INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Neural networks consist of inhibitory and excitatory neurons that use fast synaptic transmission to produce a basic motor output (see Marder and Calabrese 1996). The properties of network neurons and their synaptic connections can be altered by relatively slow-acting neuromodulators, resulting in changes in network and behavioral outputs (see Harris-Warrick and Marder 1991; Marder and Calabrese 1996; Sillar et al. 1997).

Neuropeptides form a major class of neuromodulators (Hökfelt 1991). Substance P belongs to the tachykinin family of neuropeptides. Members of this peptide family have been found in both vertebrate and invertebrate nervous systems (see Maggio 1988). In the lamprey spinal cord, immunohistochemical studies have shown tachykinin immunoreactivity in the dorsal, ventral, and lateral axon columns. The functionally important C-terminal sequence of lamprey tachykinins shows strong homologies to mammalian substance P and neurokinin A (NKA) (Waugh et al. 1996). Particularly high levels of tachykinins are found in the dorsal horn and in a ventromedial plexus that contains the co-localized amines serotonin (5-HT) and dopamine (Schotland et al. 1995, 1996; Van Dongen et al. 1985b, 1986). Tachykinin immunoreactivity is also found in close apposition to cell bodies and dendrites of motor neurons (Van Dongen et al. 1985a; Svensson, unpublished observations).

The effects of substance P have been analyzed directly on the lamprey locomotor network and its sensory and descending brain stem inputs (Parker and Grillner 1996, 1998, 1999a,b; Parker et al. 1997, 1998; Ullström et al. 1999). Substance P increases the amplitude of glutamatergic reticulospinal synaptic inputs and thus potentiates descending inputs to the spinal cord. It also potentiates sensory inputs by depolarizing mechanosensory afferents and by increasing their excitability and action-potential duration by acting through a pertussis-toxin-insensitive G protein and protein kinase C (PKC)-dependent mechanism (Parker and Grillner 1996; Parker et al. 1997). It also potentiates excitatory, but reduces inhibitory, sensory synaptic transmission to spinobulbar neurons and increases the excitability of spinobulbar neurons (Parker and Grillner 1996). These effects are associated with a net excitatory effect on sensory inputs, shown by the substance-P-mediated potentiation of skin stimulation-evoked reflex responses (Ullström et al. 1999).

At the segmental locomotor network level, a single 10-min application of substance P increases the frequency and improves the regularity of N-methyl-D-aspartate (NMDA)-evoked ventral root bursts, effects that last in excess of 24 h (Parker et al. 1998). There are at least three phases to the burst frequency modulation. An initial induction phase (<2 h) requires the PKC-dependent potentiation of NMDA responses and increased calcium levels in network neurons. This is followed by an intermediate maintenance phase (2-15 h) that requires de novo protein, but not RNA synthesis, and by a final phase (>15-20 h) that does require RNA synthesis (Parker and Grillner 1999b; Parker et al. 1998). The burst regularity modulation is, in contrast, not affected by protein synthesis inhibitors, but may require tonic protein kinase A (PKA) activation (Parker and Grillner 1999b).

Substance P has several effects on the cellular and synaptic properties of presumed network neurons. These include modulation of excitability and the slow afterhyperpolarization following the action potential (Parker and Grillner 1998), the potentiation of monosynaptic glutamatergic transmission from excitatory network interneurons, and the activity-dependent facilitation or depression of excitatory and inhibitory synaptic transmission, respectively (Parker and Grillner 1999a).

Preliminary data showed that substance P evokes membrane potential oscillations in motor neurons and certain network interneurons (see Parker and Grillner 1998). However, the mechanisms underlying these oscillations were not examined in any detail, and thus it is not known whether they were due to an effect of substance P on intrinsic cellular properties or if they were synaptically generated. These oscillations have thus been examined in detail here.

Motor neurons in the lamprey are assumed to be pure output elements (Wallén and Lansner 1984). The effects of substance P on motor neurons will thus reflect potential effects on intrinsic motor neuron properties as well as on the segmental premotor network. The results show that the membrane potential oscillations evoked in motor neurons by substance P are due to a distributed excitatory effect on spinal neurons. These oscillations cannot generate a behaviorally relevant network output in quiescent preparations (i.e., those examined in the absence of NMDA and ongoing network activity). The oscillations are not due to effects on intrinsic motor neuron properties but are synaptically generated by glycinergic and NMDA- and non-NMDA-mediated glutamatergic inputs. These inputs presumably arise as a result of a direct substance-P-mediated depolarization of inhibitory and excitatory premotor interneurons. Substance P thus appears to have a general activating effect on the locomotor network; this could contribute to its modulatory effects during ongoing activity.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Adult river lampreys (Lampetra fluviatilis) were anesthetized with tricaine methane sulfonate (MS 222 100 mg/l, Sandoz, Basel), and the spinal cord and notochord were removed in Ringer containing (in mM) 138 NaCl, 2.1 KCl, 1.8 CaCl₂, 1.2 MgCl₂, 4 glucose, 2 HEPES, and 0.5 L-glutamine, bubbled with O₂; the pH was adjusted to 7.4 with 1 mM NaOH. Pieces of spinal cord were isolated from the notochord and pinned ventral side up in a silicone elastomer (Sylgard)-lined chamber and superperfused with Ringer.

Intracellular recordings were made from motor neurons or unidentified gray matter neurons using thin-walled micropipettes filled with 4 M potassium acetate and 0.1 M potassium chloride. Motor neurons were identified by recording orthodromic spikes in adjacent ventral root following current injection into their somata. An Axoclamp 2A amplifier (Axon Instruments) was used for amplification and in discontinuous current-clamp (DCC) mode for injecting depolarizing and hyperpolarizing current and for keeping the resting membrane potential at the control level when the Ringer composition was altered or drugs were applied. Extracellular ventral root recordings were made by sucking the cut ventral roots into glass suction electrodes. Data was recorded on 486 PC equipped with an A/D interface (Digidata 1200) and analyzed using Axon Instruments software (Axotape, pClamp6).

Drugs were dissolved in the Ringer and rapidly perfused to the bath (5 ml/min) using a peristaltic pump to give a fast onset and termination of drug application. The following drugs were used (in µM): 1 substance P, 1.5 tetrodotoxin (TTX), 200 CdCl₂, 1,000-2,000 kynurenic acid (Sigma, Stockholm), 100 D-AP5, 10 6-cyano-7-nitroquinoxalene-2,3-dione (CNQX; Tocris, Bristol, UK), 5 strychnine (Apoteks Bolaget AB, Stockholm), 2 spantide II (Peninsula Laboratories Europe LTD), 20 chelerythrine, and 10-20 phorbol 12,13-dibutyrate (RBI, Natick, MA). In experiments using low-potassium Ringer, the potassium chloride concentration was reduced to 50% and replaced with sodium chloride to maintain osmolarity. Drugs and altered Ringer solutions were applied for 10-30 min before substance P. Substance P was applied at 1-h intervals. Application of substance P at intervals of <1 h resulted in reduced responses, presumably as a result of desensitization.

The amplitude and frequency of substance-P-evoked oscillations were measured during the initial 4-min period after substance P (30 s-2 min) application. The amplitude was measured from the baseline preceding substance P application to the peak amplitude of the highest depolarization, excluding spikes. Data are presented as means ± SE. Statistical significance was analyzed using Student's paired t-test.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Preliminary data showed that substance P enhanced synaptic inputs and evoked spiking and oscillatory activity in motor neurons and three types of spinal interneurons (see Parker and Grillner 1998). These oscillations have been examined in detail here.

Substance-P-evoked membrane potential oscillations in motor neurons

The effects of substance P were studied using short-lasting bath application (30 s to 2 min). In 65 of 71 neurons (39 motor neurons and 32 unidentified gray matter neurons), substance P increased spontaneous synaptic activity (shown by the thickened baseline in Fig. 1A) and evoked irregular membrane potential oscillations on which spikes could occur. Ventral-root activity occurred in phase with depolarizing oscillations in ipsilateral motor neurons (see Figs. 5 and 6, A and C). The oscillations were characterized by depolarizing plateaus with a mean peak amplitude of 8.1 ± 1.1 mV. The plateaus had a mean frequency of 0.06 ± 0.01 Hz and alternated with hyperpolarizing periods. The oscillation episode lasted for 6.6 ± 0.4 min. The oscillations varied in different preparations. However, in individual cells, they had a similar amplitude and frequency over repeated trials, providing that there was an interval of 1 h between substance P applications (see following text). Oscillations could be evoked 10-20 min after the wash-off of substance P, although the effect at this time was reduced, possibly due to desensitization. The early rapid hyperpolarization seen in some traces was an artifact of the perfusion system, as it did not occur consistently, and where it occurred it was not influenced by changes in membrane potential or Ringer composition.

View larger version (16K): [in this window] [in a new window]   Fig. 1. A: rapid perfusion of the spinal cord (30 s) with 1 µM substance P induces slow and irregular membrane potential oscillations in motor neurons. The bars underneath the traces on this and other figures indicate the onset and duration of substance P application. B: the effect of substance P was blocked by the selective tachykinin receptor antagonist spantide II (2 µM). C: the substance-P-induced oscillations recovered after washout of spantide II for 1.5 h.

The oscillations were blocked by the general tachykinin antagonist spantide II (2 µM; n = 3 unidentified neurons) (Parker et al. 1997), suggesting that substance P acted through tachykinin receptors (Fig. 1B). Responses to substance P recovered after washing out spantide II for 1.5 h (Fig. 1C).

Substance-P-induced oscillations are blocked by TTX

Preliminary data from excitatory network interneurons (EIN) showed that substance P could evoke a large depolarization that was resistant to TTX (Parker and Grillner 1998), thus suggesting a direct depolarizing effect of substance P. The substance-P-evoked oscillations in motor neurons could be mediated synaptically by the activation of EINs or alternatively could be due to the activation of intrinsic membrane properties in the motor neurons themselves. To examine these possibilities, TTX (1.5 µM) was bath applied to block action potential-evoked synaptic inputs. Substance P failed to evoke any oscillatory activity in the presence of TTX (Fig. 2A, bottom, n = 7). The oscillations were thus not due to a direct effect of substance P on intrinsic cellular properties but were presumably generated synaptically. However, a slow substance-P-mediated membrane potential depolarization persisted in the presence of TTX, suggesting that substance P also had a direct depolarizing effect on the membrane potential. This depolarization was 2.1 ± 0.2 mV (n = 7) and thus considerably smaller than that suggested by preliminary experiments in EINs (7.8 ± 2.1 mV) (see Parker and Grillner 1998).

View larger version (25K): [in this window] [in a new window]   Fig. 2. A: the substance-P-induced oscillations are blocked by TTX (1.5 µM), leaving only a depolarization of the motor neurons (bottom). Substance P does not affect the motor neuron input resistance, shown by the response to a 3.0 nA current pulse (see inset). In the presence of low-potassium Ringer, the amplitude of the depolarization was increased. B: graph showing the effect of low-potassium Ringer on the amplitude of the substance-P-induced depolarization. The peak amplitude was increased by 59% (n = 7, P < 0.05).

Substance P does not significantly affect the input resistance of motor neurons at resting membrane potentials (Fig. 2A) (see also Parker and Grillner 1998). This suggests either that the conductance underlying the direct depolarization occurs at sites distant from the recording site in the soma or that it is due to opposing effects on two or more conductances that thus result in no net change in input resistance (Jiang et al. 1994). Ion substitution experiments and ion channel antagonists were used to examine the conductances underlying the substance-P-mediated depolarization. Because the depolarization persisted in TTX, it could not be due to Na⁺ entry through TTX-sensitive sodium channels. In the presence of TTX, low-potassium Ringer (see METHODS) increased the amplitude of the substance-P-evoked depolarization by 59 ± 18% (Fig. 2, A and B; n = 7, 6 motor neurons, 1 unidentified neuron, P < 0.05 paired t-test), suggesting that a reduction of voltage-dependent potassium conductances could contribute to the depolarization. Low-potassium Ringer hyperpolarized the cells by ~4 mV (data not shown). Because the membrane potential was kept at the control level by injecting positive current, the increased depolarization was not simply due to the increased depolarizing drive resulting from the hyperpolarization. Cd²⁺ (200 µM), also in the presence of TTX, had no consistent or significant effect on the amplitude of the depolarization (n = 4, data not shown), suggesting that Ca²⁺ or calcium-dependent K⁺ channels do not contribute to the depolarization.

Substance-P-induced oscillations are PKC dependent

Substance P acts through PKC in lamprey sensory neurons, resulting in broadening of the action potential and an increase in excitability through the reduction of a 4-AP-sensitive potassium conductance (Parker et al. 1997). PKC also mediates the substance-P-mediated potentiation of NMDA responses in motoneurons and the modulation of the NMDA-evoked segmental network output (Parker et al. 1998). The involvement of PKC in the substance-P-induced membrane potential oscillations was examined using the specific PKC antagonist chelerythrine (20 µM) (Parker et al. 1997). Chelerythrine blocked the substance-P-induced oscillations. A membrane potential depolarization persisted (Fig. 3, A and B; n = 5, 4 motor neurons, 1 unidentified neuron), although its amplitude was reduced to 54.6 ± 6.1% of control (Fig. 3C, n = 5; P < 0.05). To determine whether it affected both the oscillations and the direct depolarizing effect of substance P, the effects of chelerythrine were examined in the presence of bath-applied TTX. Its amplitude was not significantly affected by chelerythrine (mean reduction 14.4 ± 9.4%, n = 5, P > 0.05; data not shown). The substance-P-induced membrane potential oscillations were thus PKC-dependent, whereas the TTX-insensitive depolarization did not significantly dependent on PKC.

View larger version (14K): [in this window] [in a new window]   Fig. 3. The specific protein kinase C antagonist chelerythrine (20 µM) blocked the substance-P-induced oscillations. A: control response to substance P. B: the oscillations were blocked in the presence of chelerythrine, although a membrane potential depolarization remained. The thickened baseline is due to noise introduced by the use of discontinuous current-clamp mode (DCC) to keep the membrane potential at the control level. C: graph showing the reduction of the amplitude of the peak depolarization in the presence of chelerythrine (n = 5).

To further investigate the role of PKC in the induction of the oscillations, spinal cords were perfused with the PKC-activating phorbol ester phorbol 12, 13-butyrate (PDBu, 10-20 µM, 8-12.5 min) (Parker et al. 1997). PDBu did not depolarize the membrane potential or induce oscillations (data not shown, n = 3, 2 motor neurons, 1 unidentified neuron), although it did markedly increase spontaneous synaptic activity, an effect that is probably related to its pre and postsynaptic facilitation of glutamatergic inputs (Parker, unpublished observations).

In addition to PKC, substance P also acts through PKA to improve the regularity of network activity (Parker et al. 1998). To examine whether PKA was involved in the substance P-induced oscillations or the membrane potential depolarization, spinal cords were incubated with the PKA and PKG antagonist H8 (10 µM). H8, however, did not affect either the induction of the oscillations or the membrane potential depolarization (n = 5, data not shown), suggesting against an involvement of these second messenger pathways.

Substance-P-induced oscillations are due to glutamatergic and glycinergic synaptic inputs to motor neurons

The results obtained with TTX suggested that the membrane potential oscillations were synaptically generated. The lamprey locomotor network consists of glutamatergic and glycinergic interneurons (see Buchanan 1982; Buchanan et al. 1989). Specific glutamatergic and glycinergic antagonists were thus used to determine the role of these transmitters in the oscillations. Glutamatergic synaptic inputs were initially examined by applying substance P in the presence of kynurenic acid (1-2 mM), an antagonist of both NMDA and AMPA/kainate receptors. Kynurenic acid abolished the substance-P-evoked oscillations (see Fig. 4B) and reduced the peak amplitude of the depolarization by 81 ± 5.1% (Fig. 4D; n = 6, 3 motor neurons and 3 unidentified neurons; P < 0.005).

View larger version (15K): [in this window] [in a new window]   Fig. 4. The glutamate receptor antagonist kynurenic acid blocks the substance-P-induced oscillations. A: control trace showing oscillations following perfusion with substance P (1 µM). B: the oscillations were blocked by kynurenic acid (2 mM), an antagonist of ionotropic glutamatergic receptors, leaving a small depolarization of the membrane potential. C: the effect of substance P partly recovered after washout of kynurenic acid. D: graph showing the effect of kynurenic acid on the substance-P-induced response. The peak amplitude of the substance P response was reduced by 81 ± 5.1% (n = 6).

The involvement of glutamatergic inputs was examined further using specific antagonists of ionotropic glutamate receptors. The AMPA/kainate-receptor antagonist CNQX at a concentration that abolishes non-NMDA-mediated inputs (10 µM) (Alford and Grillner 1990) either abolished the oscillations (n = 5 of 7), or markedly reduced their frequency and amplitude (see Fig. 6) and also inhibited the substance-P-evoked ventral root activity (see Fig. 6B). Where oscillations remained, CNQX reduced the peak amplitude of the depolarization by 56 ± 7.4% (n = 7; P < 0.05) and the frequency by 83 ± 16.7% (n = 7; 3 motor neurons, 4 unidentified neurons, P < 0.05).

Blocking NMDA inputs with the selective NMDA-receptor antagonist D-AP5 (100 µmM) reduced the frequency of the oscillations by 56 ± 3.3% (Fig. 5, A-D; n = 6; P < 0.005) and the peak amplitude by 30 ± 7.0% (Fig. 5E; n = 6, 4 motor neurons, 2 unidentified neurons; P < 0.05). However, in contrast to CNQX, in no case were the oscillations abolished by D-AP5. D-AP5 also failed to block the increase in synaptic noise (Fig. 5B) and the ventral root activity that occurs during the depolarizing plateaus (Fig. 5, F and G).

View larger version (42K): [in this window] [in a new window]   Fig. 5. Effects of the N-methyl-D-aspartate (NMDA)-receptor antagonist D-AP5 (100 µM) on the substance-P-induced oscillations. A: control response evoked by 1 µM substance P. B: in the presence of D-AP5, the frequency and the amplitude of the oscillations were reduced but not abolished. C: the oscillations recovered after washout of D-AP5 for 1 h. The rapid perfusion system caused an early hyperpolarization of the membrane potential in this experiment. D and E: graphs showing the reduction of the frequency and peak amplitude of the oscillations (n = 6). D-AP5 did not block the ventral root activity induced by substance P. F: control trace from a motoneuron showing substance-P-induced oscillations and corresponding ventral root activity. G: D-AP5 reduced the ventral root activity but did not abolish it.

Glutamate antagonists thus markedly reduced the depolarizing component of the oscillations, with non-NMDA receptors playing a greater role. In the presence of CNQX, no phasic excitatory input was seen (Fig. 6B). However, irregular hyperpolarizing shifts in membrane potential still occurred when glutamatergic inputs were blocked although their amplitude was significantly reduced by 25 ± 4.8% of control (Fig. 6, B and F; n = 7; P < 0.005). The glycine receptor antagonist strychnine (5 µM), which abolishes inputs from glycinergic inhibitory interneurons (McPherson et al. 1994), was used to determine if these inputs were glycinergic. Strychnine alone did not evoke any oscillatory activity, but it abolished the hyperpolarizing inputs evoked by substance P (see Fig. 6B) and resulted in large plateau depolarizations and bursts of spikes instead of the characteristic irregular oscillations (see Fig. 7, A and B). Strychnine increased the peak amplitude of the substance-P-induced depolarization by 294 ± 93% of control (Fig. 7C, n = 9, 3 motor neurons and 6 unidentified neurons; P < 0.005).

View larger version (44K): [in this window] [in a new window]   Fig. 6. Effects of the AMPA/kainate receptor antagonist 6-cyano-7-nitroquinoxalene-2,3-dione (CNQX, 10 µM) on the substance-P-induced oscillations. Extracellular recordings from left and right ventral roots (L-VR, R-VR) together with intracellular recording (IC) from a motoneuron ipsilateral to the left ventral root is shown. A: control effect of substance P. The oscillations induced by substance P are associated with activity in the left ventral root. Note that the ventral root activity is slow and irregular but alternates between the left and right sides. B: CNQX markedly reduced the frequency and peak amplitude of the substance-P-induced oscillations and abolished the ventral root activity. Note, however, that some hyperpolarizing membrane potential deflections still occur in CNQX (arrow in B). C: the response to substance P partly recovered after washout of CNQX. Graphs showing the reduction of the frequency (D) and amplitude (E) of the oscillations (n = 7). F: the amplitude of the hyperpolarizing membrane potential fluctuations was also reduced by CNQX.

View larger version (13K): [in this window] [in a new window]   Fig. 7. The effect of the glycinergic receptor antagonist strychnine on the substance-P-induced oscillations. A: control response to substance P. B: in the presence of strychnine (5 µM), substance-P-induced larger depolarizations (notice change in scale in B) on which bursts of spikes were evoked. The spikes are clipped in B. C: graph showing the increased amplitude of the oscillations by substance P in the presence of strychnine (n = 9).

Glutamatergic and glycinergic inputs thus underlie the substance-P-induced membrane potential oscillations. That these inputs were sufficient to account for the oscillations was confirmed by the combined application of CNQX, D-AP5, and strychnine. This combination of antagonists abolished all fast synaptic inputs and oscillatory activity, to leave only the slow TTX-resistant depolarization of the membrane potential (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study shows that substance P depolarizes and evokes membrane potential oscillations in lamprey motor neurons. The oscillations, but not the depolarization, are blocked by TTX. They are thus mediated synaptically and not by an effect on intrinsic motor neuron properties. Substance P thus appears to have a general excitatory effect on the premotor network although on its own is not able to evoke a behaviorally relevant network output.

Substance P depolarizes spinal neurons and induces membrane potential oscillations

Brief application of substance P resulted in an increase in synaptic noise, membrane potential oscillations, and spiking in motor neurons, and associated activity in ventral roots. The activity could to some extent alternate between left and right ventral root (Fig. 6A). Substance P can also evoke oscillatory activity in inhibitory and excitatory spinal interneurons (see Parker and Grillner 1998) although its interneuron effects have not been examined in any detail. The oscillations had similar peak amplitudes and frequencies in individual cells, providing that an interval of 1 h was left between substance P applications. This interval was presumably required to avoid desensitization of tachykinin receptors or effector (e.g., second messenger) pathways. The effects could vary markedly in different preparations (see Fig. 5). This variability could be due to differences in the access of substance P to the spinal cord, differences in substance P breakdown mechanisms, or differences in resting electrophysiological (e.g., membrane potential, excitability) or biochemical (e.g., protein kinases or protein phosphatases) conditions.

The effects of substance P were blocked by the general tachykinin receptor antagonist spantide II, which blocks the sensory and network effects of substance P in the lamprey (Parker et al. 1997, 1998). This suggests that the oscillations were mediated through the activation of tachykinin receptors. The membrane potential oscillations were blocked by TTX and thus were generated synaptically. A substance-P-mediated membrane potential depolarization persisted in the presence of TTX, however, thus revealing a direct depolarizing effect on the membrane potential. The TTX-resistant membrane potential depolarization was increased in low-potassium Ringer, suggesting that it was associated with the reduction of potassium conductances active at rest. In rat motor neurons, substance P reduces a persistent potassium conductance to evoke TTX-sensitive ventral root activity and membrane potential oscillations (Fisher and Nistri 1993; Fisher et al. 1994). The input resistance of lamprey motor neurons was not significantly affected by substance P as would be expected if there was a conductance increase or decrease. The reduction of potassium conductances underlying the depolarization may thus occur distally on dendrites, making a conductance change invisible at the recording site in the soma. Alternatively, the depolarization may not solely depend on a reduction in potassium conductances but may depend on a mixed conductance increase and decrease that results in no net change in input resistance (see Jiang et al. 1994) or effects on electrogenic pumps, which will again not affect the input resistance (Morita et al. 1993; Parker et al. 1996). Since Cd²⁺ did not affect the amplitude of the depolarization, Ca²⁺ conductances or Ca²⁺-dependent K⁺ channels presumably do not contribute.

Synaptic basis of the oscillations

The general glutamatergic receptor antagonist kynurenic acid abolished the oscillations. CNQX, which antagonizes both AMPA and kainate receptors as well as the glycine site on NMDA receptors (Birch et al. 1988), also abolished or significantly reduced the oscillations. The NMDA-receptor antagonist D-AP5 reduced the amplitude and frequency of the oscillations although in no case did it abolish the oscillations or block the ventral root activity. Glutamatergic inputs thus underlie the oscillations, with non-NMDA receptors apparently having a more significant role.

Substance P evokes a large TTX-resistant membrane potential depolarization (~10 mV) and spiking in glutamatergic EINs (Parker and Grillner 1998). However, as with motor neurons, there are no intrinsic oscillations of the EIN membrane potential in TTX. The direct depolarization of the EINs could mediate the increase in glutamatergic inputs that evoke the oscillations in motor neurons. The phasic strychnine-sensitive inhibitory inputs that occur during the oscillations could arise either as a result of a similar direct excitatory effect of substance P on inhibitory interneurons or the glutamatergic feedforward excitation of these interneurons following EIN activation. Hyperpolarizing inputs were reduced but persisted when glutamatergic inputs were blocked (see Fig. 6). This suggests that substance P has a direct excitatory effect on inhibitory interneurons but also that there is a feedforward excitation of these interneurons as a result of the activation of glutamatergic interneurons. This further emphasizes the multiple distributed effects of substance P on the locomotor network (see also Parker and Grillner 1998) that all need to be considered in any attempt to explain its network effects. Inhibitory inputs could arise from crossed caudal interneurons (CCIN), although substance P usually reduces the excitability of these cells (Parker and Grillner 1998). Other potential sources include lateral interneurons, which are excited by substance P (Parker and Grillner 1998), or small crossing (ScIN) or ipsilateral (SiIN) inhibitory interneurons (see Parker and Grillner 2000b). The ScINs can evoke significantly larger glycinergic inhibitory inputs than CCINs and lateral interneurons (LINs), stimulation of a single ScIN inhibiting ongoing activity in motor neurons (Parker, unpublished observation). They are thus potential sources of the powerful phasic inhibitory inputs underlying the repolarizing phase of the oscillations suggested by the large enhancement of the effects of substance P in the presence of strychnine. Thus taken together it would appear that the direct depolarizing effects of substance P directly or indirectly activates premotor interneurons that feed onto motor neurons to evoke the oscillations. While we cannot rule out an effect on intrinsic oscillatory properties in premotor interneurons, a network effect of this sort is supported by the antiphasic ventral-root activity evoked by substance P (see Fig. 6).

Substance-P-induced effects on the oscillations are PKC dependent

Tachykinin receptors belong to the family of seven transmembrane spanning G-protein-coupled receptors (Nakanishi 1991). The specific PKC antagonist chelerythrine, which antagonizes the effects of substance P and PKC-activating phorbol esters on lamprey sensory neurons (Parker et al. 1997) and the locomotor network (Parker et al. 1998), blocked the substance-P-induced oscillations. Chelerythrine, however, at a concentration that occludes the sensory and network effects of substance P (Parker et al. 1997, 1998), only partially reduced the amplitude of the direct substance-P-mediated membrane potential depolarization, suggesting an underlying mechanism that was independent of PKC. Neither the oscillations nor the membrane potential depolarization were significantly affected by blocking PKA or protein kinase G (PKG).

The PKC-activating phorbol ester PDBu failed to evoke a direct membrane potential depolarization, further suggesting against the involvement of a PKC-dependent pathway underlying this effect. PDBu, however, also failed to induce membrane potential oscillations, although it increased spontaneous synaptic activity, presumably through its pre- and postsynaptic facilitatory effect on glutamatergic transmission (Parker, unpublished observations). While the failure of PDBu to mimic the oscillatory or depolarizing effects of substance P suggests against the involvement of PKC, it must be remembered that the indiscriminate activation of second-messenger pathways will not necessarily mimic the discrete effects evoked by localized transmitter-mediated activation of second-messenger pathways (see Gustafsson et al. 1988).

The oscillations are thus dependent on PKC activation, while the intracellular pathway responsible for the depolarization is PKC, PKA, and PKG independent. This suggests that an intracellular pathway separate to the PKC- and PKA-dependent effects shown previously (Parker et al. 1998) underlies the depolarizing effect of substance P. A single modulator may thus act through several intracellular pathways to affect the network output. Alternatively, the ability of chelerythrine to block the membrane potential oscillations but not the direct depolarization could be due to a more significant effect on the depolarization in the premotor interneurons that evoke the oscillations. This has not yet been studied. Alternatively, the nonsignificant effect of chelerythrine on the depolarization (~15%) may be sufficient to block the oscillations when summed across a number of premotor interneurons. Finally, PKC-mediated effects not directly dependent on the membrane potential depolarization may evoke the synaptic inputs underlying the oscillations. These effects could include the modulation of premotor interneuron excitability or the efficacy of synaptic transmission (see Parker and Grillner 1998; Parker, unpublished observations).

The modulatory effect of substance P on the frequency of ongoing network activity is PKC dependent, whereas the burst regularity effect is PKA dependent (see Parker et al. 1998). The oscillations could thus contribute to the burst frequency modulation. However, because the oscillatory activity is brief, it could contribute to the early phase of this network modulation. However, an increase in presynaptic activity at this time could result in enhanced glutamate release and NMDA receptor activation that contributes to the NMDA and calcium-dependent induction phase of the burst FM (Parker et al. 1988).

Role of the substance-P-mediated modulation

Substance P may thus affect different cellular properties in different components of the locomotor network. This occurs for the effects of 5-HT on the gill withdrawal reflex in Aplysia (Sugita et al. 1992; Sutton and Carew 2000) and could allow the selective recruitment of different component effects under different circumstances. This could occur through modulator interactions that affect intracellular pathways (see Bhalla and Iyengar 1999). Support for this is provided by the ability of the GABA_B receptor agonist baclofen to block the oscillations, but not the direct depolarizing effect of substance P (Svensson, unpublished observations). In addition, 5-HT can abolish the effects of substance P, whereas dopamine can gate certain components of its repertoire (Svensson et al. 2001). Selection could also occur through modulatory effects arising from the activation of different receptor subtypes (see Katz 1998). The potential for effects being mediated through separate tachykinin receptors has not been investigated. Preliminary data suggest that the receptors cannot be separated on the basis of a classification using mammalian agonists and antagonists (Parker, unpublished observations). However, their future use may allow separate receptor-mediated effects to be identified.

Substance P cannot itself evoke a coordinated network output. In the absence of the activation of other pathways, substance P produces slow, irregular network driven activity. Similar effects of substance P occur on locomotor activity in the neonatal rat (see Barthe and Clarac 1997). While the general excitatory effects of substance P shown here are not able to evoke behaviorally relevant network activity, its excitatory effects could contribute to the short-term potentiation of ongoing network activity, the induction phase of the long-term network modulation or the potentiation of reflex responses evoked by cutaneous stimulation (Ullström et al. 1999).