Contribution of Persistent Sodium Current to Locomotor Pattern Generation in Neonatal Rats

Sabrina Tazerart, Jean-Charles Viemari, Pascal Darbon, Laurent Vinay, Frédéric Brocard


The persistent sodium current (INaP) is known to play a role in rhythm generation in different systems. Here, we investigated its contribution to locomotor pattern generation in the neonatal rat spinal cord. The locomotor network is mainly located in the ventromedial gray matter of upper lumbar segments. By means of whole cell recordings in slices, we characterized membrane and INaP biophysical properties of interneurons located in this area. Compared with motoneurons, interneurons were more excitable, because of higher input resistance and membrane time constant, and displayed lower firing frequency arising from broader spikes and longer AHPs. Ramp voltage-clamp protocols revealed a riluzole- or TTX-sensitive inward current, presumably INaP, three times smaller in interneurons than in motoneurons. However, in contrast to motoneurons, INaP mediated a prolonged plateau potential in interneurons after reducing K+ and Ca2+ currents. We further used in vitro isolated spinal cord preparations to investigate the contribution of INaP to locomotor pattern. Application of riluzole (10 μM) to the whole spinal cord or to the upper lumbar segments disturbed fictive locomotion, whereas application of riluzole over the caudal lumbar segments had no effect. The effects of riluzole appeared to arise from a specific blockade of INaP because action potential waveform, dorsal root–evoked potentials, and miniature excitatory postsynaptic currents were not affected. This study provides new functional features of ventromedial interneurons, with the first description of INaP-mediated plateau potentials, and new insights into the operation of the locomotor network with a critical implication of INaP in stabilizing the locomotor pattern.


In mammals, the locomotor pattern in hindlimb muscles is generated by spinal neuronal networks, referred to as central pattern generators (CPGs) (Kiehn 2006). Studies of CPG networks have shown that the intrinsic membrane properties of neurons are critical for pattern generation (Grillner et al. 2000; Marder and Bucher 2001). In rodents, there are increasing evidences that interneurons located in the ventromedial gray matter of upper lumbar segments constitute key elements of the hindlimb CPG (Butt et al. 2002; Cazalets et al. 1995; Hinckley et al. 2005; Hochman et al. 1994; Kiehn et al. 1996; MacLean et al. 1995; Wilson et al. 2005). To gain insight into the function of the hindlimb CPG, it is important to characterize individual ion channels of these interneurons and determine their role in generating the motor pattern.

Active conductances, such as potassium and calcium voltage-dependent currents, appear to regulate the locomotor pattern in vertebrates (Cazalets et al. 1999; Dale and Kuenzi 1997; Grillner et al. 2001; Kiehn et al. 2000). The contribution of sodium currents to motor activity is somewhat difficult to investigate because of their roles in the generation and propagation of action potentials. However, a persistent sodium current (INaP) is involved in the regulation of subthreshold excitability in a variety of cells (Crill 1996). Recent experimental and modeling studies indicate that INaP plays a role in generating rhythm in various CNS neurons such as some present in the trigeminal sensory nucleus (Brocard et al. 2006b), the pre-Bötzinger complex (Butera Jr. et al. 1999; Del Negro et al. 2005; Pena et al. 2004; Rybak et al. 2003), the hippocampus (Jinno et al. 2003), the subthalamic nucleus (Beurrier et al. 2000), the neocortex (Guatteo et al. 1996; van Drongelen et al. 2006), and the embryonic spinal cord (Darbon et al. 2004). Surprisingly, its contribution to locomotor pattern generation remains unexplored.

The present study was aimed at identifying and characterizing an INaP in the neonatal rat spinal cord, determining its importance in shaping the firing properties and its role in the operation of the locomotor circuitry. By using whole cell patch-clamp techniques in slice preparation, we show that INaP is present in ventromedial interneurons in upper lumbar segments and induces plateau potentials in these cells. In the isolated spinal cord preparation, the blockade of INaP alters fictive locomotion, suggesting a critical role of this current in the generation of locomotor behavior. Preliminary reports of these findings have been published in abstract form (Brocard et al. 2006a; Darbon et al. 2006).


In vitro preparations

Experiments were performed on 93 neonatal Wistar rats (1–5 days old). All surgical and experimental procedures conformed to guidelines from the French Ministry for Agriculture and Fisheries. Rats were anesthetized by hypothermia and decapitated.


The lumbar spinal cord was quickly removed and placed in cold (∼4°C) sucrose-based artificial cerebrospinal fluid [ACSF composition (in mM): 252 sucrose, 3 KCl, 1.25 KH2PO4, 4 MgSO4, 0.2 CaCl2, 26 NaHCO3, and 25 d-glucose] bubbled with 95% O2-5% CO2 (pH 7.4). In the same medium, transverse slices (350 μm) through the L1–L2 lumbar segments were prepared using a Vibratome (VT1000 S, Leica, Wetzlar, Germany). Slices were then incubated at room temperature (21–24°C) in the holding chamber filled with normal ACSF (in mM: 120 NaCl, 3 KCl, 1.25 NaH2PO4, 1.3 MgSO4, 1.2 CaCl2, 25 NaHCO3, and 20 d-glucose) for ≥1 h before recording. Slices were then transferred to an immersion slice chamber and perfused with the ACSF thermoregulated at 25°C (2 ml/min).


The spinal cord was isolated as described previously (see Norreel et al. 2003). Briefly, a ventral laminectomy was performed and the ventral and dorsal roots were cut. The spinal cord was transected at T10 and removed. The preparation was transferred to the recording chamber and perfused with normal Krebs solution (in mM: 130 NaCl, 4 KCl, 3.75 CaCl2, 1.3 MgSO4, 0.58 NaH2PO4, 25 NaHCO3, 10 glucose) saturated with 95% O2-5% CO2 (pH 7.4 at 25°C). The preparation was allowed to equilibrate for about 1 h.

Stimulation and recordings

Electrophysiological data were acquired through a Digidata 1322A interface using the Clampex 9 software (Molecular Devices).

In the slice preparation, neurons were visualized using a fixed-stage microscope (Eclipse E600FN, Nikon) coupled with a ×40 water-immersion lens. The image was enhanced with an infrared-sensitive CCD camera and displayed on a video monitor. Whole cell patch-clamp recordings in current- and voltage-clamp mode were performed with a Multiclamp 700B amplifier (Molecular Devices). Patch electrodes (6–9 MΩ) were pulled from borosilicate glass capillaries (1.5 mm OD, 1.12 mm ID; World Precision Instruments) on a Sutter P-97 puller (Sutter Instrument). Patch electrodes were filled with a K+-gluconate–based solution (in mM): 140 K+-gluconate, 5 NaCl, 2 MgCl2, 10 HEPES, 0.5 EGTA, 2 ATP, and 0.4 GTP (pH 7.3 with KOH; osmolarity ranged from 280 to 300 mOsm). To record the voltage-dependent sodium currents the internal solution for filling electrodes contained (in mM): 120 CsCl, 40 KCl, 5 NaCl, 2 MgCl2, 10 HEPES, 0.5 EGTA, 2 ATP, and 0.4 GTP (pH 7.3 with CsOH). After the establishment of a gigaseal, the pipette resistance and capacitance were compensated electronically. Recordings were sampled at 10 kHz and low-pass filtered at 10 kHz in the current-clamp mode and at 4 kHz in the voltage-clamp mode. To ensure the accuracy of the neuron location, the ventral half of the spinal cord was divided in quadrants as defined by previous authors (Theiss and Heckman 2005). All interneurons were recorded in the dorsomedial quadrant but the lamina X around the central canal was avoided. Thus most of the interneurons were recorded in lamina VII without excluding that some cells may be recorded in the lamina VIII. Motoneurons were identified on the basis of the size criterion as the largest cells in the ventrolateral quadrant. Neuron location was determined by examining electrode tip position under low magnification. To record miniature excitatory postsynaptic currents (mEPSCs) a K+-gluconate–based intracellular solution was used and tetrodotoxin (TTX, 1 μM) was added to the extracellular solution to disable spike activity. To acquire a sufficient sample of mEPSCs, experiments were performed in high extracellular K+ (9 mM). Recordings were carried out at a holding potential of −50 mV. Inhibitory postsynaptic currents were blocked by strychnine (1 μM), a glycine receptor antagonist, and bicuculline (20 μM), a γ-aminobutyric acid (GABA) receptor antagonist.

In the isolated spinal cord preparation, the locomotor-like activity was recorded (bandwidth: 70 Hz to 1 kHz) using extracellular stainless steel electrodes placed in contact with lumbar ventral roots (left/right L3 and/or left/right L5) and insulated with Vaseline. Afferent volleys from dorsal root were recorded (bandwidth: 70 Hz to 3 kHz) with suction electrodes, filled with normal saline. The tip of the suction electrodes was placed on the surface of the dorsal funiculus at the entry zone of the dorsal root. Intracellular recordings from L3–L5 motoneurons were obtained with glass microelectrodes (70–90 MΩ) filled with 3 M potassium acetate. Intracellular potentials were recorded using Axoclamp 2b (Molecular Devices). Motoneurons were identified by the antidromic response to stimulation of a ventral root. In some experiments, monopolar stainless steel electrodes were placed in contact with the dorsal roots and insulated with Vaseline for stimulation (0.3-ms duration, 0.5–3.5 V). All stimulations were delivered at a low frequency (every 30 s).

Data analysis

Electrophysiological data were analyzed with Clampfit 9 software (Molecular Devices). Passive membrane properties of cells were measured by determining from the holding potential the largest voltage deflections induced by small currents pulses to avoid the activation of voltage-sensitive currents. The input resistance was measured by the slope of the linear portion of the current–voltage (IV) relationship. The membrane time constant was determined by fitting an exponential function to the rising phase of the voltage trace used for determining the input resistance. In some cells, there was evidence of inward rectification (“sag”) during strong hyperpolarization. The size of the sag was expressed as the ratio of the negative voltage peak to the steady-state membrane potential. The rheobase was defined as the minimum current intensity necessary to induce an action potential during a 1-s pulse. Single-spike analysis was performed on the first spike elicited near the rheobase. Peak spike amplitude was measured from the threshold potential, and spike duration was defined as the time to fall to half-maximum peak. To investigate the afterhyperpolarizations (AHPs), single spikes were evoked by brief intracellular pulses at holding potential. The peak amplitude and duration (to half of the peak height) of AHPs were measured from the holding potential of −60 mV. Firing patterns were investigated with 1-s-long depolarizing current pulses of varying amplitudes. The average instantaneous firing frequency during the last 500 ms of the 1-s pulse was defined as the steady-state firing frequency. After seal rupture, the whole cell capacitance was assessed on-line from the integral of the current transient after a 10-mV voltage step (membrane test function, pClamp 9; Molecular Devices). Current density was calculated by dividing the peak current amplitude by cell capacitance. Voltage dependency and kinetics of whole cell currents were analyzed from normalized voltage-ramp data by fitting them with Boltzmann functions. The junction potential was corrected off-line based on the composition of the internal and external solutions used for recordings. mEPSCs were detected and analyzed using the MiniAnalysis Program (Synaptosoft, Fort Lee, NJ). Events were detected by setting the threshold value for detection at threefold the level of the root-mean-square noise (∼3–4 pA); therefore the detection threshold was 8–12 pA. The average values of mEPSC amplitude and frequency during the control period (10–15 min), and under the drug (35 min), were calculated over a 5-min time window.

The dorsal root–evoked potentials in lumbar motoneurons are composed of a purely glutamatergic monosynaptic component followed by mixed excitatory/inhibitory polysynaptic inputs (Seebach and Ziskind-Conhaim 1994; Wu et al. 1992; Ziskind-Conhaim 1990). Because the minimum delay from the mono- to the polysynaptic inputs has been estimated to be 3 to 5 ms during the first postnatal week (Kudo and Yamada 1987; Seebach et al. 1999), we measured the magnitude of the monosynaptic response by considering the area under the curve over the first 3 ms of the synaptic potential. We also determined the polysynaptic response by measuring the area of the synaptic potential over the 30 ms after the monosynaptic response. With stimulation of the dorsal root, a negative–positive–negative complex response, recorded with a suction electrode from the dorsal funiculus, coincided with the arrival of the afferent volley. The number of afferent fibers recruited was evaluated by measuring the area of the positive component of the complex response (Lidierth 2006). The stimulus–response relationships were plotted and fitted by Boltzmann functions. The strength of stimulation for any given shock was normalized to the threshold stimulus (T) required to evoke a response in control conditions. The latency was measured from the stimulus artifact to the onset of the response.

Alternating activity between opposite (right/left L3 or right/left L5) and ipsilateral (left or right L3/L5) recordings was taken to be indicative of fictive locomotion. During an episode of fictive locomotion, cycle periods shortened progressively and reached a steady state within 5 min (data not shown) (Cazalets et al. 1999; Sqalli-Houssaini et al. 1993). From this time, three successive 5-min periods of locomotor-like activity were analyzed 15–20 min after its pharmacological induction (see Drug application below). Raw extracellular recordings from ventral roots were rectified and resampled at 50 Hz. Amplitude and duration of ventral root bursts were measured by a threshold function that determines the peak, the onset, and end of bursts of activity. The threshold was usually set to about 30% of the peak value. An autocorrelation analysis was performed to measure stability of the rhythm. The regularity was estimated by measuring the positive coefficient correlation at zero-phase lag. The cycle period was calculated by measuring the distance between the first two peaks of the autocorrelogram. Cross-correlation analysis was performed to measure the coupling between the left and right L3 or L5 ventral bursts during the different experimental conditions. The quality of the alternation was estimated by measuring the negative correlation coefficient at zero-phase lag (center of the cross-correlogram). Details about the auto- and cross-correlation analyses have been described elsewhere (Madriaga et al. 2004; Pearlstein et al. 2005).

Data are presented as means ± SE. A Student's t-test was used for statistical analysis when two groups were compared and a one-way ANOVA followed by a Tukey test was used for multiple group comparisons. Values of P < 0.05 were considered significant (GraphPad Prism 4.0, GraphPad Software, San Diego, CA).

Drug application

All drugs were purchased from Sigma–Aldrich, kept as concentrated stock solutions, and diluted to their final concentration. The following pharmacological agents were used: tetrodotoxin (TTX, 0.5–1 μM); strychnine (1 μM); bicuculline (20 μM); riluzole (10–20 μM); N-methyl-dl-aspartic acid (NMA, 18–22 μM); and 5-hydroxytryptamine creatinine sulfate (5-HT, 5 μM). In isolated spinal cord preparations, fictive locomotion was elicited by an application of NMA/5-HT lasting 30–40 min. In some experiments, Vaseline barriers were built at the L2–L3 level to superfuse the locomotor network located in the rostral lumbar cord independently from the more caudally located motoneurons. To study the effect of blocking INaP on fictive locomotion, riluzole was preincubated for 10 min to reach a steady drug concentration in the recording chamber before its superfusion together with NMA/5-HT. Riluzole was bath applied only once for each preparation because it is difficult to wash. Note that riluzole and TTX were randomly used to block INaP in both interneurons and motoneurons recorded from slice preparations.


Electrophysiological characterization of ventromedial interneurons

The first step in this study was to compare the membrane properties of interneurons, recorded in the ventromedial gray matter, with those of motoneurons. We recorded 68 cells (41 interneurons and 27 motoneurons) from L1–L2 lumbar slices using whole cell patch-clamp techniques. Data are summarized in Table 1. Only cells exhibiting a stable resting or holding membrane potential and an action potential amplitude >50 mV were considered.

View this table:

Passive and active membrane properties of interneurons and motoneurons

In contrast to motoneurons, most interneurons (∼70%) fired spontaneously with a mean frequency of 7.5 ± 0.8 Hz, which precluded determination of their resting membrane potential. Injection of a small negative current (−10 to −20 pA) prevented them from firing (data not shown). The remaining silent interneurons exhibited a resting membrane potential more depolarized than that of motoneurons. Analysis of the voltage responses to subthreshold current pulses (Fig. 1, A1 and A2) revealed that interneurons had larger time constants. Slopes of voltage–current (VI) relationships were steeper (Fig. 1A3), indicating that their input resistance was larger. In >90% of cells, application of a large hyperpolarizing current pulse revealed the presence of an inward rectifying response (see arrowheads in Fig. 1, A1 and A2), but the magnitude of this rectification was more pronounced for interneurons than for motoneurons. Although the firing threshold was significantly more depolarized in interneurons, their rheobase was smaller, suggesting a greater excitability. Interneurons and motoneurons generated a train of spikes in response to suprathreshold current injection (Fig. 1, B1 and B2). Spikes in interneurons, compared with motoneurons, were characterized by a smaller amplitude, a longer duration (Fig. 1C), and were followed by small-amplitude and long-lasting AHPs (Fig. 1D). These distinguishing features may account for the lower firing frequency of interneurons compared with that of motoneurons (Fig. 1, B1 and B2 and Table 1).

FIG. 1.

Passive membrane properties and firing characteristics of ventromedial interneurons and motoneurons. A: typical responses of an interneuron (A1) and a motoneuron (A2) to incrementing 1-s current injections (bottom traces) with their respective voltage–current relationship (A3). Large hyperpolarizing currents revealed the presence of an inward rectification causing a depolarizing “sag” (see arrowheads). Initial potential was held at −80 mV. B: responses of an interneuron (B1, gray traces) and a motoneuron (B2, black traces) to a 1-s depolarizing current pulse (twice the rheobase). Note that the more hyperpolarized threshold of the spike in motoneurons than that in interneurons may give the first wrong impression that the spike was higher in interneurons than that in motoneurons. C: action potentials aligned at the threshold potentials to enable comparisons. D: afterhyperpolarizations (AHPs) following a single action potential (truncated). Initial holding potential: −60 mV.

Characterization of INaP in ventromedial interneurons

Here, we tested the presence of the INaP in ventromedial interneurons (n = 17) and motoneurons (n = 10) by means of a voltage-clamp protocol ramping the voltage command from −80 to +10 mV over 5 s. Slow ramps (<20 mV/s) were initially selected because they prevented activation of fast-decaying Na+-current component. K+ conductances were also minimized by an intracellular CsCl loading while the external CaCl2 was isosmotically replaced by MgCl2 to suppress voltage-dependent Ca2+ currents and synaptic transmission. In both cell types, a slow voltage-clamp command produced an inward sodium current (Fig. 2, A1 and A2), presumably INaP, that was blocked by riluzole (10–20 μM) or TTX (1 μM). Riluzole or TTX subtraction facilitated isolation of INaP (Fig. 2, B1 and B2). The continuous IV relationship showed that the voltage-dependent activation threshold of the INaP was similar in both types of neurons (interneuron: −59.3 ± 0.8 mV, n = 17; motoneuron: −59.8 ± 1.8 mV, n = 10; P > 0.05, t-test). To characterize the voltage-dependency kinetics of INaP, we fitted a Boltzmann function to conductance–voltage data normalized to the maximum peak current (Fig. 2, C1 and C2). Average half-activation potentials and slope factors of the fitted curves were not significantly different (P > 0.05, t-test) in interneurons (V1/2 = −49.5 ± 2 mV, k = −4.9 ± 0.7, n = 17) and motoneurons (V1/2 = −52.6 ± 2 mV, k = −5.8 ± 0.8, n = 10). However, INaP amplitude measured at the peak of the IV relationship was significantly smaller in interneurons (−54.5 ± 5.1 vs. −163.5 ± 23.9 pA; P < 0.001, t-test). To assess the differences between the amplitude of the currents regardless of cell size of the two cell populations, INaP was normalized with respect to the whole cell surface area derived from measurements of the membrane capacitance. The current density tended to be higher in interneurons (1.82 ± 0.31 vs. 0.98 ± 0.07 pA/pF; P < 0.05, t-test).

FIG. 2.

Basic features of persistent sodium current (INaP) in ventromedial interneurons and motoneurons. A: currents evoked by voltage ramps from −80 to +10 mV over 5 s, under control conditions and in the presence of riluzole (10 μM) or tetrodotoxin (TTX, 1 μM) in ventromedial interneuron (A1) and motoneuron (A2). B: subtracted currents obtained by digitally subtracting the traces recorded in 10 μM riluzole or 1 μM TTX from control traces. B1 and B2: same cells as in A1 and A2, respectively. C: voltage dependency of the activation of INaP in a ventromedial interneuron (C1) and a motoneuron (C2). Sodium conductances were derived from the early parts of the subtracted currents and normalized to the maximal values and fitted with single Boltzmann functions. Fitting parameters were: half-activation potential (V1/2) = −53 mV, slope factor of fitted curve (k) = −6 mV (C1); V1/2 = −52.3 mV, k = −6 mV (C2). Dotted lines indicate the half-maximal activation and the corresponding V1/2.

To test whether the INaP differentially depolarizes interneurons and motoneurons, we performed a series of experiments in which a current-clamp manifestation of INaP—the sodium plateau potential—was induced by a brief (2-ms) suprathreshold depolarizing pulse (Astman et al. 2006; Brocard et al. 2006b; Kim and McCormick 1998; Su et al. 2001). In most interneurons (10 of 12 cells), in which K+ currents were blocked by an intracellular CsCl loading while the external CaCl2 was replaced by MgCl2, a brief depolarizing pulse evoked a compound response consisting of a fast spike followed by a plateau potential that outlasted the current pulse (mean duration: 2.6 ± 0.5 s; Fig. 3, A1 and A2). By contrast, in all motoneurons tested (n = 10), the fast spike was not followed by the plateau potential irrespective of the duration and the amplitude of current pulses used (Fig. 3A3). TTX applied for a short time (5–10 min, 0.5 μM, n = 5) appreciably abolished the plateau (Fig. 3A1), without affecting the initial spike (Fig. 3A1, inset), but a longer superfusion of TTX fully abolished spikes (Fig. 3A1). This suggests that the current underlying the plateau was sensitive to low concentrations of TTX, likely attributable to INaP. To examine this issue further, we bath-applied riluzole (10–20 μM, n = 5). The TTX-sensitive plateau was substantially shortened in the presence of riluzole without affecting the initial spike (Fig. 3A2 and inset). Effects of TTX and riluzole were partially reversible within 30–60 min (Fig. 3B).

FIG. 3.

Contribution of INaP to plateau potentials in ventromedial interneurons. A: voltage responses of ventromedial interneurons (A1 and A2) or motoneurons (A3) to a brief current pulse (500 pA, 2 ms) before and during a bath application of TTX (0.5 μM, A1) or riluzole (10 μM, A2). Responses were recorded in Ca2+-free artificial cerebrospinal fluid (ACSF) with a cesium-based intracellular pipette solution. Insets: enlargements of shaded areas of recordings in A1 and A2. Note in insets the persistence of the spike, whereas the plateau potential was suppressed after the riluzole or the TTX application. B: partial recovery of the plateau after a washout period (30–60 min).

Absence of side effects of riluzole at 10 μM

Before evaluating the role of INaP on NMA/5-HT–induced fictive locomotion, it was important to selectively block this current in the isolated spinal cord. Aside from its effects on INaP (Niespodziany et al. 2004; Urbani and Belluzzi 2000), riluzole has been reported to have multiple effects on voltage-gated Na+, K+, and Ca2+ channels underlying the action potential waveform (Ahn et al. 2005; Cao et al. 2002; Huang et al. 1997; Stefani et al. 1997; Urbani and Belluzzi 2000; Zona et al. 1998). Riluzole has also been reported to depress both excitatory (Cheramy et al. 1992; Doble 1996; Jehle et al. 2000; Martin et al. 1993; Pace et al. 2007) and inhibitory (Mohammadi et al. 2001) synaptic transmission. To test the specificity of riluzole on INaP in isolated spinal cord preparations, we first analyzed the effects of different concentrations of this drug on the shape of the antidromic action potentials evoked in L3–L5 motoneurons (n = 10) by ventral root stimulation (Fig. 4 A). Increasing the concentration ≤10 μM affected neither the amplitude (80.4 ± 1.6 vs. 79.0 ± 1.3 mV; n = 10, P > 0.05, paired t-test; Fig. 4, A and B) nor the rise slope (104 ± 7 vs. 105 ± 6; n = 10, P > 0.05, paired t-test; Fig. 4B), the decay slope (−54.3 ± 3.8 vs. −55.4 ± 4.2; n = 10, P > 0.05, paired t-test; Fig. 4B), and the threshold (−46.6 ± 1.9 vs. −46.8 ± 2.4 mV; n = 10, P > 0.05, paired t-test) of the antidromic spike; however, it significantly increased its latency (2.9 ± 0.4 vs. 3.5 ± 0.5 ms; n = 10, P < 0.001, paired t-test; Fig. 4B). Furthermore, the input resistance of motoneurons was not altered (103.5 ± 6.3% of control value; P > 0.05, paired t-test). These results suggest that the availability of calcium, potassium, and fast sodium channels at the soma of motoneurons was not compromised by riluzole at 10 μM. Action potentials elicited in lumbar motoneurons could be detected by ventral root recording (Fig. 4C), suggesting that riluzole did not prevent the propagation of orthodromic action potentials along the motor axons. Note that, when riluzole was bath-applied at a higher concentration (20 μM), the amplitude of the antidromic spike was significantly reduced (28 ± 5%, n = 3, P < 0.05, paired t-test; Fig. 4A).

FIG. 4.

Effects of riluzole on action potentials recorded from lumbar motoneurons. A: time course of the effect of different concentrations of riluzole (from 5 to 20 μM) on the amplitude of the antidromic action potential. B: antidromic action potentials recorded before (black trace) and 40 min after adding riluzole (10 μM, gray trace). C: mean (n = 100 sweeps) action potential initiated by current injection into motoneuron (top trace) and the resulting extracellular signal recorded from the homonymous ventral root (bottom trace) 40 min after adding riluzole.

We further analyzed the shape of the dorsal root–evoked postsynaptic potentials in L3–L5 motoneurons (n = 6) to evaluate the effect of riluzole (10 μM) on synaptic transmission. Under control conditions, single submaximal dorsal root stimulations evoked large postsynaptic potentials (PSPs) in motoneurons. After 35 min of bath application, riluzole significantly decreased the area of PSPs (45 ± 18% of control, n = 5 preparations, P < 0.05, paired t-test; Fig. 5 A1). The time course of this effect varied from cell to cell, but, in most cases, the reduction in the magnitude of PSPs started with a 10-min latency (Fig. 5A2). Afterward, the area of PSPs decreased gradually during the remaining perfusion time. To discriminate possible alteration of excitatory synaptic transmission from a depression in the overall excitability of the spinal cord, the glutamatergic monosynaptic component of the dorsal root–evoked potentials, measured at 3 ms from response onset (see methods), was assessed for a stimulus–response relationship (Fig. 5, B1B4). Under control conditions, the monosynaptic potentials increased as a function of stimulus intensity from threshold and reached a plateau for levels of stimulation >1.5-fold that of threshold (Fig. 5, B1B4 and C1). The measured input–output property demonstrates a sigmoidal shape and is characterized by threshold, maximum slope, and plateau values (Fig. 5C1). By 35 min after supplementary riluzole, in addition to an increased latency of monosynaptic potentials (7.9 ± 0.6 vs. 8.7 ± 0.7 ms; n = 6, P < 0.001, paired t-test; see inset in Fig. 5B4), the stimulus–response relationships was shifted to the right (Fig. 5C1). The threshold intensity for evoking monosynaptic potentials thus increased from 0.73 ± 0.09 to 0.97 ± 0.12 V (n = 6, P < 0.05, paired t-test). The shift to the right of relationships was observed in the absence of changes in slope factors (0.05 ± 0.02 vs. 0.05 ± 0.01; n = 6, P > 0.05, paired t-test) and in maximum area (21.1 ± 5.5 vs. 21.8 ± 6.3 mV·ms; n = 6, P > 0.05, paired t-test) of the Boltzmann equation. These results suggest that direct glutamate release machinery was only minimally compromised by riluzole. Similar results were obtained for the mixed excitatory–inhibitory polysynaptic component of the dorsal root–evoked potentials, measured 30 ms after the monosynaptic response (Fig. 5, B1B4 and C2). Slope factors (0.043 ± 0.003 vs. 0.036 ± 0.006; n = 6, P > 0.05, paired t-test) and maximum areas (371 ± 46 vs. 384 ± 47 mV · ms; n = 6, P > 0.05, paired t-test) of the Boltzmann equation were unchanged. These results suggest that, in addition to the excitatory neurotransmission, inhibitory synapses impinging on motoneurons also were only minimally compromised by riluzole. However, to further test the possibility that riluzole may affect central synaptic transmission, we analyzed miniature excitatory postsynaptic currents (mEPSCs) recorded from L1–L2 ventromedial interneurons (n = 6) in slice preparations (Fig. 6 A). Under control conditions, the mean interval between mEPSCs was 1.8 ± 0.5 s and mean amplitude of mEPSCs was 15.4 ± 1.5 pA at a holding potential of −60 mV. We found that neither the mEPSC intervals (1.8 ± 0.2 s, P > 0.05, paired t-test; Fig. 6, A and C1) nor the mEPSC amplitudes (15.4 ± 1.2 pA, P > 0.05, paired t-test; Fig. 6, B and C2) were affected by riluzole (10 μM). It is therefore very likely that riluzole at the concentration of 10 μM has no effect on synaptic transmission in the spinal cord.

FIG. 5.

Effects of riluzole on dorsal root–evoked synaptic potentials in lumbar motoneurons. A1: mean submaximal dorsal root–evoked synaptic potentials recorded in a motoneuron at resting potential (VR = −70 mV) before (black trace) and 30 min after adding riluzole (10 μM, gray traces). A2: time course of the area of dorsal root–evoked synaptic potentials (normalized to control value, i.e., the average of the last 3 measurements before onset of riluzole application) after adding 10 μM riluzole (time t = 0; area was measured every minute). Data were cumulated from 5 motoneurons. B: graded responses to increasing stimulus intensities (stimulus strengths: 0.7, 1, 1.1, and 1.3 V in B1, B2, B3, and B4, respectively) in a lumbar motoneuron at resting membrane potential before (black traces) and 40 min after adding riluzole (10 μM, gray traces). Inset: enlargement of shaded area of the recordings in B4 to illustrate the increased latency of postsynaptic potentials in the presence of riluzole. C: relationship between the stimulus intensity and the postsynaptic response, measured at 3 ms (C1) and 30 ms (C2) from onset, before (black squares and trace) and 40 min after adding riluzole (10 μM, gray circles and trace). Sigmoidal curves in C were fitted to the stimulus–response data obtained from synaptic responses illustrated in B. Magnitude of postsynaptic potentials was expressed in percentage of the area of the maximum subthreshold response, whereas the stimulation intensity was normalized to the threshold intensity (T) eliciting a synaptic response in control conditions.

FIG. 6.

Effects of riluzole on miniature excitatory postsynaptic currents (mEPSCs) in ventromedial interneurons. A: traces of continuously recorded mEPSCs in a ventromedial interneuron voltage clamped at −60 mV. mEPSCs recorded before and 30 min after adding riluzole (10 μM) were pharmacologically isolated in the presence of TTX (1 μM), strychnine (1 μM), and bicuculline (20 μM). B: averaged mEPSCs recorded before and 30 min after adding riluzole. C: mean interevent interval (C1) and amplitude (C2) of mEPSCs before (black bars) and 30 min after the riluzole was bath-applied (gray bars).

The shift to the right of the stimulus–response relationships seems then to reflect a decrease in the excitability of the nervous system. Some evidence for excitability changes at a peripheral level comes from the increased latency of both the dorsal root–evoked potentials and the antidromic action potential of motoneurons. To test this issue, we examined extracellular presynaptic afferent volleys in response to stimulation of the dorsal root in six preparations (Fig. 7 A). After adding riluzole, the latency of afferent volleys increased (3.6 ± 0.4 vs. 4.7 ± 0.6 ms; P < 0.01, paired t-test), indicating a depression in the conduction velocity of action potentials. Changes of the stimulus–response relationship of the afferent volleys paralleled those previously observed for the dorsal root–evoked postsynaptic potentials. The stimulus–response relationships were shifted to the right in the presence of riluzole (Fig. 7B). The threshold stimulus intensity to recruit the most excitable axons was thus increased from 0.32 ± 0.07 to 0.51 ± 0.8 V (P < 0.05, paired t-test). Slope factors (0.57 ± 0.11 vs. 0.53 ± 0.13; P > 0.05, paired t-test) and maximum areas (7 ± 1.1 vs. 6.4 ± 1 mV · ms) of the Boltzmann equation were unchanged, demonstrating that the gain of axonal recruitment and the number of axons recruited were not significantly affected. We conclude that the shift of the stimulus–response relationship for the dorsal root–evoked potentials toward higher stimulus threshold likely reflects a depression of presynaptic axonal excitability induced by riluzole without an obvious effect on central synapses.

FIG. 7.

Effects of the riluzole on dorsal root–evoked afferent volleys. A: typical graded responses to increasing stimulus intensity before (black traces) and 40 min after adding riluzole (10 μM, gray traces). B: relationship between the stimulus intensity of the dorsal root and the area of the positive component of the afferent volley before (black squares and trace) and 40 min after adding riluzole (10 μM, gray circles and trace). A sigmoidal curve was fitted to the stimulus–response data obtained from the afferent volleys illustrated in A. Magnitude of responses was expressed in percentage of the area of the maximum positive component of the afferent volley, whereas the stimulation intensity was expressed in multiples of the threshold intensity (T) for eliciting an afferent volley in control conditions.

Altogether we assume that riluzole, used at 10 μM over the isolated spinal cord, preferentially affects INaP. All the following experiments to test the effects of riluzole on fictive locomotion were thus performed at this concentration.

Effects of blocking INaP on NMA/5-HT–induced fictive locomotion

In nine isolated spinal cord preparations, a locomotor-like activity was induced by bath application of NMA/5-HT (18 μM/5 μM; Fig. 8 A1). The cycle periods were initially long and shortened progressively until a steady state that was reached within 5 min (data not shown). Starting at this time, time series analysis of the locomotor-like activity was performed over three successive time windows (5 min each). NMA/5-HT induced stable, long-lasting locomotor activity because correlation coefficients and burst parameters (period, duration, and amplitude) did not change with time (Fig. 8B, black bars; P > 0.05, one-way ANOVA; see also Sqalli-Houssaini et al. 1993). At the concentration of NMA/5-HT used, mean auto- and cross-correlation coefficients were: 0.64 ± 0.015 and −0.69 ± 0.018, respectively; locomotor period and burst duration were: 2.3 ± 0.09 s and 1.4 ± 0.06 s, respectively.

FIG. 8.

Effects of riluzole on fictive locomotion. A: locomotor-like activity (at different times: ∼7.5, 12.5, and 17.5 min after the fictive locomotion started) induced by N-methyl-dl-aspartate/5-hydroxytryptamine creatinine sulfate (NMA/5-HT, 18 μM/5 μM, respectively) alone (A1) or together with riluzole (10 μM, A2). Extracellular recordings of right (rL3 and rL5) and left (lL3) ventral root activities illustrated in A1 and A2 are collected from 2 different preparations. B: autocorrelation coefficient (B1), cross-correlation coefficient (B2), burst amplitude (B3), and burst duration (B4) in the 3 time windows analyzed (same as in A), in the absence (black bars) or presence of riluzole (white bars). Values measured during the first time window (5 to 10 min after the fictive locomotion started) are taken as the “control” responses (100%) in both groups.

Because the parameters and the stability of the locomotor-like activity were different on the second application of NMA/5-HT, compared with the first one (data not shown), the effects of riluzole were studied by performing a second series of experiments (n = 8 animals) in which riluzole was preincubated over the whole spinal cord 10 min before addition of NMA/5-HT (Fig. 8A2). Riluzole did not prevent fictive locomotion. However, in contrast with the control experiments, the locomotor-like activity was not stable throughout the application of riluzole. Auto- and cross-correlation coefficients decreased with time from 0.54 ± 0.02 to 0.25 ± 0.04 (53 ± 8% reduction, Fig. 8B1, white bars; P < 0.001, one-way ANOVA) and from −0.54 ± 0.03 to −0.32 ± 0.03 (41 ± 6% reduction, Fig. 8B2, white bars; P < 0.001, one-way ANOVA), respectively. These results show that the regularity of the locomotor rhythm and the left–right alternation declined throughout the application of riluzole. Reductions in both the amplitude (21 ± 17%; Fig. 8B3, white bars; P < 0.05, one-way ANOVA) and the duration (27 ± 4% going from 1.3 ± 0.1 to 0.94 ± 0.05 s; Fig. 8B4, white bars; P < 0.001, one-way ANOVA) of the locomotor bursts were also observed. After 90 min of washout, NMA/5-HT was able to reinduce a short episode of rhythm (<5 min) but without locomotor pattern recovery (data not shown). Note that the locomotor period was not significantly changed by riluzole (mean value: 2.7 ± 0.1 s; P > 0.05, one-way ANOVA).

The riluzole-induced disruption of fictive locomotion might result from an overall depression of neuronal excitability. Increasing this excitability by raising the concentration of NMA should prevent this depression and would be expected to, at least partly, restore a stable, long-lasting locomotor activity. To test this hypothesis, a third series of experiments was performed and the control locomotor pattern was investigated by bath-applying 22 μM of NMA and 5 μM of 5-HT (Fig. 9 A1). In this condition, a few minutes after fictive locomotion started (Fig. 9A2), the frequency of bursting increased considerably (Fig. 9A3) until the locomotor pattern was disrupted (Fig. 9A4) and further replaced by a tonic activity (Fig. 9A5). The locomotor episode, characterized by left–right alternation, lasted on average 8.4 ± 0.4 min (n = 8 animals). In separate experiments (n = 6 animals), riluzole was preincubated and then superfused together with a high concentration of NMA (Fig. 9B1). This significantly shortened the locomotor episode (5.5 ± 1.4 min; P < 0.05, unpaired t-test; Fig. 9, B2B4) and did not prevent the emergence of a tonic activity (Fig. 9B5). These results suggest that disruption of fictive locomotion by riluzole may not be the result of a reduced excitability of locomotor networks but rather of a specific blockade of the capability of the CPG to generate locomotor patterns.

FIG. 9.

Riluzole-induced disruption of the locomotor pattern was not prevented by increasing the excitability of spinal networks with high concentration of NMA. A and B: locomotor-like activity as a function of time during bath application of NMA/5-HT (22 μM/5 μM) in the absence (A1) or presence (B1) of riluzole (10 μM). Horizontal bars below extracellular recordings indicate the duration of fictive locomotion. Shaded areas are shown at a faster timescale in A or B2B5. Recordings in A and B are from 2 different preparations.

To determine whether the riluzole-induced disruption of fictive locomotion results from a blockade of INaP in motoneurons and/or premotor interneurons, the bath was partitioned by a Vaseline barrier built at the L2–L3 level of the spinal cord (Fig. 10 A). Thus the thoracolumbar part of the spinal cord (T13–L2), where the CPG is mostly located (Kiehn 2006), and the caudal segments (L3–L6), which contain most of the motoneurons innervating the hindlimb muscles (Nicolopoulos-Stournaras and Iles 1983), could be perfused separately. In these experiments, motor bursts recorded from left–right L5 ventral roots were analyzed. A control locomotor-like activity was first induced by bath application of NMA/5-HT (18 μM/5 μM) in both compartments (Fig. 10A1). In the 10 preparations tested, the NMA/5-HT induced stable, long-lasting locomotor activity (Fig. 10, B1 and C1C4, black bars; P > 0,05, one-way ANOVA). Mean auto- and cross-correlation coefficients were: 0.53 ± 0.021 and −0.58 ± 0.023, respectively; mean locomotor period and burst duration were: 1.93 ± 0.07 and 1.2 ± 0.1 s, respectively. When riluzole was preincubated and superfused in the caudal bath (Fig. 10A2), the locomotor pattern was not altered in any of the four preparations tested (Fig. 10, B2 and C1C4, gray bars; P > 0.05, one-way ANOVA). Auto- and cross-correlation coefficients (0.54 ± 0.021 and −0.54 ± 0.016, respectively), period, and burst duration (2.4 ± 0.4 and 1.3 ± 0.16 s, respectively) were close to control values. Only a slight nonsignificant decrease in the amplitude of motor bursts was noted throughout the application of riluzole (9.2 ± 16.8%; Fig. 10C3, gray bars). By contrast, when riluzole was preincubated and superfused rostral to the barrier (Fig. 10A3), the quality of the locomotor pattern progressively decreased before its disruption (Fig. 10B3). Auto- and cross-correlation coefficients decreased from 0.45 ± 0.04 to 0.21 ± 0.05 (52.2 ± 11.3% of decrease; Fig. 10C1, white bars; P < 0.001, one-way ANOVA) and from −0.49 ± 0.02 to −0.26 ± 0.05 (45.9 ± 11.2% of decrease; Fig. 10C2, white bars; P < 0.001, one-way ANOVA), respectively. Concomitant reductions in both the amplitude (19.4 ± 21.7%; Fig. 10C3, white bars; P < 0.05, one-way ANOVA) and the duration (22.7 ± 11.7% going from 1.2 ± 0.12 to 0.94 ± 0.15 s; Fig. 10C4, white bars; P < 0.001, one-way ANOVA) of locomotor bursts were also observed. Only the locomotor period was not significantly changed by riluzole during the analyzed time window (mean value: 2.1 ± 0.2 s; P > 0.05, one-way ANOVA).

FIG. 10.

Differential effects of riluzole on fictive locomotion when selectively superfused over the caudal (below L3 segment) or the rostral (above the L3 segment) lumbar segments. A1A3: diagrams of the experimental paradigm. A Vaseline barrier was built at L2/L3 level. B1B3: locomotor-like activity as a function of time during bath application of NMA/5-HT (18 μM/5 μM) in both compartments (B1) and after adding riluzole in the caudal (B2) or rostral (B3) baths. Raw recordings in B1, B2, and B3 are collected from opposite ventral roots (lL5 and rL5) in 3 different preparations. C: autocorrelation coefficient (C1), cross-correlation coefficient (C2), burst amplitude (C3), and burst duration (C4) in the 3 experimental conditions: NMA/5-HT alone (black bars), riluzole superfused caudal (gray bars), or rostral (white bars) to the barrier. Values measured during the first time window analyzed (5 to 10 min after fictive locomotion started) are taken as the “control” responses (100%) in all groups.

Altogether, our results suggest that INaP in premotor interneurons, presumably taking part in the CPG, plays a more important role in the generation of coordinated locomotor activity than INaP in motoneurons.


Our study describes a consistent electrophysiological phenotype of ventromedial interneurons, discriminating them from motoneurons, in particular an INaP that takes part in generating plateau potentials. Our results also provide new insights into the operation of the CPG with a critical implication of INaP. We suggest that INaP from ventromedial interneurons plays a key role in pattern generation during locomotion.

Electrophysiological signature of ventromedial interneurons

On the basis of their electrophysiological properties, ventromedial interneurons could be easily distinguished from motoneurons. They were more excitable, likely as a result of their higher input resistance and membrane time constant. These larger values, close to those found by Hochman et al. (1994) in interneurons located in a similar region, may be attributable to a smaller size of cells as suggested by their lower cell capacitance. Lumbar interneurons in laminae V–VII have been classified based on the characteristics of their firing properties in neonatal rat slice preparation (Szucs et al. 2003). Among them, a unique category of interneurons designated “slow” (restricted to the ventromedial gray matter) shares some similar characteristics with interneurons recorded in the present study: long duration of action potential and AHPs, which limits the firing rate. We presume that “slow” interneurons are homologous to those recorded herein. Biophysical properties underlying the action potential and AHP waveforms were not investigated. It is likely that the longer time constant of interneurons underlies, at least partly, the slower time course of both action potentials and AHPs. Distinct properties, densities, and/or locations of Na+ and K+ channels may also contribute to this heterogeneity.

The persistent sodium current in the lumbar spinal cord

Our voltage-clamp recordings revealed that the INaP amplitude is smaller in ventromedial gray interneurons than that in motoneurons, which apparently results from differences in the cell size because the INaP density tended to be larger in interneurons. Despite its activation below the firing threshold, it appears that INaP does not contribute to differences in the shape of the action potential between interneurons and motoneurons. Indeed, riluzole at 10 μM had no effect on the action potential in either group of neurons. These results are consistent with those of Kuo et al. (2006) who showed that riluzole (10 μM) affects neither the spike nor the AHP waveforms in mouse spinal motoneurons. Although INaP has only a slight contribution, if any, to the inward current flow that accompanies the action potential, it takes a large part in generating plateau potentials exclusively observed in ventromedial interneurons after reducing K+ and Ca2+ currents. None of the motoneurons recorded in the present study exhibited such INaP-dependent behavior. Strong excitability of ventromedial interneurons, resulting from passive membrane properties, associated with their high INaP density may be major factors in generating plateau potential. However, adult mammalian motoneurons display TTX-sensitive plateau potentials (Hsiao et al. 1998; Li and Bennett 2003). This discrepancy with our data likely results from immature INaP in neonatal rat motoneurons because a postnatal maturation of INaP was observed in spinal motoneurons (Garcia et al. 1998).

The presence of INaP in upper lumbar ventromedial interneurons is consistent with a recent study showing that ventral horn interneurons through lumbar spinal cord exhibit such INaP at the end of the second postnatal week (Theiss et al. 2007). Our data are also consistent with those of a previous study showing that locomotor-related neurons exhibited a negative-slope current response when a slow depolarizing voltage ramp was applied (Kiehn et al. 1996). The ionic nature of the current involved was not investigated, although its biophysical properties were similar to those of the INaP characterized in the present study and in other preparations (Crill 1996; Theiss et al. 2007). Sodium channel subunits NaV1.6, widely distributed throughout the nervous system (Krzemien et al. 2000; Schaller and Caldwell 2000; Schaller et al. 1995; Tzoumaka et al. 2000), are proposed to be major molecular determinants of the INaP (Smith et al. 1998). In cultured motoneurons, NaV1.6 channels predominate at the axonal initial segment (Alessandri-Haber et al. 2002). This subcellular distribution profile suggests an involvement of INaP in regulating motoneuron outputs rather than inputs. This assumption is in line with the observation in the present study that the maximal dorsal root–evoked synaptic potentials were unchanged after INaP blockade but also with studies of Kuo et al. (2006), Miles et al. (2005), and Zhong et al. (2007), demonstrating that repetitive firing during sustained intracellular depolarization is disrupted by riluzole in motoneurons. In accordance with the presence of INaP in myelinated nerve fiber (Benoit and Escande 1991), NaV1.6 is heavily expressed at nodes of Ranvier in both sensory and motor axons of the peripheral nervous system (Caldwell et al. 2000; Krzemien et al. 2000; Tzoumaka et al. 2000), suggesting a major function of this channel in regulating the excitability and the conduction velocity of peripheral nerves. This is consistent with the higher stimulation threshold to evoke dorsal root–evoked potentials and the longer latencies of both the dorsal root–evoked afferent volleys and the antidromic action potential of motoneurons after the blockade of INaP (Figs. 4 and 5). Similarly, administration of riluzole in patients with amyotrophic lateral sclerosis increases the latency of flexor reflex (Riepe et al. 1997).

Functional implications of INaP in locomotor pattern generation and future directions

It is widely accepted that the CPG for hindlimb movements is mainly located in the rostral-most segments of the neonatal rat lumbar spinal cord (Cazalets et al. 1995). Lesion experiments and functional labeling studies argue that the groups of neurons involved in the generation of the locomotor patterns are located in the ventromedial gray matter (Kjaerulff and Kiehn 1996; Kjaerulff et al. 1994). This is the area where the interneurons we recorded from were located, raising the possibility that they may be part of the hindlimb CPG. Most interneurons recorded in this region are rhythmically active during fictive locomotion (Hinckley et al. 2005; Hochman et al. 1994; Kiehn et al. 1996; Wilson et al. 2005; Zhong et al. 2006b). Moreover, the findings that “slow” interneurons send their axons to the ipsilateral motor column or to the contralateral gray matter (Szucs et al. 2003) provide further support to the assumption that interneurons recorded in this study may play an important role in the generation of locomotor patterns by acting either directly or indirectly on motoneurons. CPG neurons are composed of ipsilaterally projecting excitatory interneurones, proposed to be important for rhythm generation (Kiehn and Butt 2003), and four populations of commissural interneurons, critical for left–right coordination of hindlimb movements (Butt and Kiehn 2003). It will therefore be interesting in future experiments to determine which classes of neuronal populations of the hindlimb CPG express INaP and whether a rostrocaudal distribution of this current throughout the lumbar enlargement matches the rostrocaudal gradient in rhythmogenic ability of the spinal cord.

Disruption of fictive locomotion by riluzole suggests that INaP plays an important role in the generation of coordinated locomotor activity. Such a conclusion was recently found for the neonatal mouse (Zhong et al. 2007). It is noteworthy that a concomitant decrease in both the duration and the amplitude of motor bursts is observed before the disruption of locomotion and that the time courses of the effects of riluzole on motor bursts, locomotor pattern, and dorsal root–evoked potentials in the isolated spinal cord preparation are quite similar (see also Zhong et al. 2007). Such decrease has been reported in the neuronal circuit underlying the respiratory function in the neonatal rat (Del Negro et al. 2002a; Ramirez and Viemari 2005). We assume that a hypothetical decrease in the excitability of motoneurons cannot account for the disruption of fictive locomotor pattern. First, no significant changes in the input resistance and action potential threshold of motoneurons were observed in the presence of riluzole. Second, the riluzole-induced disruption of fictive locomotion was not prevented by increasing the network excitability with high concentrations of NMA. Third, the selective application of riluzole over the caudal-most lumbar motoneurons did not abolish the locomotor pattern despite a slight, nonsignificant decrease in the amplitude of motor bursts. Two mechanisms may account for the latter reduction. The riluzole-induced reduction in conduction velocity in motor axons may cause a dispersion of action potentials, thereby slightly decreasing the amplitude of the motor output. In addition, although riluzole did not affect the antidromic spike of motoneurons, we cannot rule out the possibility that the drug reduces their repetitive spiking over locomotor drive oscillations because INaP is stated to be important in spike initiation during slowly rising inputs (Kuo et al. 2006). The specific disruption of locomotion after the selective application of riluzole over the rostral-most lumbar segments demonstrates that INaP plays a key role in CPG operation. The mechanisms by which INaP regulates locomotion remains unclear. INaP may boost the locomotor drive potentials (Hu et al. 2002) from premotor interneurons by sustaining their repetitive firing, as recently demonstrated in ventral horn interneurons (Theiss et al. 2007; Zhong et al. 2007), or by generating plateau potentials, previously recorded in interneurons displaying locomotor-related activity (for review see Schmidt et al. 1998), and thereby shape stable locomotor activity at slow rates of locomotion (Grillner et al. 2001). This is consistent with our findings that INaP is important for stabilizing the locomotor rhythm.

The higher INaP density and input resistance, found in ventromedial gray matter interneurons, are features thought to be fundamental in the generation of bursting behavior (Del Negro et al. 2002b; Taddese and Bean 2002). The intrinsic ability of CPG interneurons to generate bursting behavior may be relevant to recruit large populations of motoneurons in synchronized activities. As we recently demonstrated, the modulation of INaP may contribute to important integrative motor functions such as mastication by regulating pacemaker-like abilities (Brocard et al. 2006b).

Modulation of INaP may represent a new tool by which the different parameters of locomotion may be finely tuned. Descending serotonergic projections shape spinal motor patterns, particularly by strengthening the locomotor-related alternations. When applied to the isolated spinal cord during NMDA-induced fictive locomotion, 5-HT improves the left–right alternations (Pearlstein et al. 2005). The left–right alternating locomotor pattern is disorganized 6 days after a neonatal spinal cord transection but recovers after the activation of 5-HT2 receptors (Norreel et al. 2003). The cellular mechanisms by which 5-HT strengthens the locomotor-related alternations remain unknown, but 5-HT has recently been shown to directly excite commissural interneurons in the neonatal mouse (Carlin et al. 2006; Zhong et al. 2006a,b). Considering that the ability of adult spinal motoneurons to activate plateaus relies on the facilitation of persistent inward currents by 5-HT (Harvey et al. 2006; Hounsgaard and Kiehn 1989; Hsiao et al. 1998) and, as for interneurons in the pre-Bötzinger complex (Pena and Ramirez 2002), 5HT2 receptors facilitate their INaP (Harvey et al. 2006), we speculate that 5-HT–induced modulation of INaP in commissural interneurons might play a key role in the dynamic reconfiguration of the locomotor network by inducing plateau potentials. Further experiments will be performed to test this hypothesis.


This work was supported by the Fondation NRJ–Institut de France, the French National Research Agency (ANR “Neurosciences, Neurologie and Psychiatrie” program), and the French Institut pour la Recherche sur la Moelle Epinière et l'Encéphale. P. Darbon was supported by the Christopher Reeve Foundation under Contract VB1-0502-2 to L. Vinay. J.-C. Viemari was supported by grants from the Ministry for Research and the Philippe Foundation.


Present address of P. Darbon: Institut des Neurosciences Cellulaires et Intégratives, UMR 7168/LC2 CNRS–Université Louis Pasteur, 21 rue René Descartes, F-67084 Strasbourg Cedex, France.


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


View Abstract