wood & Slater, 1995). As illustrated in Fig. 7, with respect to the control RMP value, a 5 mV hyperpolarization (−78.8 ± 0.9 mV, n = 66, P < 0.05) was observed after a 10 min application of CNTF (2 ng ml⁻¹) in FDB fibres. In the presence of PKC inhibitors, CNTF failed to hyperpolarize FDB fibres and even caused a small fibre depolarization with respect to control conditions (Fig. 7), the RMP values being −65.3 ± 0.6 mV (n = 43, P < 0.05) with 1 μm chelerythrine and −65.0 ± 1.0 mV (n = 23; P < 0.05) with 0.3 μm bis-indolylmaleimide I, another PKC inhibitor (Toullec et al. 1991). It is noteworthy that in the absence of CNTF, the RMP values recorded with chelerythrine (−69.8 ± 3.5 mV; n = 6) or with bis-indolylmaleimide I (−73.5 ± 3.5 mV; n = 5) are similar to those obtained in control conditions.

Figure 7. CNTF effects on resting membrane potential in rat skeletal muscle fibres.

The intrinsic resting membrane potential of FDB muscle fibres was measured with a 3 m KCl single intracellular microelectrode in control conditions (open bar), after the 10 min application of CNTF (2 ng ml⁻¹) in the absence (filled bar) or in the presence of 1 μm chelerythrine (striped bar) or 0.3 μm bis-indolylmaleimide I (cross-hatched bar). *Statistical difference from control value with P < 0.05.

Discussion

This is the first study carried out to evaluate the short-term effects of CNTF application on the activity of voltage-gated sodium channels in native skeletal muscle fibres. The main findings of this study are that macroscopic sodium currents measured in cell-attached patches are rapidly reduced by 30% after a 10 min application of CNTF. Under these conditions, the attenuation of sodium currents by CNTF is dose dependent and occurs without significant modification of activation and inactivation rates. Furthermore, CNTF alters the voltage dependence of channel inactivation processes by shifting both slow and steady-state fast inactivation curves to more negative potentials. The CNTF effects could be paralleled with those of another cytokine, IL-2 (interleukin-2), previously reported as inhibiting sodium currents recorded in the whole-cell mode in human cultured myoballs (Brinkmeier et al. 1992). Within a few seconds, IL-2 caused a 50% decrease in the I_Na amplitude and shifted the steady-state inactivation curve to the left. However, when the authors have thereafter used the attached-patch configuration, as we have done here, they failed to observe any inhibition of sodium currents by IL-2 (Kaspar et al. 1994). Thus, in contrast to the action of IL-2 of inhibiting sodium channels in muscle by binding directly with the channels and keeping them in an inactivated state, as is the case with local anaesthetic-like drugs, the CNTF cytokine used in our study most likely requires an intracellular second messenger system to act.

A possible second messenger system is the PKC-dependent signalling pathway. Supporting this hypothesis, we observed that the effects of CNTF on sodium currents were blocked in the presence of chelerythrine, a PKC inhibitor, whereas they were similar to those obtained with OAG, a PKC activator. It has already been shown that PKC activation modulates the activity of different voltage-dependent sodium channel isoforms (Numann et al. 1991, 1994; Qu et al. 1994; Bendahhou et al. 1995). In particular, in HEK293 cells expressing the Na_v1.4 (or μ₁) sodium channel, the application of OAG or DOG (1,2-dioctanoyl-sn-glycerol), another diacylglycerol-like compound, induces a maximal peak sodium current reduction of about 40% (Bendahhou et al. 1995), a value close to the one that we observed (Fig. 6). Furthermore, the effects of diacylglycerol-like substances occurred within 5 min, reversed upon washing and were blocked in the presence of PKC inhibitors (Numann et al. 1994; Qu et al. 1994; Bendahhou et al. 1995). Thus, even if none of the native fibres examined here in the presence of CNTF exhibited a slowing or a speeding of inactivation as previously reported with PKC activators (Numann et al. 1994; Bendahhou et al. 1995), the pattern of current suppression by CNTF strongly shares the properties of Na⁺ current inhibition induced by the PKC activators. Moreover, our experiments showed that CNTF, as OAG, failed to alter the steady-state activation of sodium channels, whereas the midpoint voltage of steady-state fast inactivation curves was shifted by −6 mV in the presence of the cytokine, a phenomenon that was not observed when CNTF was applied with PKC inhibitors. Thus, as previously proposed by others with PKC activators (Qu et al. 1994; Bendahhou et al. 1995; Franceschetti et al. 2000), the inhibitory effect of CNTF on sodium currents can be attributable to a reduction in sodium channel availability combined with a hyperpolarizing shift in the voltage dependence of the fast-inactivation of residual currents.

The sodium channel availability at resting potential depends on the number of sodium channels present at the membrane surface and on the fraction of activable sodium channels, i.e. able to open in response to membrane depolarization (Ruff, 1996a, b). One cannot exclude the hypothesis that CNTF by inducing PKC activation causes a decrease in the Na⁺ channel expression. Indeed, Yanagita et al. (1996) have previously reported that the PKC activation induced a down-regulation of sodium channels leading to a decrease in channel density. However, the cytokine incubation time used in our experiments (10 min) is quite short compared to the several hours (up to 15 h) of application of the PKC activator used by Yanagita. It is generally thought that CNTF acts through the activation of the JAK/STAT, PI 3-kinase and MAP kinase pathways (Heinrich et al. 1998; Sleeman et al. 2000; Lelievre et al. 2001) that could affect the expression of Na⁺ channels. However, such an assumption is unlikely since the different kinases mentioned above are activated within 10 min following the CNTF application (Guillet et al. 1999) and their action on genes would then occur within a longer time. Moreover, the CNTF effects observed here could differ slightly from the short-term actions of cytokine previously reported in neurones that took place over a time scale of a few hours, during which CNTF would have time to alter protein expression (Stoop & Poo, 1995; Yan et al. 2000; Holm et al. 2002). Thus, our study argues in favour of a rapid CNTF action mediated by a PKC-sensitive pathway modulating the channel open probability rather than involving a variation of the channel density.

In skeletal muscle, slow inactivation, a very slow process compared to fast inactivation, would not participate in action potential termination but would regulate the number of excitable Na⁺ channels as a function of the membrane potential without changing the single-channel conductance or single-channel open time (Ruff, 1996a). Our results show that CNTF application shifts the voltage dependence of slow inactivation towards hyperpolarizing potentials. This could lead to a greater fraction of slow-inactivated and unexcitable channels at potentials close to membrane resting potential. It is possible that the changes in the slow inactivation process following CNTF application could be related to the phosphorylation of sodium channels, particularly by PKC, as recently described in brain neurones (Carr et al. 2003). These authors have shown that a slow, voltage-dependent process, that strikingly resembles slow inactivation, mediates the reductions in sodium channel availability induced by PKC, PKA and the activation of G protein-coupled receptors.

In cultured neurones, a 5 min application of CNTF induces PKC activation, probably by increasing intracellular diacylglycerol levels (Kalberg et al. 1993). The activity of PKC may be induced via the activation of phospholipase C-γ a well-known target for tyrosine phosphorylation after stimulation by neurotrophic factors, including CNTF (Widmer et al. 1993; Boulton et al. 1994). Moreover, it was previously reported that stimulation of PKC δ activity occurs in less than 10 min in the IL-6-stimulated cells (Jain et al. 1999). There are multiple consensus sequences for PKC phosphorylation in the Na_v1.4 (or μ₁) skeletal muscle sodium channel (Trimmer et al. 1989); one of them is the serine at position 1321, a site highly conserved in other types of sodium channels (Ser-1506 in rat brain Na_v1.2 (or rBIIA) sodium channel and Ser-1505 (or Ser-1503) in rat cardiac Na_v1.5 (or rH1) sodium channel) (Numann et al. 1991; Qu et al. 1994; Tateyama et al. 2003). This site is located in the centre of the intracellular loop between homologous domains III and IV of sodium channels, an important region in fast inactivation gating (Vassilev et al. 1988; Patton et al. 1992). Whereas this consensus PKC site is required for PKC effects on rat brain sodium channels (West et al. 1991), a mutation of the equivalent serine in cardiac and skeletal sodium channels only partially alters or completely fails to modify the PKC effects on sodium currents suggesting the existence of other PKC phosphorylation sites on these channel isoforms (Bendahhou et al. 1995; Murray et al. 1997; Tateyama et al. 2003). The Ser-906 located in the II–III loop of skeletal muscle sodium channel α-subunits could be a possible candidate. It was recently shown that mutations of this amino acid caused alterations in the slow inactivation process (Kuzmenkin et al. 2003).

Bendahhou et al. (1995) have also noticed that the addition of a phosphatase inhibitor in combination with OAG or DOG further inhibited sodium current by up to 90%, while applied alone, the phosphatase inhibitor failed to significantly affect sodium current. This suggests that the reduction of peak sodium current by a diacylglycerol-like substance mediated by PKC activation reflects a balance between enzymatic phosphorylation and dephosphorylation of sodium channels. Likewise, a similar mechanism could explain why, in the present study, CNTF failed to fully suppress sodium currents. We tried to check this hypothesis, but after pre-treatment with okadaic acid (a phosphatase inhibitor), the effects of CNTF showed a greater variability than the effects of CNTF alone (data not shown). A similar or higher (sometimes total) inhibition was observed. This could be explained by the fact that phosphatases can interact with a large number of metabolic pathways other than that of PKC, leading to stochastic effects from one cell to another.

A fibre difference in the enzymatic phosphorylation/dephosphorylation balance and/or in the expression rate of CNTF receptor could also account for the variability in CNTF-induced I_Na,max decrease that we observed between fibres. In the present study, the CNTF-induced negative shift of steady-state fast inactivation curves was greater in the patches in which the control half-maximal fast inactivation potentials were less negative (Fig. 4B). Considering the balance between enzymatic phosphorylation and dephosphorylation of sodium channels, this phenomenon could reflect fibre-specific differences in the initial phosphorylated state of the channels. For example, a greater number of phosphorylated channels may have already shifted the voltage dependence of fast inactivation curves in a hyperpolarizing direction, and in consequence may have prevented further hyperpolarization induced by CNTF. In this case, the difference in maximal I_Na,peak amplitude observed after CNTF application would be minor with respect to that obtained with fibres showing less negative sodium current fast inactivation curves. The linear correlation between the V_h,1/2 and the difference in maximal I_Na amplitude (Fig. 4B) supports the idea that CNTF effects depend on the initial state of phosphorylation of sodium channels.

CNTF is abundantly expressed in the cytoplasm of peripheral nerve Schwann cells (Friedman et al. 1992); nevertheless, the cytokine is not secreted but could be released uniquely by damaged cells, for instance after nerve injury (Lin et al. 1989; Stöckli et al. 1989; Sendtner et al. 1992a, b). Such a hypothesis implies that CNTF may exist in a wide range of concentrations near skeletal muscles in vivo, depending on physiological or pathophysiological conditions. A rapid PKC-dependent decrease of availability of excitable sodium channels as a result of CNTF in skeletal muscle could cause a decrease in fibre excitability and consequently modify contractile responses. In line with this idea, we have observed in our experiments that a 10 min application of CNTF causes resting membrane hyperpolarization, a phenomenon that does not occur in the presence of PKC inhibitors (Fig. 7). Taken together, these data lead us to propose that the CNTF released as a consequence of muscle innervation injury may produce a rapid decrease in muscle excitability, in particular by hyperpolarizing fibres. By acting on sodium channels, CNTF could also limit the firing rate of muscle fibres, a phenomenon that would protect muscle fibres from overstimulation. Then, in addition to its long-term myotrophic effects in denervated (Helgren et al. 1994) and in normally innervated muscles (Guillet et al. 1999; Fraysse et al. 2000), the present study suggests that CNTF could also play the role of a short-term skeletal muscle-protecting agent. For instance, in denervated muscles, CNTF could rapidly counteract fibre hyperexcitability by deepening, in the short-term, through a PKC-dependent pathway, the voltage dependence of slow inactivation of sodium channels, which is shallower after denervation due to the appearance of a juvenile and cardiac-like sodium channel isoform (Pappone, 1980; Kallen et al. 1990; Richmond et al. 1998). Further investigations are required to discover how CNTF activates PKC and whether this cytokine also modifies the activity of other ionic channels involved in electrical and/or contractile properties of skeletal muscle.

Although our present results show the possibility of a physiological role of CNTF in muscle contraction, the modification of muscle strength in CNTF-deficient mice, certainly significant, is small and is only prevalent postnatally (Masu et al. 1993). By contrast, mice lacking the CNTF receptor die shortly after birth (DeChiara et al. 1995). These contrasting phenotypes point to the existence of a second ligand for the CNTF receptor. The cardiotrophin-like cytokine/cytokine-like factor-1 composite cytokine (CLC/CLF) and the neuropoietin, both recently shown to activate the CNTF receptor (Lelievre et al. 2001; Derouet et al. 2004), could be good candidates. It remains to be seen whether these cytokines are also able to modify ionic channel activities in muscle.

The present results demonstrating the short-term effects of CNTF on sodium channel activity, probably through a PKC-activation pathway, could contribute to the understanding of the role of endogenous CNTF in skeletal muscle in normal and pathophysiological conditions.