Interplay between neuromodulator-induced switching of short-term plasticity at sensorimotor synapses in the neonatal rat spinal cord

Abstract

In the present study, we investigated the modulation of short-term depression (STD) at synapses between sensory afferents and rat motoneurons by serotonin, dopamine and noradrenaline. STD was elicited with trains of 15 stimuli at 1, 5 and 10 Hz and investigated using whole-cell voltage-clamp recordings from identified motoneurons in the neonatal rat spinal cord in vitro. STD was differentially modulated by the amines. Dopamine was effective at all stimulation frequencies, whereas serotonin affected STD only during 5 and 10 Hz stimulus trains and noradrenaline during 1 and 5 Hz trains. Dopamine and serotonin homogenized the degree of depression observed with the different stimulation modalities, in contrast to noradrenaline, which amplified the rate differences. The different modulatory profiles observed with the amines were partly due to GABAergic interneuron activity. In the presence of GABA_A and GABA_B receptor antagonists, the rate and/or kinetics of STD did not vary with the stimulation frequency in contrast to the control condition, and noradrenaline failed to alter either synaptic amplitude or STD, suggesting indirect actions. Dopamine and serotonin strongly decreased STD and converted depression to facilitation at 5 and 10 Hz during the blockade of the GABAergic receptors in 50% of the neurons tested. Altogether, these results show that STD expressed at sensorimotor synapses in the neonatal rat not only is a function of the frequency of afferent firing but also closely depends on the neuromodulatory state of these connections, with a major contribution from GABAergic transmission.

Synaptic transmission varies according to the previous activity of the synapses. This change in synaptic efficacy is called activity-dependent plasticity (ADP) and results in either an increase (potentiation) or a decrease (depression) in connection strength between neurons. ADP in turn can be modulated by previous synaptic activity or by neuromodulatory inputs. This second degree of plasticity has been termed metaplasticity (Abraham & Tate, 1997; Fischer et al. 1997). ADP and metaplasticity are now regarded as essential for normal information processing in neuronal networks (for review see Abraham & Tate, 1997; Abbott & Regehr, 2004).

Short-term depression (STD) has been shown to bestow neuronal networks with functional capabilities such as filtering and gain control to increase their sensitivity to sudden changes (Abbott & Regehr, 2004). To date much attention has focused on understanding the cellular basis underlying STD (Thomson, 2000; Zucker & Regehr, 2002), although little is known about its neuromodulation. Recent studies, however, hypothesized that dysfunctioning of STD modulation could be of particular relevance in pathophysiology, as in Parkinson's disease (Tecuapetla et al. 2007) and epilepsy (Doherty & Dingledine, 2001).

ADP and metaplasticity have long been considered to be restricted to supraspinal structures where they were thought to be the basis of learning and memory (Abraham & Tate, 1997; Bruel-Jungerman et al. 2007). In contrast, the spinal cord was rather seen as a non-plastic structure whose single flexibility was manifest in the modulation of motor reflexes by descending pathways. In fact, plastic phenomena occur in the spinal circuits that control posture and locomotion in the healthy and injured spinal cord (Wolpaw & Carp, 1990; Rossignol, 2000; Wolpaw & Tennissen, 2001; Edgerton et al. 2004; Ding et al. 2005). In the in vitro spinal cord of neonatal rats, glutamatergic synapses between sensory afferents and lumbar motoneurons express STD when the afferents are repetitively stimulated at time intervals of less than 30 s (Lev-Tov & Pinco, 1992; Bertrand et al. 2000). Although STD has been well described, no data are available regarding the metaplasticity of these sensorimotor connections. The sensory afferent terminals are, however, submitted to numerous neuromodulatory influences coming from both the spinal interneurons and the supraspinal pathways (for review see Rudomin & Schmidt, 1999; Mendell et al. 2001). The description of ADP neuromodulation at sensorimotor connections therefore appears to be of particular relevance to further understand the input–output relationships in motoneurons, as well as the profound reorganization of sensory afferent input processing following the loss of descending pathways subsequent to spinal cord injury (Frigon & Rossignol, 2006).

Since several motor disabilities like restless leg syndrome and spasticity are thought to be due to perturbations of the aminergic control of sensory afferent inflow (Barriere et al. 2005b), we analysed the effects on STD of three neuromodulators that control sensorimotor transmission in neonatal rodents: serotonin (Crick & Wallis, 1991; Yomono et al. 1992; Schmidt & Jordan, 2000), dopamine (Clemens & Hochman, 2004), and noradrenaline (Kitazawa et al. 1985). We also characterized the involvement of the GABAergic system in these modulations. We report that STD could be differentially altered and even converted to short-term facilitation depending on the neuromodulatory environment of sensorimotor synapses.

Methods

Dissection

Experiments were performed on isolated spinal cord preparations (n = 68) obtained from newborn Wistar rats aged 1–4 days bred in our laboratory. All experiments were carried out in accordance with the guidelines of the Institutional Animal Care and Use Committee of Bordeaux 2 University (agreement number AP 2/5/2006). The animals were chilled by hypothermia until reflexes were lost, then decapitated. A laminectomy was performed to expose and remove the spinal cord. The spinal cord was cut at the thoracic level and placed in a recording chamber. The preparation was superfused (3 ml min⁻¹) with oxygenated (95% O₂–5% CO₂) saline containing (mm): NaCl 130, KCl 3, CaCl₂ 3.75, MgSO₄ 1.3, NaHPO₄ 0.58, NaHCO₃ 25, and glucose 10, adjusted to pH 7.4 with HCl. All experiments were performed at room temperature (25°C).

Extracellular stimulation

Extracellular stainless steel pin electrodes placed in contact with dorsal or ventral roots and insulated with Vaseline were used for stimulations (Sqalli-Houssaini et al. 1991).

Whole-cell recordings

To gain access to the motoneurons, the pia was removed, then whole-cell patch-clamp recordings were made using glass microelectrodes (20–30 mΩ) filled with a solution containing (mm): potassium gluconate 120, KCl 20, CaCl₂ 0.1, MgCl₂ 0.1, EGTA 1, Hepes 10, GTP 0.1, cAMP 0.2, leupeptin 0.1, Na₂-ATP 3, and d-mannitol 77, pH 7.3. Motoneurons were identified by antidromic action potentials in response to ventral root stimulation. Whole-cell currents were amplified using a RK400 amplifier (Biologic), digitized with an Instrutech ITC-16 interface (Great Neck, NY, USA) and stored on an Apple computer. The membrane potential of the motoneurons was held at −60 mV and monosynaptic excitatory postsynaptic currents (EPSCs) were pharmacologically isolated throughout the experiments by adding the myorelaxant mephenesin (1 mm) (Berry & Pentreath, 1976; Pinco & Lev-Tov, 1993a) and strychnine (1 μm) to the extracellular saline. Trains of 15 successive stimuli (500 μs) were applied to the ipsilateral dorsal root at three frequencies: 1, 5 and 10 Hz with increasing voltages (range for all neurons tested: 0.26–4 V) to evoke monosynaptic EPSCs in the motoneurons (Lev-Tov & Pinco, 1992). The trains were repeated at 30 s intervals to obtain a full recovery of the first EPSC amplitude (Lev-Tov & Pinco, 1992; Bertrand et al. 2000). Responses were fully reproducible from trial to trial at a given stimulation voltage (not shown). Series resistance was monitored throughout the experiments and was not compensated.

Drugs

Fresh stock solutions of noradrenaline and dopamine were prepared on the day of each experiment and were protected from light exposure. Dopamine was coapplied with the antioxidant sodium metabisulfite at a final concentration of 5 μm (Barriere et al. 2004). All other drugs were prepared as stock solutions, aliquoted and frozen until use. To avoid activation of the central pattern generator (CPG) for locomotion, dopamine, serotonin and noradrenaline were tested at 50 μm, 5 μm and 5 μm, respectively, which are subthreshold concentrations for generating fictive locomotion (Sqalli-Houssaini et al. 1993; Kiehn et al. 1999; Sqalli-Houssaini & Cazalets, 2000; Barriere et al. 2005a). All reagents were obtained from Sigma (St Louis, MO, USA) except dl-AP5, SR95531 hydrobromide (gabazine), CGP55845 and cyclothiazide, which were from Tocris Cookson (Ellisville, MO, USA).

Data analysis

Recordings were analysed using the AxoGraph software program (AxoGraph Scientific, Sydney, Australia, axographx.com). The peak amplitude of each EPSC in the train was measured from the maximum peak to the baseline just before each individual EPSC. The responses were measured in defined time windows based on the frequency of stimulation and designed to eliminate the stimulus artifact: 0.99 s, 0.19 s and 0.09 s for 1, 5 and 10 Hz, respectively. To calculate the amount of STD, the amplitude of the 15th EPSC was normalized to the amplitude of the first EPSC in the train according to the formula: ((EPSC₁₅/EPSC₁) × 100) − 100.

Statistical analysis

Statistical analyses were performed on raw data. The data were checked for normal distribution. Student's t test was used to compare two series of data. An analysis of variance (ANOVA) was carried out to test for significant effects between the different frequencies of stimulation and/or drugs. Pairwise comparisons were performed using Tukey's test. The level of significance was set at P ≤ 0.05. Mean values are given ±s.e.m.

Results

The results of this study are based on recordings from 65 motoneurons (17 from the lumbar segment 4 (L4) and 48 from L5). The mean resting membrane potential and the mean input resistance were −64.1 ± 0.5 mV and 184.6 ± 12 MΩ (n = 47), respectively, in normal saline.

STD at sensorimotor synapses

STD at synapses between sensory afferents and lumbar motoneurons has been previously described by Lev-Tov & Pinco (1992). In the present study, trains of 15 stimuli were applied to dorsal roots at three different frequencies: 1, 5 and 10 Hz based on our previous study (Bertrand et al. 2000). This protocol induced a pronounced, gradual and reproducible decrease in EPSCs in the ipsilateral lumbar motoneurons in the same segment (Fig. 1A). Frequency-dependent STD was expressed in two different ways. EPSC amplitude was plotted as a function of time (Fig. 1Ba) or as a function of stimulus number in the train (Fig. 1Bb). EPSC amplitude decreased significantly during the trains at all stimulation frequencies used. The average EPSC amplitude normalized to the first EPSC was plotted as a function of the stimulus number for the 47 neurons tested (Fig. 1C). When the dorsal root was stimulated at 1 Hz, the EPSCs were depressed and immediately reached a steady state after the second stimulus (Fig. 1C). The amount of synaptic depression increased with the stimulation frequency (inset in Fig. 1C) and the EPSC amplitude reached a steady state only after five or six stimuli (Fig. 1C). EPSCs were depressed on average by 44 ± 4%, 67 ± 4% and 71 ± 4%, n = 47, during 1, 5 and 10 Hz stimulation trains, respectively (inset in Fig. 1C).

Figure 1. Frequency-dependent short-term synaptic depression of sensory afferent to ipsilateral motoneuron synapses.

A, representative excitatory postsynaptic currents (EPSCs) recorded from a motoneuron held at −60 mV (V_hold = −60 mV) show significant depression in response to trains of stimuli applied to the ipsilateral dorsal root at increasing frequencies. B, EPSC amplitude plotted versus time (a) and versus stimulus number (b) in the train. C, graph of mean EPSC values as a function of the stimulation frequency (n = 47 neurons). EPSC amplitude was normalized against EPSC₁ at each frequency. The inset shows the mean percentage of depression for the three frequencies of stimulation used. *P ≤ 0.05; ns: non-significant.

It has been shown that both the sign of short-term plasticity, i.e. whether depression or facilitation occurs, and the degree of plasticity depend on the amplitude of the first response (Debanne et al. 1996; Jiang et al. 2000; Bertrand & Lacaille, 2001). To investigate how the depression varied with the amplitude of the first EPSC (EPSC₁), we varied the stimulation voltage. The percentage of depression expressed as a function of the amplitude of EPSC₁ was plotted for all the neurons tested (n = 47, at least 157 values were computed; data not shown). There was a small but significant positive correlation (r = 0.21) between the amplitude of the first EPSC and the percentage of depression for a 1 Hz stimulation (data not shown). In contrast, no significant correlation was observed between the amplitude of EPSC₁ and the amount of depression at 5 and 10 Hz (data not shown).

Several distinct mechanisms, either alone or in combination, may cause short-term synaptic depression. These include presynaptic depletion of releasable vesicles, other presynaptic mechanisms that decrease vesicle release, and postsynaptic receptor desensitization or saturation. Because glutamate receptors exhibit pronounced desensitization during high frequency bursts (Jones & Westbrook, 1996), desensitization may contribute to STD of neonatal rat sensorimotor synapses. To investigate a putative postsynaptic locus for depression, we therefore tested the effects of cyclothiazide, which blocks the desensitization of AMPA receptors (Wong et al. 2003). At 50 μm, cyclothiazide significantly and irreversibly increased the peak amplitude of EPSC₁ in the neonatal rat spinal cord (Fig. 2Aa and b) and significantly slowed the decay of EPSCs measured from monoexponential fits starting from the region where EPSCs began to decay to the baseline (Fig. 2Aa–c). In contrast, cyclothiazide modified neither the percentage (Fig. 2Ba) nor the kinetics (Fig. 2Bb, data not shown for 5 and 10 Hz) of the STD, whatever the stimulation frequency used. These results are in agreement with those of Wan et al. (2006), who reported similar effects of cyclothiazide in the dorsal part of the cord on the synapses between primary afferents and neurons of lamina III–V. Cyclothiazide have been reported to affect the neurotransmitter release probability (Bellingham & Walmsley, 1999; Wall, 2005). The lack of effect of cyclothiazide on STD reported here suggests, however, that such non-specific presynaptic actions do not occur at sensorimotor synapses in the neonatal rat spinal cord.

Figure 2. Effects of cyclothiazide on both EPSCs and short-term depression.

Aa, superimposed EPSCs in control condition, in the presence of cyclothiazide and after a 35 min wash-out of the drug. The continuous lines represent the monoexponential fits used to calculate the decay time constant. Cyclothiazide significantly and irreversibly increased the amplitude (Ab) and the decay time constant (Ac) of the first EPSC but failed to affect both STD levels (Ba) and the time course of depression (Bb). *P ≤ 0.05; ns: not significant. n = 5 motoneurons.

Effects of exogenous amines on sensorimotor transmission

We then examined the effects of dopamine, serotonin and noradrenaline on both motoneurons and glutamatergic transmission between sensory afferents. As previously reported in other studies, all three amines depolarized the motoneurons and inhibited the sensorimotor synapses (Kitazawa et al. 1985; Crick & Wallis, 1991; Yomono et al. 1992; Schmidt & Jordan, 2000; Gordon & Whelan, 2006). In our experimental conditions, when bath-applied onto the spinal cord, all three amines induced a stable inward current in motoneurons held at −60 mV (shown for noradrenaline in Fig. 5A) that persisted throughout drug application (mean inward current: 50 ± 7 pA (n = 15) for dopamine, 31 ± 4 pA (n = 12) for serotonin and 45 ± 5 pA (n = 17) for noradrenaline at the concentration tested; see Methods). This inward current was associated with a significant increase in the input resistance in lumbar motoneurons: 156 ± 18 MΩ in control, 200 ± 24 MΩ with dopamine (n = 15); 211 ± 24 in control, 229 ± 26 MΩ with serotonin (n = 14); and 188 ± 19 in control, 219 ± 21 MΩ with noradrenaline (n = 17). The action of the aminergic drugs was then tested on sensorimotor glutamatergic transmission. The three neuromodulators significantly reduced the amplitude of the EPSC₁ (trace panels in Figs 3–5) by 74.3 ± 3% (n = 14) with dopamine; 71.1 ± 5% (n = 14) with serotonin and 23.2 ± 6% (n = 12) in the presence of noradrenaline. We computed the paired-pulse ratio (EPSC₂/EPSC₁) in control conditions and in the presence of the aminergic compounds, since an increase in the paired-pulse ratio suggests that there is a presynaptic action of the drug tested (for example see Kamiya & Ozawa, 1998). Table 1 summarizes the ratio obtained for stimulations at 1, 5 and 10 Hz. At all frequencies tested, the aminergic drugs significantly increased or showed a tendency to enhance the paired-pulse ratio.

Figure 5. Effects of noradrenaline on short-term depression.

A, representative traces of EPSCs recorded from lumbar motoneurons during trains of 15 stimuli applied to the dorsal root at 1, 5 and 10 Hz in control conditions (left panel in Aa–c), when noradrenaline (5 μm) was added to the bathing medium (right panel in Aa–c). B, plots of the averaged and normalized EPSC amplitude over stimulus number at 1 (a), 5 (b) and 10 Hz (c) in the absence (circles) or the presence of noradrenaline (squares). C, mean percentage of depression during stimulation at the three frequencies tested before and in the presence of noradrenaline. No effect of noradrenaline was observed during 10 Hz stimulation. The numbers in each bar indicate the number of cells tested in each condition. *P ≤ 0.05, ns: not significant compared to control conditions; *P ≤ 0.05, NS: not significant between the different stimulation frequencies. n = 12 neurons.

Figure 3. Effects of dopamine on short-term depression.

A–C, representative recordings showing the effects of dopamine (50 μm) on EPSCs evoked by repetitive stimulation of the dorsal root at 1 Hz (A), 5 Hz (B) and 10 Hz (C). D, bar graphs of the mean percentage of depression for the three stimulation frequencies tested (1 Hz filled bars, 5 Hz grey bars and 10 Hz open bars) in control conditions, in the presence of dopamine, when the stimulation was increased to evoke an EPSC₁ close to the predopamine amplitude (DA + increased stimulation) and after wash-out of the drug (1 h). The numbers in each bar indicate the number of cells tested in each condition. *P ≤ 0.05; ns: not significant compared to control conditions, *P ≤ 0.05; NS: not significant between the different stimulation modalities. E, graphs of the relative amplitude of the EPSC over the stimulus number in control conditions (circles), during the bath-application of dopamine (DA, triangles; DA + increased stimulation, squares) at 1 Hz (a), 5 Hz (b) and 10 Hz (c) stimulation trains.

Table 1.

Mean paired-pulse depression ratio (EPSC₂/EPSC₁) during 1, 5 and 10 Hz stimulation in control conditions and in the presence of the different amines

	1 Hz	5 Hz	10 Hz
Control	0.5 ± 0.06	0.7 ± 0.08	0.7 ± 0.1
Dopamine	0.9 ± 0.06*	0.8 ± 0.06	0.9 ± 0.07
n	14	11	10
Control	0.6 ± 0.03	0.5 ± 0.05	0.5 ± 0.05
5-HT	0.7 ± 0.05*	0.8 ± 0.08*	0.9 ± 0.1*
n	14	13	13
Control	0.7 ± 0.04	0.6 ± 0.04	0.4 ± 0.05
Noradrenaline	0.75 ± 0.05	0.9 ± 0.05*	0.8 ± 0.05*
n	12	12	12

^*P ≤ 0.05.

Effects of amines on STD

Figure 3A–C illustrates the action of dopamine on STD for the three stimulation frequencies tested in a representative motoneuron (1 Hz: Fig. 3Aa and b, 5 Hz: Fig. 3Ba and b, 10 Hz: Fig. 3Ca and b). Bath-application of dopamine dramatically depressed the amplitude of EPSC₁ and significantly decreased the EPSC depression reached during the trains at all frequencies tested (Fig. 3D, small stars). The percentage of depression was as follows: at 1 Hz: 60 ± 3% in control, 26 ± 7% with dopamine (n = 13); at 5 Hz: 74 ± 3% in control, 36 ± 10% with dopamine (n = 12); at 10 Hz: 75 ± 4% in control, 41 ± 5% with dopamine (n = 12) (Fig. 3D). This effect of dopamine was partly reversed during a 1 h wash-out with normal saline (Fig. 3D). In contrast to control conditions where the amount of depression was frequency dependent, EPSC depression in the presence of dopamine always reached the same level irrespective of the frequency of stimulation. There was no significant difference in the degree of depression in the presence of dopamine obtained with 1, 5 and 10 Hz trains (large stars versus NS in Fig. 3D). Before and during the bath-application of dopamine, the global shape of the STD time course was similar for the three frequencies of stimulation tested (Fig. 3Ea–c). A question that arose was whether the change in EPSC depression seen in the presence of dopamine is due to the decrease in EPSC₁ amplitude. We therefore increased the stimulation voltage in the presence of dopamine so that the amplitude of EPSC₁ was similar to the control amplitude (Fig. 3Ac, Bc and Cc). The depression still remained less pronounced and independent of stimulation frequency in the presence of dopamine (Fig. 3D). Furthermore, the shape of the depression time course was not altered compared to control conditions (Fig. 3E).

Serotonin, like dopamine, also decreased the EPSC₁ amplitude but the statistical analysis revealed only a tendency for this drug to affect EPSC depression during 1 Hz trains (Fig. 4A and D). This lack of significant effect persisted even when the amplitude of EPSC₁ was restored to the control value by increasing the stimulus voltage in the presence of serotonin (Fig. 4Ac and D). At 5 and 10 Hz, however, serotonin strongly reduced the degree of depression. Serotonin flattened the depression level reached for these two frequencies for both reduced EPSC₁ (Fig. 4Bb–Cb and D) and EPSC₁ amplitudes restored to control values (Fig. 4Bc–Cc and D: serotonin + increased stimulation). The level of STD was clamped to around 40% in the presence of serotonin, whatever the frequency of stimulation applied to the dorsal roots. This value was comparable to that obtained in the presence of dopamine (Fig. 3D). This action of serotonin was reversible (Fig. 4D). The global shape of STD time course was not altered in the presence of serotonin compared to control condition (Fig. 4E).

Figure 4. Effects of serotonin on short-term depression.

A–C, representative EPSCs recorded from one motoneuron during repetitive stimulation of the sensory afferents at 1 (A), 5 (B) and 10 Hz (C) in the absence (Aa, Ba and Ca) and in the presence of serotonin (5 μm; Ab–Cc). D, bar graphs of the mean percentage of depression reached in the different conditions (wash-out: 1 h). As serotonin decreased the amplitude of the first EPSC in the train, the voltage of stimulation was boosted to recover an EPSC₁ amplitude similar to that obtained in control conditions (Serotonin + increased stimulation). Note that serotonin failed to alter the depression rate during 1 Hz stimulation. The numbers in each bar indicate the number of cells tested in each condition. *P ≤ 0.05, ns: not significant compared to control conditions; E, graphs of the relative amplitude of the EPSC over the stimulus number in control conditions (circles), during the bath-application of serotonine (5HT, stars; 5HT + increased stimulation, squares) at 1 Hz (a), 5 Hz (b) and 10 Hz (c) stimulation trains. *P ≤ 0.05, NS: not significant between the different stimulation frequencies.

Next, we analysed the effects of noradrenaline on STD (Fig. 5). Although noradrenaline had a weaker effect than the other two amines on sensory afferent synapses when tested at subthreshold concentrations for fictive locomotion generation (see below), it significantly altered the degree of EPSC depression observed with 1 and 5 Hz stimulation trains (Fig. 5Aa and b and C) but failed to affect the STD obtained during 10 Hz stimulations (Fig. 5Ac, Bc and C). In contrast to dopamine and serotonin, which flattened the percentage of depression reached, different STD levels were significantly reached for the three frequencies tested in the presence of noradrenaline (Fig. 5C). Although the global dynamics of STD appeared unchanged during 5 Hz stimulation in the presence of the drug (Fig. 5Bb), the plots of normalized EPSCs over stimulus number for 1 Hz stimulation exhibited a sag in the presence of noradrenaline that corresponded to a tendency for EPSC amplitude values to return to the EPSC₁ value (Fig. 5Ba).

Altogether these results indicate that dopamine, serotonin and noradrenaline, in addition to inhibiting the segmental afferent inputs onto motoneurons, differentially modulate the STD expressed at these synapses.

An inward current induced by serotonin and noradrenaline

Inspection of the above revealed that responses to repetitive dorsal root stimulation in the presence of noradrenaline and serotonin showed signs of a concomitant inward current that outlasted the sensory afferent stimulation (dashed lines in Figs 4A–C and 5A). To determine whether the expression of this current was dependent on the stimulation frequency and the presence of amine, the amplitude of this current was correlated with the amplitude of EPSC₁ (Fig. 6). The amplitude itself was determined as the difference between the baseline prior to stimulation and the current 100 ms after the last stimulus (arrows in Fig. 6A). Stimulation trains were evoked at different voltage amplitudes (as illustrated in the example in Fig. 6Aa and b) and frequencies (Fig. 6B–D). Since the amplitude of EPSC₁ could vary according to the stimulus voltage, we combined the data into 40 pA bins (mean number of EPSC amplitudes per bin: 9.8 ± 0.8). This analysis revealed that the inward current was also present in control conditions (filled circles in Fig. 6B–D) and that its amplitude was positively correlated with the amplitude of EPSC₁ (Fig. 6B–D). A one-way ANOVA showed that the relationship between EPSC₁ amplitude and the amplitude of the inward current was independent of the stimulation frequency in the control conditions and reached a value of about 10 pA for the largest EPSC1 amplitudes (filled circles in Fig. 6B–D). In a second step, we found that this inward current was differentially modulated by the three amines. Dopamine and serotonin were so powerful in depressing the sensorimotor synapses that it was even impossible to reach the highest EPSCs amplitudes recorded in the control conditions. This explains the lack of values in Fig. 6B–D for the largest EPSC₁ amplitude bins (220–260 pA). The overlapping black and grey circles in Fig. 6B shows that dopamine did not significantly modulate the inward current (see also Fig. 3) at any of the frequencies tested. In the presence of serotonin, however, the current increased during 5 and 10 Hz stimulation but the effect did not reach statistical significance (Fig. 6C). In contrast, noradrenaline significantly amplified this current during 5 and 10 Hz stimulation trains (Fig. 6A and D).

Figure 6. Expression of an AP5-sensitive inward current during stimulation trains and its modulation by amines.

A, examples of the development of an inward current during 5 Hz stimulation of the dorsal roots in the presence of noradrenaline at two stimulation voltages. The current amplitude was calculated as the difference between the baseline prior to stimulation and the current 100 ms after the last stimulus in the train (arrows on Aa and b). B–D, plots of the inward current as a function of the first EPSC amplitude in control conditions (filled circles), in the presence of dopamine (50 μm; B, grey circles), serotonin (5 μm; C, 5-HT, open squares) and noradrenaline (5 μm; D, NA, open circles) during 1 (Ba, Ca and Da), 5 (Bb, Cb, Db) and 10 Hz (Bc, Cc, Dc) stimulation trains. Superfusion of the NMDA antagonist dl-AP5 (AP5, 50 μm) reversibly decreases both the amplitude of the first EPSC and the inward current evoked during the train (Ea, Eb and Ed). The inward current was still strongly depressed when the stimulation was increased to restore the amplitude of the first EPSC (Ec).

We further investigated the pharmacology of this inward current under optimal conditions for its activation, i.e. in the presence of noradrenaline using 5 Hz trains (Fig. 6E). Under these conditions, all the motoneurons tested developed an inward current (n = 6; Fig. 6Ea). Bath-application of the NMDA receptor antagonist dl-AP5 (50 μm) reversibly decreased both the inward current and the EPSC₁ amplitude (Fig. 6Eb–d). The dl-AP5-induced decrease of the inward current also remained when the stimulus voltage was increased to evoke an EPSC₁ with an amplitude similar to the control value (Fig. 6Ec). Therefore, these data strongly suggest that the slow inward current that develops during stimulation trains and is amplified by noradrenaline and serotonin is mediated by NMDA receptors.

Contribution of GABAergic interneurons to STD and to its aminergic modulation

The previous experiments were conducted in the presence of strychnine to suppress glycinergic transmission and mephenesin to eliminate the main polysynaptic pathways activated by dorsal root stimulations. In these experimental conditions, however, we could not exclude that some local circuits involving a few synapses might remain active. It is well documented that primary afferent terminals are under the strong control of spinal GABAergic interneurons that presynaptically filter the sensory volley entering the spinal cord via the activation of GABA_A and GABA_B receptors (for example see Peshori et al. 1998; Rudomin & Schmidt, 1999). When bath-applied on the whole spinal cord, dopamine, serotonin and noradrenaline would all be expected to produce spiking in GABAergic interneurons, which would in turn impact onto sensorimotor transmission. A differential activation of these GABAergic neurons by the amines could therefore take part in the differential modulation of STD reported here. To determine the possible contribution of the GABAergic transmission in STD and its aminergic modulation, we therefore tested the effects of the three amines in the presence of blockers of GABA_A and GABA_B receptors: SR95531 hydrobromide (gabazine) and CGP55845, respectively. We first checked whether CGP55845 and gabazine used at a 1 μm concentration efficiently blocked GABA receptors in our preparation. In the neonatal rat spinal cord, bath-application of GABA has been shown to slow down pharmacologically induced motor activity in a dose-dependent manner, and to modulate the information conveyed by the thoracic 13 (T13) and L2 network to its target motoneurons via GABA_A and GABA_B receptor activation (Cazalets et al. 1994; Bertrand & Cazalets, 1999). In the present study, locomotor activity was induced by using a mixture of 1.8 μm N-methyl-dl-aspartic acid (NMA) and 1.8 μm serotonin (Sqalli-Houssaini et al. 1993) and recorded extracellularly from the right and left L2 ventral roots in three spinal cord preparations (data not shown). This activity consisted of alternating bursts of action potentials between the right and left part of the cord. The locomotor period, which corresponds to the time between two action potential bursts, was significantly and reversibly decreased by the addition of 200 μm GABA to the saline: 4.5 ± 0.3 s in control condition, 5.5 ± 0.1 s in the presence of GABA and 4.1 ± 0.1 s after a 1 h wash-out of GABA (n = 3; data not shown). In the presence of 1 μm CGP55845 and 1 μm gabazine, bath-application of NMA/5HT triggered synchronized action potential bursts between the left and right part of the cord as previously reported in the presence of the other GABA_A receptor antagonist, bicuculline (Pflieger et al. 2002). The period of this synchronized activity was no longer significantly affected by the subsequent superfusion of GABA: 17.3 ± 5 s in the presence of CGP5584 and gabazine and 15.9 ± 2 with GABA (n = 3; data not shown). These results suggest that CGP55845 and gabazine applied at 1 μm actually suppress the effects of GABA on the lumbar motor networks in the neonatal rat spinal cord preparation in vitro.

Next we sought whether STD expression was modified in the presence of the GABA_A and GABA_B receptor antagonists. For this purpose, motoneurons were directly impaled in CGP55845–gabazine-containing saline to reduce acquisition times. The spinal cord preparations were globally more excitable in the presence of the GABA_A and GABA_B receptor antagonists. Slight increases in the stimulation voltage of sensory afferents often triggered bursts of action potential that outlasted the stimulation trains (data not shown).

The blockage of the GABA receptors strongly amplified STD during 1 Hz stimulation (compare inset in Figs 1C and 7A) leading to an absence of significant difference in the percentage of depression reached between the three stimulation frequencies tested (Fig. 7A). The STD dynamics was also modified when GABAergic receptors were blocked. In control conditions, the mean amplitude of EPSC₂ was very similar for the three frequencies of stimulation tested (Fig. 1D), while in the presence of CGP55845 and gabazine, mean EPSC₂ amplitude values exhibited a high level of variability (Fig. 7B). This variability was introduced by neurons that did not present a classically depressed EPSC₂. As exemplified in Fig. 7Ca, at 10 Hz stimulation one neuron showed a depressed second postsynaptic current and the other one a facilitated EPSC₂ (Fig. 7Cb). Three motoneurons out of 17 (18%) tested at 5 Hz and 4 motoneurons out of 14 (29%) tested at 10 Hz exhibited a facilitation instead of a depression of EPSC₂ amplitude in the presence of CGP55845 and gabazine.

Figure 7. STD expression in the presence of GABAergic receptor antagonists.

A, bar graph of depression at 1, 5 and 10 Hz in the presence of 1 μm of GABA_A antagonist SR95531 hydrobromide (gabazine) and 1 μm of GABA_B antagonist CGP55845. In these conditions, the depression amount is not frequency-dependent. The numbers in each bar indicate the number of cells tested in each condition. B, plots of the averaged normalized EPSC amplitude as a function of stimulus number at 1, 5 and 10 Hz. Note the large variability of the second EPSC (EPSC₂) during 5 and 10 Hz stimulations. Representative traces of two different motoneurons held at −60 mV expressing a depressed (Ca) and a facilitated (Cb) EPSC₂ in the presence of CGP55845 and gabazine during 10 Hz dorsal root stimulation.

We further investigated the action of noradrenaline on STD when the GABA_A and GABA_B receptors were blocked. In CGP55845 and gabazine containing saline, noradrenaline still induced a stable inward current in motoneurons (mean inward current: 81 ± 10 pA (n = 10); Fig. 8A) and an increase in input resistance (121.6 ± 10 MΩ in CGP–gabazine and 143 ± 10 MΩ (n = 10) in the presence of noradrenaline). In these conditions, noradrenaline failed to decrease the sensorimotor transmission significantly but on the contrary increased it in 3 of the 10 neurons tested (Fig. 8A; mean reduction of ESPC₁: 9.9 ± 11%, n = 10). Similarly in the presence of CGP55845 and gabazine, the effects of noradrenaline on STD were completely abolished, whatever the stimulation frequency (Fig. 8A). Neither the kinetics (Fig. 8Ba–c) nor the amount of depression (Fig. 8C) was significantly affected by noradrenaline when GABA_A and GABA_B receptors were blocked.

Figure 8. Inhibition of the noradrenergic modulation of both sensorimotor transmission and STD in the presence of GABAergic receptor antagonists.

A, representative traces of EPSCs depression recorded during 1, 5 and 10 Hz stimulation in CGP55845 (1 μm)–gabazine (1 μm)-containing saline (left panel in Aa–c) and in the presence of noradrenaline (5 μm; right panel in Aa–c). B, plots of the relative amplitude of EPSCs expressed as a function of stimulus number in the absence (circles) and the presence (squares) of noradrenaline at 1 (Ba), 5 (Bb) and 10 Hz (Bc) trains. C, bar graph of the mean percentage of depression in control conditions (CGP–gabazine) and in the presence of noradrenaline. The numbers in each bar indicate the number of cells tested in each condition.

We then examined the effects of dopamine and serotonin on STD in the presence of CGP55845 and gabazine. Both drugs induced an inward current in motoneurones (53.3 ± 13 pA (n = 9) in the presence of dopamine, 62.5 ± 12 pA (n = 8) in the presence of serotonin) and an increase in input resistance (160 ± 18 MΩ in CGP–gabazine and 184 ± 24 (n = 9) with dopamine, 112 ± 7.6 MΩ in CGP–gabazine and 122 ± 9 (n = 8) with serotonin). Figure 9 shows dopaminergic and serotoninergic effects for 1 Hz stimulation trains. The two amines strongly diminished the EPSC₁ amplitude and it was almost impossible to elicit visible EPSCs using the same stimulation voltage range before and during dopamine or serotonin superfusion. This strong inhibition together with the increased excitability of the cord in the presence of CGP55845 and gabazine (see above) did not make it possible to increase the stimulation voltage in order to trigger an EPSC₁ amplitude value similar to the control amplitude. Bath-application of dopamine or serotonin strongly reduced STD in all the motoneurons tested (Fig. 9A–C). These inhibitory effects on STD expression were not significantly different from control conditions (Figs 3D and 4D). The shape of the depression time course was no different in the presence of dopamine (Fig. 9Da) or serotonin (Fig. 9Db) compared to CGP55845–gabazine control conditions.

Figure 9. Modulatory effects of dopamine (50 μm) or serotonin (5 μm) on STD induced by 1 Hz stimulation in the presence of GABAergic receptor blockers.

A and B, representative traces of EPSCs recorded from lumbar motoneurons during trains of 15 stimuli applied to the dorsal root at 1 Hz in the presence of CGP55845–gabazine (gbz) and with dopamine (A) or serotonin (B). C, mean percentage of depression during 1 Hz stimulation before and in the presence of dopamine or serotonin during GABAergic receptor inhibition. The numbers in each bar indicate the number of cells tested in each condition. D, plots of the normalized EPSC amplitude over stimulus number in the presence of CGP55845–gabazine (a and b) and during the subsequent addition of dopamine (a) or serotonin (b) to the saline. *P ≤ 0.05.

Interestingly, under GABA_A and GABA_B receptor blockage, two distinct populations of motoneurons could be discriminated on the basis of the effects of dopamine and serotonin during 5 and 10 Hz stimulation trains. At 5 Hz, in the presence of dopamine, 33% of the neurons exhibited STD (Fig. 10Aa, n = 9) whereas in the remaining 66%, STD was converted to short-term potentiation (STP, data not shown). In the presence of serotonin, the same stimulation frequency induced STD in 50% of the motoneurons (data not shown) and STP in the remaining ones (Fig. 10Ba). The same dichotomy was observed during 10 Hz stimulations: 56% of the neurons in the presence of dopamine (n = 9) and 50% in the presence of serotonin (n = 8; Fig. 10Ab) expressed STD, whereas in the remaining ones the STD was shifted to STP (Fig. 10Bb). The data obtained in the presence of dopamine and with serotonin were pooled to yield results with a strong statistical value. The bar graphs in Fig. 10Ac show that STD was significantly and equally reduced in the presence of dopamine or serotonin during 5 and 10 Hz dorsal root stimulations. The global shape of depression kinetics was not modified (Fig. 10Ad and e). At facilitating synapses, the extent of the plasticity was not frequency dependent (Fig. 10Bc). During 5 Hz stimulation, facilitation gradually developed (Fig. 10Bd) whereas 10 Hz stimulations triggered a pronounced facilitation of EPSC₂ and EPSC₃ which then stabilized around an intermediate potentiated amplitude value (Fig. 10Be). The development of STD or STP at sensory afferent motoneuron synapses was not significantly correlated to the input membrane potential of the motoneurons, to the segmental level where they were recorded (L4 versus L5 segment) or to the rat age. The amount of STD in CGP–gabazine containing saline was also not significantly different between depressing and facilitating connections before the superfusion of dopamine or serotonin (Fig. 10Ac–Bc, black bars). It is noteworthy, however, that facilitating synapses exhibited a greater tendency to EPSC₂ facilitation in the control condition compared to depressing connections (compare the filled circles in Fig. 10Ad–e and Bd–e).

Figure 10. Dopamine and serotonin strongly reduced STD or converted it into short-term potentiation during 5 and 10 Hz trains in the absence of GABAergic receptor activation.

Aa and b, representative traces of depressing synapses in CGP55845–gabazine-containing medium and in the presence of dopamine (50 μm) at 5 Hz (a) and serotonin (5 μm) at 10 Hz (b). Ba and b, examples of representative recordings from sensory afferent motorneuron connections switched from depression to facilitation in the presence of serotonin at 5 Hz (a) and dopamine at 10 Hz (b). Ac and Bc, pooled depression (Ac) or plasticity extent (Bc) for depressing or facilitating synapses, respectively, in the presence of one of the two drugs. The numbers in each bar indicate the number of cells tested in each condition. Ad and e and Bd and e, plots of averaged and normalized EPSC amplitude over stimulus number at 5 and 10 Hz in the presence of CGP55845–gabazine (•) and in the presence of serotonin or dopamine (□) for depressing (Ad and e) and facilitating synapses (Bd and e). Note the tendency for the facilitation of the second ESCP in control conditions in synaptic connections converted from STD to STP by dopamine or serotonin.

Discussion

While segmental reflexes have been traditionally considered to be very simple hardwired circuits, our study shows that they are in fact highly flexible circuits that express different forms of short-term plasticity depending on their neuromodulatory environment. In this study, we demonstrate (1) that the STD expressed by the segmental reflex circuits in the mammalian ventral spinal cord can be differentially modulated by amines and (2) that depending on the activity of GABAergic interneurons, dopamine and serotonin can convert STD to STP.

STD mechanisms in neonatal rat spinal cord

Our study extends the work by Lev-Tov and Pinco that described the STD of the sensorimotor connections in the neonatal rat spinal cord (Lev-Tov & Pinco, 1992; Pinco & Lev-Tov, 1993a,b). These authors hypothesized that STD was due to presynaptic mechanisms (Lev-Tov & Pinco, 1992). In the present work, we excluded the possible involvement of postsynaptic desensitization of AMPA receptors by using cyclothiazide (Wong et al. 2003; Wall, 2005). Quantal analysis will nevertheless be required to confirm the presynaptic locus of STD in the neonatal rat spinal cord.

Aminergic modulation of STD

A main finding of the present study is that dopamine, serotonin and noradrenaline differentially modulate STD at sensorimotor synapses. In control conditions, dopamine was effective at all stimulation frequencies, whereas serotonin affected STD only during 5 and 10 Hz trains and noradrenaline during 1 and 5 Hz stimulus trains. The differential effects of the amines observed here could be due to the differing degrees of activation of their respective receptors since the three amines were each tested at a single concentration. Although it was not realistic to perform a dose-dependent response analysis of STD in the different conditions tested here, we showed that 50 μm dopamine and 5 μm serotonin similarly depressed sensorimotor transmission (see Results) but exhibited different modulatory profiles on STD (Figs 3 and 4). As discussed below, this differential action of dopamine and serotonin is partly due to the activation of GABAergic interneurons.

STD and GABA

Our study sheds light on the crucial role played by the GABAergic interneurons in shaping and modulating short-term plasticity at sensorimotor synapses. The changes reported in STD expression in the presence of GABAergic receptor antagonists (Fig. 7) suggest the involvement of a tonic GABAergic inhibition or of local synaptic pathways involving GABAergic interneurons since GABAergic interneurons activated by sensory afferents are responsible for presynaptic inhibition (Hultborn et al. 1971; Rudomin & Schmidt, 1999). STD induced by 1, 5 and 10 Hz trains expressed differential sensitivities to the GABAergic antagonists (Fig. 7) that could be in part explained by a differential activation of GABAergic interneurons. Although mephenesin, which strongly reduces the activity in spinal polysynaptic pathways (Wright, 1954; Fulton & Walton, 1986; Pinco & Lev-Tov, 1993b; Cazalets et al. 1996), was bath-applied throughout the experiments, we cannot rule out that some GABAergic pathways involving a few synapses could remain active during sensory afferent stimulation.

Interestingly, however, the aminergic modulation of STD was also strongly modified during the blockage of the GABAergic receptors. The lack of noradrenaline effect on both sensory transmission and STD in the presence of CGP55845 and gabazine (Fig. 8) suggests that its action on sensory terminals is indirect and is due only to the activation of the GABAergic interneurons. In the substantia gelatinosa, the analgesic action of noradrenaline has also been shown to be mediated by a facilitation of GABAergic and glycinergic transmitter release (Baba et al. 2000a,b). In the presence of CGP–gabazine, dopamine and serotonin strongly reduced the STD amount during 1 Hz dorsal root stimulation (Fig. 9), while serotonin failed to affect it in control conditions (compare Figs 4 and 9). During 5 and 10 Hz stimulations, serotonin and dopamine reduced the amount of depression and even converted it into facilitation in about 50% of the motoneurons in the absence of GABAergic receptor activation (Fig. 10). It is already known that facilitation and depression may be expressed at the same synaptic terminals with their relative weight depending largely on the initial probability of neurotransmitter release (p: high p favours depression, low p favours facilitation; for review see Abbott & Regehr, 2004). This shift has been shown to occur spontaneously (Debanne et al. 1996; Bertrand & Lacaille, 2001) or under modulatory control (Parker, 2003; Baimoukhametova et al. 2004; Bevan & Parker, 2004) in other systems. In the neonatal rat spinal cord in vitro, lowering the release probability with baclofen or increasing the Mg²⁺/Ca²⁺ ratio of the saline alters the amount of depression and even converts it to facilitation (Lev-Tov & Pinco, 1992). Our experimental data suggest that in the neonatal rat sensory terminals, there is a continuum in the synaptic state leading from STD to STP. Synapses expressing EPSC₂ depression in the presence of GABAergic antagonists certainly present a high p that is probably reduced in the presence of serotonin and dopamine (Fig. 10) but not enough to topple synapses toward facilitation. In contrast, p should be smaller in synapses presenting an EPSC₂ facilitation in CGP55485–gabazine-containing saline and the further decrease of p by the amines converts these synapses from depression to facilitation. The short-term plasticity expressed at a given time is therefore the result of both afferent input rates and dynamic integration of neuromodulatory influences at sensory terminals.

Different types of STD?

The differential actions of amines on STD and its different sensitivity to the GABAergic antagonists suggest that 1, 5 and 10 Hz stimulation trains induce STD sustained by different mechanisms.

These differences may partly depend on presynaptic calcium dynamics. The spatiotemporal profile of Ca²⁺ influx into presynaptic boutons determines the plasticity expressed at the synapse (for example see Ismailov et al. 2004). Depending on the presynaptic firing rate, the proteins controlling neurotransmitter release could differ and thus lead to the expression of different STDs (for review see Thomson, 2003; Leenders & Sheng, 2005).

The differential aminergic actions could be due to differential selectivity for both calcium channels and/or differential modulation of the vesicular release machinery. We previously showed that N- and P/Q-type Ca²⁺ channels are the main Ca²⁺ channel subtypes underlying excitatory transmission at sensorimotor synapses (Bertrand et al. 2000). Throughout the CNS, serotonin and dopamine exert their biological effects via G-protein coupled receptors (Neve et al. 2004; Kitson, 2007). Depending on the second messenger pathways activated by the amines and/or the GABA, distinct steps of the exocytosis process could be modulated, thus accounting for the different short-term plasticity reported herein.

Physiological relevance

STD plays an essential role in filtering sensory information (O'Donovan & Rinzel, 1997; Zucker & Regehr, 2002; Thomson, 2003; Abbott & Regehr, 2004). In spinal motor networks, the impact of STD in both motoneuronal processing of sensory afferent volley and subsequent muscle contraction are unknown. It has been recently shown that STD of reciprocal inhibition allows the struggling rhythm to be generated in the hatching tadpode hindbrain and spinal cord (Li et al. 2007). In pathologies such as the restless leg syndrome, the emergence of abnormal motor activity is thought to be due to a dysfunction of the modulatory descending pathways, in particular the dopaminergic one, controlling sensory afferents (Barriere et al. 2005b). In this context, recent studies have demonstrated that modulation of spinal reflex through the dopaminergic D₃ receptor could be implicated (Clemens & Hochman, 2004). Following a spinal cord injury (SCI), the dramatic loss of supraspinal neuromodulatory inputs, which finely tune short-term plasticity, engenders exaggerated reflex responsiveness (for review see Frigon & Rossignol, 2006) leading to spastic symptoms. Sensory information is also crucial for functional locomotor recovery after SCI and locomotor training has been shown to influence transmission directly in sensory pathways (Frigon & Rossignol, 2006; Petruska et al. 2007).

In conclusion, our study is the first to show a differential control of sensory afferent inflow by amines and to demonstrate that they may be implicated in shifting the synaptic plasticity from depression to facilitation. The timing of the descending aminergic pathway inputs onto both GABAergic interneurons and sensorimotor terminals then appears crucial to determine the short-term plasticity expressed at sensory afferent–motoneuron synapses.