Cholinergic control of excitability of spinal motoneurones in the salamander

Slice preparation

Animals were deeply anaesthetized by immersion in a 0.1% aqueous solution of tricaine methanesulphonate (MS-222; Sigma). After evisceration, the spinal cord was exposed by a dorsal laminectomy in a dissection dish containing ice-cold control Ringer solution (mm: NaCl, 130; KCl, 2.1; CaCl₂, 2.6; MgCl₂, 0.2; Hepes, 4; glucose, 5; NaHCO₃, 1), saturated with O₂ (pH 7.4). A portion of the cord comprising the 14th–18th segments was isolated and longitudinal slices containing the ventral part of the cord, over two spinal segments, were prepared by carefully removing two-thirds of the most dorsal part of the cord using a microscalpel (Sharpoint; Perouse Implant). The slices obtained (around 150 μm thick) were kept in ice-cold oxygenated control Ringer solution during at least 1–2 h prior recording session.

A single slice was transferred to the thermoregulated recording chamber and viewed under infrared (IR) light using a CCD camera attached to an upright Olympus Optical fluorescent microscope BX51WI (Olympus France, Rungis, France) equipped with a 40× water immersion lens (LUMPlanFI/IR). The slice was continuously perfused at 2 ml min⁻¹ with control oxygenated Ringer solution (see above) and maintained at 18–19°C. Ringer solution could be exchanged with drug-containing solution within 2–3 min.

Whole-cell recordings

Whole-cell recordings from IR visualized neurones were performed in either the voltage- or current-clamp configuration using 15 MΩ electrodes containing (mm): potassium gluconate, 110; KCl, 10; MgCl₂, 1; EGTA–KOH, 1; Hepes, 10; CaCl₂, 0.1; GTPNa₃, 0.10; AMPc, 0.20; Na₂-ATP, 4; and 2% biocytin. The pH was adjusted to 7.3 with KOH and the osmolarity was adjusted to 280 mosmol l⁻¹ (i.e. 5 mosmol l⁻¹ above the osmolarity of the perfused medium) with mannitol.

The voltage- and current-clamp recordings were made with an Axoclamp-2B intracellular amplifier (Axon Instruments, Union City, CA, USA) and monitored on an oscilloscope (DSO 630, Gould, Ilford, Essex, UK). Analog signals were filtered (0–30 kHz) and sampled (5–50 kHz) with the Digidata 1322-A analog-to-digital board (Axon Instruments) and stored on hard disk for subsequent analysis. pCLAMP 9 software (Axon Instruments) was used to control the current command outputs and to acquire data. Bridge balance and input resistance were monitored throughout the experiments by means of hyperpolarizing (−20 pA; 1.5 s) current pulses. Voltages were not corrected for the liquid junction potential which was close to 0 mV with the intracellular and various extracellular solutions used in the present experiments.

Whole-cell recordings from 68 neurones were performed in stable conditions characterized by the maintenance of a resting potential more negative than −60 mV and spike amplitude > 60 mV, thereby attesting to the absence of excessive damage to the cell.

Identification of motoneurones

All the recorded cells were intracellularly labelled after recording with biocytin in the patch pipette.

In 28 animals, a retrograde tracing technique was used in order to identify motoneurones. Two to seven days before electrophysiological experiment recording, animals were deeply anaesthetized by immersion in MS-222 as described above. A solution of carbocyanin at 15% in ethanol (Fast DiI, Molecular Probes) was injected with a Hamilton syringe into the hindlimb muscles. Then, animals were kept individually in empty tanks in which the bottom was covered with a fresh wet paper until the electrophyiological experiments. In total, 32 cells, located in the ventral horn of the lumbar spinal cord, were identified as motoneurones by their fluorescence after retrograde labelling with FDiI (Russier et al. 2002) (Fig. 1).

In 19 animals, 36 biocytin-labelled cells located in the ventral horn of the spinal cord were identified as putative motoneurones by their large soma diameter (i.e. > 20 μm) and morphology (monopolar or bipolar; Fetcho, 1986). A subset (4/36) of the biocytin-labelled neurones were reliably identified as motoneurones by the additional observation of their axon exiting the ventral root. Furthermore, 36/36 biocytin-labelled neurones displayed the characteristic-firing pattern of motoneurones (i.e. repetitive firing) in response to sustained injection of depolarizing current (Russo & Hounsgaard, 1999) and no significant difference between their intrinsic electrophysiological properties and those of FDiI backfilled motoneurones could be detected.

Biocytin labelling

After recording, longitudinal slices were incubated overnight in a fixative solution (4% paraformaldehyde in 0.1 mm phosphate buffer). Biocytin was then revealed with the following fluorescence protocol: longitudinal slices were rapidly rinsed in Tris–NaCl buffer, preincubated for 45 min in Tris–Triton X-100 and incubated for 2 h at room temperature with Alexa-fluor 488-conjugated streptavidin (Molecular Probes) diluted at 1: 200 in Tris–Triton. Spinal cord slices were finally rinsed, mounting in DAKO fluorescence mounting medium and coverslipped. Staining was observed in a Zeiss Axiophot 2 fluorescence microscope (Zeiss, Jena, Germany) equipped with the appropriate filter sets.

Solutions and drugs

In all experiments, fast excitatory and inhibitory synaptic transmission were eliminated with (±)-2-amino-5-phosphonopentanoic acid (AP-5, 50 μm; Sigma), 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX, 20 μm; Tocris Cookson, Ltd, Bristol, UK) and 1 μm strychnine (Sigma). The drugs were added to the extracellular medium and continuously superfused over the preparation. Acetylcholine (ACh, 20 μm; Sigma) was freshly prepared before experiments. The cholinesterase inhibitor eserine (physostigmine, 20 μm; Sigma) was added to enhance the effect of ACh through limitation of degradation. Other drugs used were: (±)-muscarine chloride (muscarine, 20 μm; Sigma); atropine (0.1–1 μm; Sigma); nicotine (20 μm; Sigma); apamine (0.2 μm; Sigma), d-tubocurarine (10 μm; Sigma), bicuculline (20 μm; Sigma), linopirdine (20 μm; Sigma), ZD 7288 (100 μm; Tocris), caesium (1 mm) and barium (500 μm).

Data analysis

Data analysis was performed using Axograph (Axon Instruments), Microsoft Excel and SigmaPlot 8 (SPSS Inc., Chicago, IL, USA). The data were processed with the use of standard statistical analyses (SigmaStat software). Values are given as the mean ±s.e.m. and n is the number of motoneurones. Differences were considered to be significant for P < 0.05.

The passive (cell-at-rest) properties measured included resting membrane potential (in mV) and ‘steady-sate’ input resistance (in MΩ). The ‘steady-sate’ input resistance was estimated as the voltage change induced by small hyperpolarizing current pulses of 1.5 s duration applied from resting membrane potential divided by the amount of current injected.

The firing of motoneurones was characterized both by calculating the instantaneous frequency and the mean frequency during their steady state discharge evoked by a 2 s depolarizing current. For each intensity of the current pulse, the mean firing frequency was plotted against the current intensity (f–I plot). For each motoneurone, the slope of the f–I relation, which expressed the gain, was measured as the slope of the straight line fitting of the f–I plot. Unless otherwise specified, a bias current was always injected to reach the same membrane potential before and after drug application. The amplitude of the medium after-hyperpolarization (mAHP) was estimated as the difference (mV) between the threshold of the action potential and the peak of the mAHP for each action potential fired at low rate (1–2 Hz). The threshold of the action potential was visually determined from the voltage record as the membrane potential at the point of maximal change of voltage (inflection point). Muscarine-induced currents were obtained using membrane potential ramps and a subtraction procedure. I–V relations were obtained during ramps from −40 to −140 mV over 1 s. The current (I_m) was averaged from four successive responses. The conductance (G_m) was determined before and after application of muscarine at −25 mV (V_m) from the measured equilibrium potential for K⁺ (E_K) using the equation G_m=I_m/(V_m–E_K). E_K was determined at the inflection point of the I–V relationship.

Effects of acetylcholine on the excitability of spinal motoneurones

The effect of acetylcholine (ACh) was tested on 13 spinal motoneurones recorded in current-clamp mode after blockade of the fast synaptic transmission (see Methods). For each motoneurone, ACh (20 μm) was bath applied in combination with eserine (20 μm) to limit the degradation of ACh. In 9 out of the 13 recorded motoneurones, application of this medium on the spinal cord slices induced a hyperpolarization of the resting membrane potential (−6.0 ± 0.9 mV, n= 9), whereas in the remaining four cells the resting membrane potential remained unchanged.

The input resistance and the spike characteristics were monitored in a subgroup of 7 out of the 13 recorded motoneurones. ACh application induced a decrease in both the input resistance (1256 ± 183 MΩversus 1434 ± 210 MΩ, n= 7/7 cells; t test, P < 0.01) and the spike amplitude (61 ± 3 mV versus 72 ± 2 mV, n= 7/7 cells; t test, P < 0.01). By contrast, no statistically significant changes in the width of the spikes were observed (1.408 ± 0.199 s versus 1.234 ± 0.120 s, n= 7/7 cells; t test, n.s.). Unexpected from the preceding results, the excitability of 12 out of the 13 motoneurones increased (+10.7 to +80.7%), as seen by the number of spikes elicited by a depolarizing current pulse applied at the same membrane potential before and after application of ACh in combination with eserine (Fig. 2A).

In all cells, the discharge pattern, under control or cholinergic conditions, consisted of an initial burst of action potentials at high frequency (20–60 Hz) followed by a sustained discharge with a slight adaptation (Fig. 2B). However, for similar sustained discharge, the initial burst of action potentials had a higher frequency in the presence of ACh and eserine than in control conditions (Fig. 2C).

Muscarinic receptors mediate the cholinergic increase in the excitability of spinal motoneurones

The effect of bath application of muscarine (20 μm) was tested on 22 motoneurones under current clamp conditions, in the presence of blockers of fast synaptic transmission. The excitability of 19 of these cells was increased (+3.1 to +62.0%), since the same depolarizing current pulse applied at the same potential induced more action potentials (Fig. 3A). In the remaining three cells, the excitability was not affected.

The input resistance and the spike characteristics were monitored in a fraction of 8 out of the 19 motoneurones in which the excitability was increased by muscarine. Similarly with the results obtained during bath application of ACh plus eserine, the increase in excitability induced by muscarine was associated with a decrease in the input resistance (868 ± 199 MΩversus 1011 ± 226 MΩ, n= 8/8 cells; t test, P < 0.05). Furthermore, as during ACh application, the amplitude of each action potential within a train of spikes in response to a constant-current stimulus was decreased (61 ± 3 mV versus 68 ± 2 mV, n= 8/8 cells; t test, P < 0.01). However, muscarine induced a membrane hyperpolarization (−6.7 ± 1.8 mV) in only a small fraction of motoneurones (3/19), whereas the resting membrane potential of the remaining cells (16/19) was not affected.

In 14 of 19 motoneurones, in which the excitability was increased by muscarine, the firing pattern during a step depolarization was similar under control conditions and in the presence of muscarine. In both cases, the firing rate adapted from an early maximum at the onset of depolarization to a lower steady state (‘adaptating pattern’) (Fig. 3B). By contrast, in the remaining 5/19 cells, the discharge pattern in presence of muscarine consisted of an initial adaptation at the onset of the depolarizing current pulse followed by a gradual increase in frequency reaching a maximum at the end of the pulse (‘accelerating pattern’) (not illustrated). Note that, for each firing pattern, in addition to the delayed increase in firing frequency, the initial rate of firing during the pulse was also facilitated by muscarine (Fig. 3B).

In 7/7 motoneurones, bath application of the muscarinic antagonist, atropine (0.3–1 μm), blocked the effects induced by muscarine (Fig. 3A and B). After addition of atropine to the bath, the firing frequency of motoneurones decreased progressively to control values (for example, in Fig. 3B, the mean firing frequency was 13.41 ± 0.15 Hz in control conditions and 14.54 ± 0.15 Hz in the presence of atropine). The muscarinic effect on the amplitude of action potentials was also reversed by atropine (Fig. 3A). The effects of muscarine and atropine were reversible (not illustrated).

In five motoneurones, the responses to depolarizing current pulses applied at the same membrane potential, before and after application of nicotine (10 μm), were investigated. No statistically significant (t test, P= 0.99) changes in the firing rate were observed. As an example, the mean firing rate of a typical motoneurone in response to a depolarizing pulse of 60 pA was 13.12 ± 0.19 Hz in control conditions versus 13.40 ± 0.22 Hz in the presence of nicotine. Furthermore, in the five motoneurones recorded, nicotine application did not change the resting membrane potential, the input resistance (939 ± 275 MΩversus 955 ± 258 MΩ, n= 5; t test, n.s.) or the spike amplitude (81 ± 5 mV versus 79 ± 4 mV, n= 5; paired t test, n.s.).

Taken as a whole, our results indicated that the acetylcholine effects on the excitability of motoneurones were mediated by muscarinic receptors.

Muscarine enhances the input–output relationship of motoneurones

The plot in Fig. 4A illustrates the relationship between the mean firing rate and the amplitude of depolarizing currents injected in a typical motoneurone (f–I plot). The f–I relation was linear over the whole range of stimulus intensities in control conditions (y= 0.141x+ 0.529, r²= 0.99), as well as in the presence of muscarine (y= 0.229x+1.023, r²= 0.99) (Fig. 4A) or acetylcholine and eserine (not illustrated). This was observed in motoneurones which exhibited an adapting firing pattern as well as in motoneurones which exhibited an accelerating firing pattern in the presence of muscarine. Furthermore, in most cases tested (33 of 37 cells), the slope of the f–I relationship was steeper during application of muscarine, indicating an increase in the gain (0.323 ± 0.023 Hz pA⁻¹versus 0.266 ± 0.019 Hz pA⁻¹; paired t test, P < 0.001). In 9 out of 9 tested motoneurones, the gain increase induced by the cholinergic agonists was reversed with atropine (0.3–1 μm; Fig. 4B).

Muscarinic modulation of the medium after hyperpolarization (mAHP)

Spinal motoneurones exhibited action potentials followed by a complex AHP (Fig. 5A). The first phase of the AHP (fast AHP: fAHP) is a direct continuation of the repolarizing phase of the action potential. The second phase peaked 15–50 ms after the action potential and displayed a more prolonged time course (20–150 ms). This component was referred to as the medium AHP (mAHP) (Barrett et al. 1980). The two phases were separated by a depolarizing waveform, the afterdepolarization (ADP). Because the mAHP was the most frequently observed and the fAHP did not always appear, we focused on the cholinergic modulation of the mAHP. As described in other species (Goh & Pennefather, 1987; Zhang & Krnjevic, 1987; Hounsgaard et al. 1988; Seutin et al. 1997) and depicted in Fig. 5B, the mAHP of salamander motoneurones was strongly reduced (−32.3 ± 2.3%, n= 3) during application of an SK blocker (i.e. apamine, d-tubocurarine or quaternary salt of bicuculline) (Fig. 5B).

Bath application of ACh plus eserine (n= 5 cells) or muscarine (n= 8 cells) clearly decreased the mAHP amplitude, as exemplified in Fig. 5C. On average, cholinergic agonists decreased significantly (t test, P < 0.001) the mAHP from 20.8 ± 0.8 mV in control conditions to 15.4 ± 0.9 mV in the presence of cholinergic agonists (n= 9; Fig. 5D). This effect was fully reversed by atropine (20.1 ± 1.1 mV, n= 9; Fig. 5D).

Previous reports have shown that a voltage-activated K⁺ current, the M-current, can contribute to the mAHP (Storm, 1989), to the discharge adaptation (Wang & McKinnon, 1995) and to the control of the motoneurone gain (Alaburda et al. 2002). Because muscarinic receptors inhibited the M-current in several type of neurones (Adams et al. 1982; Halliwell & Adams, 1982), we tested the effects of linopirdine (20 μm), a specific blocker of the M-current (Aiken et al. 1995; Schnee & Brown, 1998). In all tested motoneurones (n= 4), we did not observe any change in the resting membrane potential and input resistance (594 ± 61 MΩversus 543 ± 48 MΩ, n= 4; t test n.s.). Also the mean firing frequency during a given current pulse applied at the same membrane potential was not significantly affected (t test, P= 0.91). As an example, the firing rate in a motoneurone was 25.69 ± 0.46 Hz in the presence of linopirdine versus 25.54 ± 0.37 Hz in control conditions, for a 140 pA depolarizing pulse applied at −60 mV. These results indicated that the M-current could not account for the muscarinic effects observed in the present experiments.

Altogether, our findings suggested that the increase in firing frequency and gain of motoneurone could partly be explained by a muscarinic inhibition of the mAHP, via a negative modulation of the SK channels.

Negative modulation of I_h

In response to hyperpolarizing current injection, 27 out of 28 motoneurones tested expressed a pronounced depolarizing sag during the pulse of current followed by a rebound depolarization at the termination of the stimulus (Fig. 6A and C). Depending on the amplitude of the hyperpolarizing current, the post-inhibitory rebound (PIR), either was subthreshold (Fig. 6A and C) or triggered an action potential (not illustrated). Furthermore, as exemplified in Fig. 6, the amplitude of the depolarizing sag (filled symbols in Fig. 6B) and that of the PIR (black bars in Fig. 6D) increased when the amplitude of the hyperpolarizing pulse current increased.

In all motoneurones tested, bath application of caesium (Cs⁺) at low concentration (1 mm) abolished or strongly reduced (−95 ± 2%, n= 6) the depolarizing sag (Fig. 6B, open symbols) and the PIR (−96 ± 3%, n= 6). At the same time, the resting membrane potential was hyperpolarized (−15 ± 1 mV, n= 7) and the input resistance was increased (1128 ± 218 MΩversus 778 ± 147 MΩ, n= 6; t test, P < 0.001). This sensitivity to low concentrations of Cs⁺ suggested the presence of an I_h current (Takahashi, 1990; Pape, 1996). Consistently, the PIR was abolished during the application of a low concentration of Cs⁺, as shown in the example in Fig. 6A. In order to further characterize the current involved in the depolarizing sag and PIR, we applied to the bath ZD 7288 (100 μm, n= 7), a selective and irreversible pharmacological blocker of I_h (BoSmith et al. 1993; Harris & Constanti, 1995; Williams et al. 1997; Hughes et al. 1998; Khakh & Henderson, 1998; Williams et al. 2002). In all cells tested, bath application of ZD 7288 strongly reduced the depolarizing sag (−81 ± 9%, n= 5) and the PIR (−88 ± 5%, n= 5; Fig. 6D). Moreover, ZD 7288, like low concentrations of Cs⁺, induced a membrane hyperpolarization (−9 ± 1 mV, n= 5) and a significant increase in the input resistance (1329 ± 166 MΩversus 928 ± 71 MΩ, n= 4; t test, P < 0.001). Altogether these results indicated that the spinal motoneurones of salamander expressed an I_h current.

Figure 7A illustrates a typical example of the responses to a square pulse of hyperpolarizing current recorded before (left) and after (middle) addition of muscarine (20 μm). During application of muscarine, both the peak voltage and the PIR of the response decreased. Figure 7A (right) further shows that in the presence of atropine (0.3 μm), the effects of muscarine were partly reversed.

Since the depolarizing sag and the PIR are voltage dependent, both increased, as the peak voltage reached during a hyperpolarizing pulse was more negative (black bars in Fig. 7B and C). Therefore, in order to determine the muscarinic effects, we compared the amplitude of the depolarizing sag and of the PIR, before and after application of muscarine, while adjusting the current amplitude to reach the same peak potential in the different pharmacological conditions (inset in Fig. 7). The bar graphs in Fig. 7B and C show that bath application of muscarine and ACh plus eserine decreased significantly the depolarizing sag and the PIR, respectively. These results indicated that muscarine inhibited the I_h current in spinal motoneurones.

Positive modulation of I_Kir

As already mentioned, activation of muscarinic receptors in salamander motoneurones caused membrane hyperpolarization and decrease in input resistance, suggesting the activation of a K⁺ current. A possible target for the cholinergic modulation could be an inward rectifier potassium current (I_Kir) (Sakmann et al. 1983; Surprenant & North, 1988). To test the hypothesis that muscarine activated a Kir conductance in motoneurones, we voltage-clamped the cells and recorded current responses to hyperpolarizing voltage ramps in the absence and in the presence of muscarine. In control conditions, the voltage ramp from −40 to −140 mV induced an inward rectifying current characterized by a steeper slope of the I–V plot at a potential more negative than −100 ± 3 mV (n= 17) (Fig. 8A), i.e. at a potential close to the K⁺ equilibrium potential (E_K predicted to be −103 mV with the intra- and extracellular solutions used) as expected for a Kir conductance. Addition of muscarine (20 μm) to the bath enhanced the inward rectification (Fig. 8A, compare black and red traces). Substraction of the two current traces yielded the I–V relationship of the muscarine-induced current (Fig. 8B), which displayed strong inward rectification and a reversal potential (around −95 mV) near the calculated E_K. The I–V relationship of the muscarine-induced current in Fig. 8B shows that muscarine enhanced the inward rectification by increasing the inward current, while exerting a relatively weak effect on the outward current. To quantify the action of muscarine on the inward rectifying current, we calculated the conductance before and after application of the drug at −25 mV with respect to the inflection point in eight motoneurones. The conductance in the presence of muscarine was significantly increased compared to the control condition (17.03 ± 4.55 nS versus 14.48 ± 3.74 nS, n= 8; Wilcoxon test, P < 0.01) (Fig. 8C). The conductance increase in the presence of muscarine was 16.73 ± 3.64% (χ² test, n= 8, P < 0.001).

To further characterize the involvement of a Kir conductance we tested for the sensitivity to a low concentration of barium (Ba²⁺). The inward current induced by muscarine was blocked at a bath concentration of 500 μm Ba²⁺ (Fig. 8A, green trace) (n= 6). This high sensitivity to Ba²⁺ is characteristic of I_Kir (Hagiwara et al. 1978; Sodickson & Bean, 1996). Altogether, these results indicated that activation of muscarinic receptors exerted a positive modulation of I_Kir.

Muscarinic modulation of the mAHP

Our results indicate that the mAHP that follows each action potential was mediated by Ca²⁺-activated K⁺ channels of the SK type since it was sensitive to apamine, d-tubocurarine and quaternary salts of bicuculline (for review see Vogalis et al. 2003). However, we cannot exclude the involvement of other K⁺ channels (Sah, 1996; Vogalis et al. 2003). Indeed, in lamprey spinal motoneurones, 20% of the sAHP (corresponding to the mAHP in the present study) are mediated by calcium-insensitive channels that might be Na⁺-activated K⁺ channels (Cangiano et al. 2002). Futhermore, in rat hippocampal neurones, several ionic currents participate in the mAHP generation (Storm, 1989). One of them, I_M, may be the main target for the blocking action by muscarinic agonists on the mAHP (Halliwell & Adams, 1982). However, since we did not evidence I_M during our experiments, this current does not contribute to the mAHP in spinal motoneurones of salamander.

We have evidenced a negative regulation of the mAHP by muscarinic agonists, like in rat hypoglossal and mouse spinal motoneurones (Lape & Nistri, 2000; Brownstone et al. 2004). This effect could result from either a direct negative modulation of the SK channels or from a negative modulation of calcium channels which in turn reduce the amplitude of the mAHP. Previous studies have reported such an indirect mechanism for the sAHP in lamprey spinal neurones (Cangiano et al. 2002) or in caudal Raphe neurones (Bayliss et al. 1994). In the present study, however, we did not observe variations in action potential duration in the presence of muscarinic agonists, thus indicating that muscarinic receptors probably modulated the SK channels directly.

Muscarinic modulation of repetitive firing

Several studies have shown the role of SK channels and of the I_M current in controlling spike frequency adaptation in Vertebrate motoneurones (Russo & Hounsgaard, 1999; Alaburda et al. 2002). Since in the present study I_M current was not detected, it is likely that firing adaptation was mainly controlled by SK channels. Furthermore, the decrease in the AHP amplitude observed in the present study probably accounted for the observed muscarine-induced increase in firing frequency.

In a few salamander motoneurones (5/22), muscarine promoted an acceleration, rather than an adaptation, of their discharge with constant-current stimulus. This behaviour, which is also present in turtle motoneurones (Svirskis & Hounsgaard, 1997; Hornby et al. 2002a), has been related to the dynamic properties underlying the generation of plateau potentials (Russo & Hounsgaard, 1999) for review). More experiments are necessary to further substantiate this idea.

In vitro experiments showed that suppression of the AHP by apamine or reduction of the AHP amplitude by 5-HT (Hounsgaard & Kiehn, 1989) induced an increase in the gain of motoneurones. A recent dynamic clamp study on cat lumbar motoneurones has demonstrated that the gain of motoneurones is inversely proportional to the AHP conductance and to the AHP time constant (Manuel et al. 2004). Therefore, it is likely that the muscarinic control of the input–output relationship in spinal motoneurones in salamander is mediated by modulation of the mAHP.

The doublet of action potentials recorded at the beginning of the response to a depolarizing current pulse reached a higher frequency in the presence of cholinergic agonists. This effect can contribute to the production of a marked and persistent enhancement of the force output. Indeed, insertion of a short, high-frequency burst into an otherwise low-frequency stimulus train can produce remarkably long-lasting enhancement of isometric force output in non-twitch muscles of arthropods (Blaschko et al. 1931; Wilson & Larimer, 1968). This ‘catch property’ was also evidenced in mammalian motor units where the output force is modulated primarily by the pattern and not by the average frequency of the input train (Burke, 1981).

Our data further evidenced a decrease in amplitude of the spikes during the activation of the muscarinic receptors. One possible explanation is that muscarine decreased the spike amplitude simply because it induced a faster firing frequency. This faster frequency resulted in a more depolarized membrane potential between spikes (see, e.g. Fig. 3A). As a result, there was probably less relief from Na⁺ channel inactivation, thereby decreasing spike height. However, if this explanation is correct, the first spike in the train should have a greater height than the subsequent spikes. Since this was not the case, a direct muscarinic effect on Na⁺ channels is probably involved.

Muscarinic modulation of inward rectifying currents

Our observation that application of Cs⁺ at low concentration or of ZD 7288 induced a hyperpolarization of the membrane potential suggests that I_h was tonically activated at rest and that it therefore contributed to the resting membrane potential in salamander motoneurones. The contribution of I_h to the resting membrane potential of spinal motoneurones has previously been evidenced in the newborn rat (Kjaerulff & Kiehn, 2001).

In addition to its inhibitory action on I_h, we have shown that activation of muscarinic receptors enhanced inward rectification of the I–V plot in motoneurones, via the activation of Kir channels. A likely candidate is the G-protein-activated inwardly rectifying K⁺ channel (Kir3) (Pfaffinger et al. 1985; Dascal, 1997). Spinal motoneurones from newborn rats (Takahashi, 1990; Kjaerulff & Kiehn, 2001) possess I_Kir too.

In control conditions, motoneurones displayed an inward rectifying current at hyperpolarizing potentials around E_K, indicating the expression of an inward rectifying K⁺ current, I_Kir, in salamander motoneurones. This raises the possibility that Kir3 channels displayed some degree of spontaneous background activity. This background activity could be due to a constitutive Kir3 channel activity in the absence of G protein activation (Takigawa & Alzheimer, 1999; Chen & Johnston, 2005). However, we cannot exclude the presence of ambient modulators of Kir3 channels such as ACh, GABA (Derjean et al. 2003) or adenosine, which could play a role in keeping these channels spontaneously opened (Takigawa & Alzheimer, 2002). On the other hand, the salamander motoneurones could express the Ba²⁺-sensitive Kir2 channels (Topert et al. 1998), which are constitutively active and contribute to the resting K⁺ conductance in many cells (Isomoto et al. 1997). However, the fact that in our study the bath application of Ba²⁺ did not change the resting potential of the cell rules out this possibility.

It is well known that activation of M₂ muscarinic receptors leads to an increase of a Kir3 conductance in mammal atrial myocytes (Sakmann et al. 1983; Surprenant & North, 1988) via Gβγ subunits (Pfaffinger et al. 1985). In the central nervous system, evidence for a similar muscarinic action via M₂ receptors has been reported from thalamic reticular neurones (McCormick & Prince, 1986) and from interneurones of striatum (Calabresi et al. 1998), neocortex (Xiang et al. 1998) and hippocampus (McQuiston & Madison, 1999). In the present study, the type of muscarinic receptors involved in the enhancement of the Kir conductance remains to be determined.

A dynamic clamp study in newborn rats suggests that the main action of I_h during an NMDA-induced locomotor activity is to increase the firing probability in motoneurones via a tonic depolarization (Kiehn et al. 2000). Several lines of evidence suggest an involvement of ACh in the modulation of ongoing locomotor activity in vertebrates (Panchin et al. 1991; Fok & Stein, 2002; Quinlan et al. 2004). Therefore, the negative regulation of I_h by the activation of muscarinic receptors evidenced in the present study will lead to a decrease in the firing probability of spinal motoneurone excitability during ongoing locomotor activity.

Our study reveals a mechanism that may amplify and prolong hyperpolarizing events. Because the activation and deactivation kinetics of Kir3 current are much faster than those of I_h, muscarine-induced Kir3 current can rapidly hyperpolarize the neurone in response to hyperpolarizing events. The inhibition of I_h by muscarine can then delay the recovery from hyperpolarizations and thus prolonged their duration. The relatively weak facilitation of I_Kir by muscarine in the range of potentials more positive than E_K suggests a minor role of muscarine modulation of I_Kir on motoneurone firing. Interestingly, a similar conclusion has previously been drawn for 5-HT modulation of I_Kir in neonatal rat motoneurones (Kjaerulff & Kiehn, 2001).

In conclusion, the present study reveals that muscarine induces two opposite effects on the excitability of salamander motoneurones. Indeed, the motoneuronal excitability could be increased by an inhibition of the mAHP while it could be decreased by an inhibition of I_h and an enhancement of I_Kir. Because these two effects are voltage dependent, muscarinic modulation can strengthen the contrast of synaptic effects by minimizing low activations and enhancing strong activations (Seeger & Alzheimer, 2001).