INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Although a considerable body of research has documented the roles of G protein signaling in intracellular signal transduction, the functions served by G proteins in rhythmic motor pattern generation are less well explored. Only recently has interest started to focus on whether G protein signaling participates directly in motor pattern generation (Delgado-Lezama et al. 1997; Krenz et al. 2000; Krieger et al. 1998), possibly mediating intrinsic neuromodulatory actions (Katz 1995; Katz and Frost 1996). A more widely recognized role for G protein signaling in central pattern generator (CPG) circuit neurons is in mediating the actions of extrinsic neuromodulatory inputs (Arata et al. 1993; Mironov et al. 1999). Such neuromodulatory inputs to CPGs can change the cellular and synaptic properties of CPG components and thereby alter the motor output (Calabrese 1998; Feldman and Smith 1989; Harris-Warrick and Marder 1991; Kiehn and Katz 1999; Marder and Calabrese 1996). Here we present evidence that G protein signaling also can be necessary for the operation of a CPG and that it can mediate some of the neuromodulatory actions of neurons intrinsic to the circuit. These findings raise the possibility that G protein signaling might not merely alter the motor pattern but that it might be an integral part of the motor pattern-generating process.

The system that we studied is the escape swim response of the nudibranch mollusk, Tritonia diomedea. The Tritonia swim response is controlled by a CPG that contains intrinsic neuromodulatory elements, the dorsal swim interneurons (DSIs) (Katz 1998). The DSIs are members of the CPG circuit; they fire bursts of action potentials in time with the rhythmic motor behavior and they synapse onto other members of the CPG (Getting 1989a; Getting et al. 1980). The DSIs are serotonergic (Fickbohm and Katz 2000; Katz et al. 1994; McClellan et al. 1994) and use serotonin (5-HT) for both neurotransmission and neuromodulation (Katz and Frost 1995a). They elicit at least two different neuromodulatory effects on another CPG neuron (interneuron C2): heterosynaptic facilitation of neurotransmitter release (Katz and Frost 1995a,b) and enhancement of excitability (Katz and Frost 1997).

Previous work demonstrated that the electrophysiological properties of the known CPG neurons and their synaptic interconnectivity can account for the generation of the rhythmic swim motor program (Getting 1989a,b; Getting et al. 1980); however, there were suggestions of a role for G protein signaling in the pattern generation process. For example, although DSI modulatory actions were absent in the original simulations of this circuit (Getting 1983, 1989b), they may be necessary to recreate some aspects of the neural circuit (Frost et al. 1997). Furthermore, many of the synaptic potentials evoked by the CPG neurons onto each other have multiple components with time courses of several seconds (Getting 1981), which is typical of synaptic potentials mediated by G-protein-coupled receptors and second messengers (Iverson and Goodman 1986; Kehoe 1990; Libet 1986). Computational models of the swim CPG suggested that the time courses of the slow synaptic potentials play crucial roles in the generation of the rhythmic motor pattern (Getting 1983, 1989b). Furthermore, pharmacological studies suggested that serotonergic neuromodulatory actions are essential for the operation of the CPG (McClellan et al. 1994). We had shown previously that the DSIs activate both ionotropic and metabotropic pathways in some follower neurons, the dorsal flexion neurons (DFN) (Clemens and Katz 2001). Therefore we sought to test whether the DSI modulatory actions within the CPG itself are also mediated through G-protein-coupled receptors and whether G protein signaling contributes directly to the production of the rhythmic motor program.

Some of these data have been published previously in abstract form (Clemens and Katz 1999a, 2000).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Dissection and electrophysiology

Experiments were performed on adult Tritonia diomedea obtained from Living Elements (Delta, British Columbia, Canada) and maintained in artificial recirculating seawater prior to experiments. Dissection protocols were as described earlier (Getting et al. 1980; Katz and Frost 1995a). The isolated CNS, consisting of the fused cerebropleural and pedal ganglia, was desheathed and superfused with normal saline [which contained (in mM) 420 NaCl, 10 KCl, 10 CaCl₂, 50 MgCl₂, 10 D-glucose, and 10 HEPES, pH 7.4]. Neurons were identified using both physical and electrophysiological criteria (Getting 1977; Getting et al. 1980). The cell bodies of the DSIs and C2 are located in the cerebral ganglia. The dorsal flexion neurons (DFNs), followers of C2 and DSI, are located in the pedal ganglion contralateral to the somata of their presynaptic neurons.

Electrophysiological recordings were obtained using Axoclamp 2B amplifiers (Axon Instruments, Union City, CA). Electrophysiological signals were digitized with a CED 1401plus and analyzed with Spike2 software (Cambridge Electronic Design, Cambridge, UK).

Experiments examining the effects of the guanine nucleotides on the DSI-elicited heterosynaptic facilitation of C2 synapses were performed in high-divalent cation saline [which contained (in mM) 285 NaCl, 10 KCl, 25 CaCl₂, 125 MgCl₂, 10 D-glucose, and 10 HEPES, pH 7.4] to increase firing thresholds and thus decrease the contribution of polysynaptic pathways. In these experiments, intracellular microelectrode recordings were made simultaneously from a C2, an ipsilateral DSI, and a contralateral DFN. The C2 electrode contained one of the drugs to be tested, dissolved in 20 mM HEPES (Clemens and Katz 2001). C2 was stimulated to fire action potentials at 60 s intervals, and a DSI was made to fire 40 action potentials at 20 Hz ending 2 s before the onset of every other C2 stimulus. Individual action potentials were elicited with brief current pulses delivered through the recording electrode (7 nA, 20 ms). The amplitude of the monosynaptic summated EPSP recorded in the DFN in response to C2 stimulation alone was compared with the amplitude of the C2-evoked excitatory postsynaptic potential (EPSP) after DSI stimulation. The stimulus protocol was repeated after a drug was injected into C2 via iontophoresis.

Tests of C2 excitability were performed in normal saline. Simultaneous intracellular recordings were obtained from a C2 and an ipsilateral DSI. C2 was depolarized every 10 s with a constant current pulse (2 nA, 2 s). The DSI was stimulated as above every 60 s, ending 2 s before the onset of the next C2 stimulus. The spiking response of C2 when stimulated alone was compared with that obtained after DSI stimulation. The protocol was repeated after drug iontophoresis into C2.

Rhythmic swim motor patterns were elicited by stimulating a body-wall nerve (pedal nerve 3) with 20 V pulses (20 ms duration at 20 Hz for 1-2 s). To avoid habituation of the swim motor program (Frost et al. 1996), nerve stimuli were applied at intervals of 10 min or more. Before drug application, the motor pattern was elicited several times to assure a constant response. The guanine nucleotides were then applied by intracellular iontophoresis. The effects of drug injection on motor pattern production generally appeared 20-30 min after the end of the injection. Under control conditions, C2 bursts usually consist of more than 20-25 action potentials with an average instantaneous spike frequency of 10-12 Hz (cf. Katz and Frost 1997). Therefore, to distinguish between bursting and nonbursting in a weakly active C2, we defined a burst as consisting of 10 action potentials at a frequency of >5 Hz.

Guanosine 5'-O-(2-thiodiphosphate) (GDP--S), guanosine 5'-O-(3-thiotriphosphate) (GTP--S) (trilithium salts), and 5'-guanylyl-imidodiphosphate (GMP-PNP, trisodium salt) were dissolved in 20 mM HEPES at 10^-4 to 10^-3 M (all chemicals: Sigma, St. Louis, MO) and injected iontophoretically into identified neurons (Clemens and Katz 2001). To improve visibility of the electrode tip, the electrode solution also contained 0.2% Fast Green, filtered with a 0.22-µm syringe filter (Millipore, Bedford, MA). All intracellular solutions were adjusted to pH 7.4. Injection pulses (5 to 7 nA, 500-ms duration) were applied at 1 Hz in bouts of 15-30 min. Stimulus test protocols were begun after recovery to the preinjection membrane potential of the injected neuron.

Data analysis

All values are given as means ± SE. SigmaStat (SPSS Science, Chicago, IL) was used to test for significant differences between samples. Parametric or nonparametric comparisons of data groups were performed where appropriate. Differences were considered significant if P < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Injection of GDP--S into a single C2 blocked motor pattern generation

To test whether G protein signaling in C2 is involved in the production of the rhythmic swim motor pattern, we intracellularly iontophoresed GDP--S into C2 and analyzed the effect on the CPG activity. Prior to injection of GDP--S, stimulation of a body wall nerve reliably evoked the rhythmic motor program, which is characterized by a repeated series of action potential bursts in CPG neurons DSI and C2 (Fig. 1A). Injection of GDP--S into one of the pair of bilaterally symmetric C2 neurons led to a significant decrease in the number of burst cycles produced by the CPG neurons in response to nerve stimulation (Fig. 1, B and C; n = 8). Although GDP--S was injected into only one C2 neuron, in the right cerebral ganglion (R-C2), this treatment eventually eliminated rhythmic activity throughout the network as seen in the simultaneous recordings from the contralateral C2 (L-C2) and from DSIs in both the left and right cerebral ganglia (L-DSI and R-DSI). Intracellular recordings from CPG follower neurons and extracellular nerve recordings also exhibited an absence of rhythmic bursting activity (not shown here). Although the injected R-C2 in this preparation fired only the initial spike in response to the nerve stimulus (arrow), the un-injected L-C2 still fired action potentials, albeit at a low frequency.

View larger version (36K): [in this window] [in a new window]   Fig. 1. Blocking G protein signaling in a single C2 interrupts rhythmic motor pattern generation. Schematic representation of the CPG (black lines) and its input and output pathways (gray lines) (adapted from Frost et al. 2001). Sensory neurons (S) transmit information both directly to the single dorsal ramp interneuron (DRI) and indirectly via a trigger neuron (Tr1). DRI excites the serotonergic dorsal swim interneurons (DSIs). The CPG that generates the rhythmic motor pattern is formed by the 3 DSIs (DSI-A, -B, and -C), the 2 ventral swim interneurons (VSI-A and -B), and cerebral neuron 2 (C2). The CPG neurons project onto efferent neurons in the pedal ganglia that relay the pattern of activity to the muscles, producing the dorsal or ventral flexion phases of the swim behavior (dorsal flexion neurons, DFN-A and -B; ventral flexion neurons, VFN). A: electrically stimulating pedal nerve 3 (20-V, 20-ms pulses at 20 Hz for 1 s, at arrow) elicited a swim motor pattern consisting of bursts of action potentials in C2 and DSI. B: iontophoretic injection of the GDP-analogue GDP--S into the one C2 (here R-C2) was sufficient to block motor pattern generation. As in the preinjection state, nerve stimulation increased DSI firing and caused an immediate depolarization of C2, suggesting that there was no change in the inputs to the circuit. C: pooled data from all experiments showing the significant reduction from 3.3 ± 0.2 to 1.0 ± 0.5 bursts in C2 after injection with GDP--S (P = 0.002, n = 8, Kruskal-Wallis with post hoc Dunn's comparison). D: time course of the effects of GDP--S injection into C2 from the experiment shown in A and B. Prior to injection of GDP--S, nerve stimulation (every 10 min) reliably evoked 3 swim cycles (defined by the number of C2 bursts). After injection, nerve stimulation abruptly failed to elicit any bursts in C2 or the other CPG neurons. A is from trial 2 and B is from trial 4.

In a subset of four experiments, GDP--S caused a complete failure of C2 bursting (Fig. 1, B and D). However, the effect of GDP--S was not always immediate; rather, the number of cycles declined to zero after a small number of nerve stimuli (Fig. 2A).

View larger version (14K): [in this window] [in a new window]   Fig. 2. Average time courses of responses to repeated nerve stimuli. Pedal nerve 3 was stimulated every 10 min (20 V, 20 ms pulses at 20 Hz for 1 s) to elicit repeated swim motor programs. A: the effects of injecting guanosine 5'-O-(2-thiodiphosphate) (GDP--S; ), the HEPES carrier solution (), and GTP analogues () into C2 were compared. Prior to injection, the average number of swim cycles was stable. After injection of GDP--S, the average number of swim cycles decreased and progressively declined with repeated stimuli. In contrast, injection of C2 with just the HEPES carrier had no effect on the number of swim cycles and did not cause a significant decline in cycle number per swim (P > 0.05, n = 3). However, as with GDP--S, after injection of GTP analogues [either guanosine 5'-O-(3-thiotriphosphate) (GTP--S) or 5'-guanylyl-imidodiphosphate (GMP-PNP)] into C2, there was a gradual decline in the average number of cycles per swim with repeated stimulation. B: in preparations where no drugs were included in the microelectrode at all, there was no change in the number of swim cycles in response to nerve stimuli delivered every 10 min (P > 0.05, n = 12). All tests Kruskal-Wallis with post hoc Dunn's comparison.

To test whether the carrier solution in the electrode might have contributed to this decline, we injected HEPES together with the marker Fast Green into C2 and tested its effects on the motor program (n = 4). Injection of the carrier for up to 30 min did not reduce the number of cycles per swim motor pattern (Fig. 2A, black circles).

The reduction of cycle number observed in response to GDP--S injection was not due to cycle number habituation. It was reported previously that the swimming behavior exhibits habituation in the form of decreased cycle number when repeatedly elicited (Frost et al. 1996). The behavior of the animal habituates even when the stimuli were applied at intervals of 30 min. We found that, in the isolated nervous system, a 10 min stimulus interval did not cause a reduction in cycle number per swim (Fig. 2B, n = 12). Rather, after a slight initial increase after the first swim, the swim cycle number remained stable for more than an hour.

Injection of GTP analogues into a single C2 also blocked motor pattern generation

The results with GDP--S suggested that G protein signaling in C2 is essential for the operation of the Tritonia swim CPG. Therefore, to test whether nonspecific activation of G proteins in C2 is sufficient to permit motor pattern production, or perhaps enhance its performance, we iontophoresed the nonhydrolyzable GTP analogues (GTP--S, n = 6) or GMP-PNP (n = 2) into C2 (Fig. 3). Prior to injection, nerve stimulation reliably evoked the swim motor program in each preparation (Figs. 2A and 3A). Injection of either GTP analogue inhibited the production of the swim motor program (Fig. 3, B and C). As with GDP--S, the number of swim cycles progressively decreased after the injection of the GTP analogues (Figs. 2A and 3D).

View larger version (27K): [in this window] [in a new window]   Fig. 3. Tonic activation of G-protein-coupled second-messenger pathways in a single C2 also blocked the generation of the rhythmic motor pattern. A: the swim motor program elicted by stimulating pedal nerve 3 (20 V, 20 ms pulses at 20 Hz for 1 s, ). B: iontophoretic injection of GMP-PNP into a single C2 blocked motor pattern generation. C: pooled data from all experiments with GTP analogues (both GTP--S and GMP-PNP show a significant reduction from 2.8 ± 0.3 to 1.2 ± 0.5 bursts in CPG neurons after injection with GTP analogues (P = 0.015, n = 7). D: time course of the effects of GMP-PNP injection into C2 from the experiment shown in A and B. Prior to injection, the preparation exhibited a reliable motor pattern consisting of 4 bursts. After GMP-PNP injection, nerve stimulation became less effective and by the third stimulus failed to elicit any bursts in the CPG neurons. A is from trial 3 and B is from trial 8. All tests Kruskal-Wallis with post hoc Dunn's comparison.

These results suggest that tonic, nonspecific activation of G protein pathways in C2 is not only insufficient to produce or sustain rhythmic motor activity but that it disrupts motor pattern generation. Thus interference in G protein signaling in C2 leads to the inhibition of the escape swim motor program.

G proteins mediate heterosynaptic facilitation of C2 synapses but not basal release

We next examined sites of action of the guanine nucleotide analogues that could potentially contribute to their ability to block motor pattern generation when injected into C2. For example, interfering with G protein signaling in C2 might simply block neurotransmitter release (Haydon and Trudeau 1998; Mirotznik et al. 2000; Takahashi et al. 2000) and thus contribute to disabling the CPG. Therefore we examined the effects of guanine nucleotide analogues on C2 synaptic strength. Alternatively, interfering with G protein signaling might not directly block C2 synapses but rather might reduce the heterosynaptic facilitation of those synapses by the serotonergic DSIs (Katz and Frost 1995b). To address both of these possible effects, we monitored the synaptic output of C2 by recording the monosynaptic EPSP that it evokes in DFN (Hume and Getting 1982) and also examined the effect on DSI modulation of that synapse.

Stimulation of C2 alone (5 Hz, 1 s, every 60 s) evoked a reliable and stable monosynaptic EPSP in DFN (Fig. 4A1, black trace). We found that blocking G protein signaling in C2 with GDP--S did not reduce C2 synaptic strength (Fig. 4A2, black trace). In the pooled data of seven experiments, we did not observe any significant change in the amplitude of EPSPs recorded in DFN after injection of GDP--S into C2 (Fig. 4B1, P > 0.5, Kruskal-Wallis with Dunn's pairwise comparison). These data indicate that GDP--S does not impede neurotransmitter release from C2.

View larger version (20K): [in this window] [in a new window]   Fig. 4. Injection of GDP--S into C2 had no effect on basal C2 synaptic strength but reduced the DSI-evoked enhancement of C2-evoked excitatory postsynaptic potentials (EPSPs). A: in high-divalent cation saline, C2 was made to fire 5 action potentials at 5 Hz every 60 s using 20 ms, 7 nA current pulses. Bottom: the times of C2 spikes; top: averages from 3 sweeps in the same DFN neuron when C2 was stimulated alone (black traces) and when C2 was stimulated 2 s after a spike train in an ipsilateral DSI (20 Hz, 2 s; gray traces). A1: preinjection. When C2 was stimulated alone, it elicited a summated monosynaptic EPSP in a contralateral DFN (black trace). When C2 was stimulated after DSI, the amplitude of the C2-evoked EPSP recorded in DFN was larger (gray trace, C2 + DSI). DSI stimulation directly depolarizes the DFN, causing the large declining depolarization seen when C2 was stimulated after DSI. A2: GDP--S. Injection of GDP--S into C2 did not alter the size of the EPSP evoked by C2 when stimulated alone (black trace). However, the amount of facilitation elicited by DSI stimulation was substantially reduced (gray trace). B1: C2 synapse. Summary of all 7 experiments showing that injection of GDP--S did not reduce the size of the C2-DFN EPSP (P = 0.742). B2: summary of the same 7 experiments depicted in B1, showing that GDP--S significantly reduced the extent of DSI-elicited heterosynaptic facilitation in these experiments (P = 0.036, all tests Kruskal-Wallis with Dunn's comparison).

In contrast, GDP--S injection into C2 reduced the heterosynaptic facilitation of C2 synapses caused by DSI stimulation (n = 4, Fig. 4, A, 1 and 2, gray trace, and B2). Prior to injection of GDP--S, DSI stimulation (20 Hz for 2 s, ending 2 s before C2) caused the amplitude of the C2 to DFN EPSP to increase, on average, to 214 ± 16% of the amplitude when C2 was stimulated alone (Fig. 4B2). After GDP--S injection, the increase was significantly reduced to only 169 ± 7% of when C2 was stimulated alone (Fig. 4B2, P = 0.036). In the one preparation where we specifically tested for possible effects of the carrier HEPES, injection of the carrier HEPES alone did not alter the DSI modulation of this synapse (data not shown).

To test whether nonspecific activation of G proteins in C2 affects its synaptic strength, we injected GTP--S into C2. A C2 was stimulated (5 pulses at 5 Hz), and the EPSP was recorded in a contralateral DFN (Fig. 5A, black trace). Unlike GDP--S, iontophoresis of GTP--S into C2 had a direct effect on C2 synaptic strength; it increased the amplitude of C2-evoked EPSPs recorded in the DFN in both experiments where we were able to maintain the recording after drug injection (Fig. 5, A, gray trace, and B).

View larger version (9K): [in this window] [in a new window]   Fig. 5. Injection of GTP--S into C2 increased C2-evoked EPSPs in DFN. A: in high divalent cation saline, C2 stimulation (7 nA pulses, 20 ms duration at 5 Hz for 1 s, every 60 s) elicited discrete EPSPs in the postsynaptic DFN (black trace). Iontophoresis of the GTP analogue, GTP--S, into C2 increased the amplitude of the postsynaptic potentials in DFN (gray trace). The traces represent averages from 4 control sweeps and 5 sweeps after injection within a single preparation. Inset: the responses of the DFN to the 1st and 2nd C2 spike before and after injection of C2 with GTP--S. GTP--S did not affect the size of the first EPSP. B: the amplitude of the C2-evoked summated EPSP recorded in DFN increased by 23 and 50% respectively, in the 2 experiments where GTP--S was injected into C2. Each data point is the average of 4 and 5 EPSPs respectively before injection and 4 and 9 EPSPs after the injection.

The enhancement of EPSPs appeared to be due primarily to an increase in homosynaptic facilitation causing the latter EPSPs in the train to grow larger; there was little or no effect on the size of the initial EPSPs in a train (see Fig. 5A, inset). This effect is similar to the effect of DSI stimulation on paired-pulse facilitation (Katz and Frost 1995b).

GDP--S reduces C2 excitability while the GTP analogues block the modulation by DSI

Excitability is another property of C2 that could be affected by guanine nucleotide injection. This property is also modulated by DSI stimulation (Katz and Frost 1997). To test excitability, a C2 was repeatedly depolarized (2 nA for 2 s every 10 s). DSI stimulation (20 Hz for 2 s), preceding C2 depolarization by 2s, increased the number of spikes produced by C2 (Fig. 6, A1 and B1).

View larger version (18K): [in this window] [in a new window]   Fig. 6. The effects of guanine nucleotides on C2 excitability and its modulation by DSI. A: GDP--S reduced C2 excitability but did not alter the ability of DSI to enhance excitability (n = 3). C2 was stimulated repeatedly in normal saline every 10 s with 2 nA, 2 s current pulses, until it displayed a consistent action potential firing frequency. A1: under control conditions, DSI stimulation (20 Hz for 2 s) substantially increased C2 the number of action potentials produced by C2 in response to the same current pulse. A2: injection of GDP--S into C2 decreased the basal spiking response of C2 but did not abolish its modulation by DSI. 1 and 2 are from the same experiment. A3: pooled data from all 3 experiments show that GDP--S caused a decrease in C2 spiking response to ~40% of preinjection (P < 0.001), while the HEPES control decreased the C2 spiking response only to ~80% (n = 3, P < 0.01). The difference between HEPES and GDP--S injection was highly significant (P < 0.001). A4: pooled data show that GDP--S did not significantly affect the ability of DSI stimulation to increase C2 excitability. Before injection, DSI increased C2 spiking to 244 ± 17% of control. After GDP--S injection, DSI stimulation caused C2 spiking to increase to 226 ± 25% of control. This is not a significant change (P = 0.3). B: GTP analogues did not alter the basal excitability of C2 but inhibited the DSI-elicited enhancement of C2 excitability (n = 4). B1: under control conditions, DSI stimulation increased the firing response of C2 (stimulated as in A). B2: injection of a GTP analogue, here GTP--S, did not alter C2 basal excitability but abolished the ability of DSI to modulate C2 excitability. B3: pooled data from all 4 preparations show that C2 excitability, when compared with the HEPES control, is not affected by the GTP analogues. B4: GTP analogue injection caused a significant decrease in the extent to which DSI stimulation enhanced C2 excitability from 145 ± 9% of control to 110 ± 7% of control (P < 0.007). All tests Kruskal-Wallis test with post hoc Dunn or Tukey comparison.

Injection of GDP--S into C2 reduced the basal excitability of C2 to ~40% of preinjection levels (n = 3), suggesting a role for G proteins in maintaining cell excitability (Fig. 6A, 1-3). Iontophoresis of the carrier HEPES alone led to only a slight decrease in C2 excitability to ~80% of the preinjection values (Fig. 6A3).

GDP--S injection did not reduce the DSI-elicited enhancement of C2 excitability (Fig. 6A, 1,2, and 4). DSI stimulation enhanced the number of action potentials fired in C2, when C2 was stimulated with a 2 s, 2 nA current pulse (Fig. 6A1). After injection with GDP--S, this modulation persisted (Fig. 6A2), and its relative increase remained unchanged (Fig. 6A4). Moreover, in one experiment, we increased the current injected into C2 to restore basal C2 frequency to its preinjection range, but this procedure did not reduce the modulatory capabilities of the DSI (data not shown). These data indicate that this neuromodulatory action may not be dependent on G protein signaling.

We next tested the effects of injecting C2 with the GTP analogues GTP--S (n = 2) or GMP-PN (n = 2). Neither GTP--S nor GMP-PNP altered the basal excitability of C2 (Fig. 6B, 1-3). These results suggest that although G protein signaling seems to be involved in the regulation of C2 excitability, nonspecific activation of G protein pathways alone is not sufficient to increase C2 excitability. However, unlike the lack of an effect of GDP--S, the GTP analogues strongly reduced the ability of DSI stimulation to enhance the C2 spiking response (Fig. 6B, 2 and 4). Thus the role of G proteins in mediating this modulatory action of the DSIs is ambiguous.

Hyperpolarization of a single C2 does not disrupt motor pattern generation.

To determine if the ability of the guanine nucleotide analogues to block the swim motor program could be related to their effect on C2 excitability, we reduced C2 firing by injecting a strong continuous hyperpolarizing current (5 to 7 nA). Previous work had shown that hyperpolarization or voltage-clamp of both C2 neurons blocks the production of the swim motor program (Lennard et al. 1980). However, it was also previously shown that voltage-clamp of just one of the bilaterally symmetric C2 neurons did not prevent rhythmic motor pattern generation (Getting and Dekin 1985). Similarly, we found that, although hyperpolarization of a single C2 greatly reduced its spiking response and dramatically increased its voltage excursions during each burst cycle, it had little effect on the swim motor program as monitored in other CPG neurons (Fig. 7). In the experiment depicted here, three intracellularly recorded CPG members (both C2s and a DSI) expressed a robust and reliable swim motor pattern when stimulated under control conditions (Fig. 7A). Continuous hyperpolarization (here: 7.5 nA) of one of the two C2s, starting ~30 s before the nerve stimulus, did not prevent motor pattern production (Fig. 7B). Similar results were obtained in five different preparations. Thus whereas injection of GDP--S, GTP--S, or GMP-PNP halted motor pattern production, hyperpolarization of a single C2 did not noticeably alter the motor pattern.

View larger version (22K): [in this window] [in a new window]   Fig. 7. Hyperpolarization of a single C2 neuron did not substantially alter motor pattern generation (n = 5 preparations). A: in response to electrical stimulation of pedal nerve 3 (arrow), simultaneous intracellular recordings of both C2s and a DSI show the pattern of action potential bursts characteristic of the escape swim motor program. B: the membrane potential of the right C2 (R-C2) was strongly hyperpolarized with current injection into its soma (-7.5 nA at open arrow) prior to the onset of the swim stimulus. This treatment greatly decreased the spiking response of R-C2, even causing it to not fire at all on the 3rd cycle. However, hyperpolarization of this 1 C2 did not reduce the number of bursts or the intensity of firing in the other neurons.

The ability of guanine nucleotides to block the production of the swim motor program when injected into a single C2 does not appear to be due to these molecules passing to the contralateral C2 through gap junctions. The electrical coupling between the two contralateral C2s is very weak; current injected into one neuron only slightly affects the voltage of the other; note that the membrane potential of the L-C2 is not affected by strong hyperpolarization of the R-C2 (Fig. 7). Furthermore, we have not observed dye coupling between the contralateral C2s using biocytin (Molecular Probes), whereas we have seen dye coupling between other electrically coupled neurons in Tritonia (data not shown). These findings are strengthened by data obtained in one other experiment where we successfully recorded from both C2s and their ipsilateral DSIs after injecting GMP-PNP into only one C2. In that case also, we did not observe any effect on the excitability or synaptic strength of the contralateral, un-injected C2.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We found that interference with G protein signaling in a single neuron of the Tritonia swim CPG disrupts motor pattern generation. We previously used iontophoretic injection of guanine nucleotide analogues in the DFNs to separate responses to DSI stimulation mediated by ionotropic receptors from those mediated by G-protein-coupled receptors (Clemens and Katz 2001). Our data in this study show that we can also successfully alter the properties of a single CPG neuron using the same approach. Although we cannot determine the final concentrations of these substances that reached the synaptic terminals of the injected neuron, the fact that we saw an effect on synaptic release suggests that we were successful at delivering high enough concentrations.

We found that G protein activation in CPG neuron C2 is necessary for motor pattern generation; intracellular iontophoresis into C2 of GDP--S, a nonhydrolyzable analogue of GDP that blocks activation of G proteins, prevented body-wall-nerve stimulation from eliciting a swim motor program. To test whether nonspecific activation of G proteins would allow motor pattern generation, we injected C2 with the nonhydrolyzable GTP analogues GTP--S or GMP-PNP. These analogues also disrupted motor pattern generation. Control injections of the carrier alone did not have any effects on the generation of the rhythmic swim motor program. We conclude from these data that G protein signaling in C2 is directly involved in the motor pattern-generating process and that any alteration in these second-messenger pathways is sufficient to halt rhythmogenesis.

The strong effect of guanine nucleotide analogues on the swim motor program was not simply due to blocking synaptic transmission from C2; GDP--S had no significant effect on the basal strength of the C2 to DFN synapse, whereas GTP--S potentiated C2 synaptic strength. Furthermore, although the GDP--S decreased C2 excitability, this is unlikely to have caused the disruption in the swim motor pattern; strong hyperpolarization of C2 reduced C2 firing substantially, but did not affect the bursting pattern in other CPG neurons.

It is also unlikely that the decrease in cycle number was caused by habituation of the response. Although Frost et al. (1996) reported a strong habituation of the swim motor behavior in vivo to repeated stimuli at 10 min intervals, we found that in the isolated nervous system, a 10 min interval resulted in a stable response for more than an hour. We did, however, observe a strong habituation, similar to the one described by Frost et al. when we decreased the swim interval to 2 min (not shown). There have been other reports of differences between the behavior of the animal and the fictive pattern recorded from the isolated nervous system. For example, in a series of studies on the behavioral repertoire of another CPG system, the stomatogastric (STG) nervous system of crustaceans, data from in vivo and in vitro experiments demonstrated that this CPG does not behave identically under such different conditions (Clemens et al. 1998, 1999).

G proteins mediate DSI neuromodulatory actions

Interfering with G protein signaling disrupts neuromodulatory actions mediated by G-protein-coupled receptors. C2 cellular and synaptic properties are modulated by the serotonergic DSIs (Katz and Frost 1995a, 1997). We found that injection of GDP--S into C2 reduced the ability of the DSIs to heterosynaptically enhance C2 synaptic strength, supporting the hypothesis that the DSI actions are mediated by G-protein-coupled receptors on C2. These data also support the previous conclusion that the DSIs act presynaptically on C2 to enhance release of neurotransmitter (Katz and Frost 1995b).

Injection of GTP--S into C2 mimicked DSI stimulation by enhancing C2 synaptic strength in the two experiments where we could hold the cells after injection of the drug. As with DSI modulation, the synaptic enhancement appeared to be caused primarily by an increase in C2 homosynaptic facilitation. However, we do not know whether GTP--S acted on the same G proteins as DSI stimulation or whether nonspecific G protein activation yields a similar synaptic enhancement through a mechanism other than serotonin receptor activation.

The effect of guanine nucleotide analogues on the ability of DSI stimulation to increase C2 excitability was ambiguous. Although GDP--S caused a general decrease in C2 excitability, it did not block the ability of DSI stimulation to enhance C2 excitability. In contrast, while the GTP analogues GTP--S and GMP-PNP did not affect basal C2 excitability, they blocked the ability of DSI to enhance C2 excitability. One possible explanation is that the DSIs might act through a pathway that is not mediated by G proteins but that is overridden by G protein activation. Another possibility might be that strong activation of guanine nucleotide-dependent signaling induces a cross-talk between different second-messenger pathways (Diversé-Pierlussi et al. 1997) that could mask the potentiating effects of the DSIs.

It could also be that GDP--S and the GTP analogues were able to elicit at least partially similar effects in C2. GDP--S and GTP analogues are generally considered to exert opposite effects on G protein signaling, with GDP--S interfering with and GTP--S prolonging metabotropic actions (e.g., Firestein et al. 1991). However, there is evidence that these guanine nucleotide analogues can also have similar effects in their target neurons. For instance, both GDP--S and GTP analogues can activate the adenylate cyclase system in rat (Rasenick et al. 1989), and both can block a serotonin-induced facilitation in isolated motor neurons of Aplysia (Chitwood et al. 2001). Thus further work is needed to determine the mechanisms underlying the DSI-evoked effects on C2 excitability in Tritonia.

Role of G protein signaling in the generation of rhythmic motor patterns

The roles of G proteins in motor pattern-generating systems are relatively unexplored. G protein-coupled receptors have been shown to mediate some actions of glutamate in the lamprey locomotor system (Krieger et al. 1998) and in the spiny lobster stomatogastric nervous system (Krenz et al. 2000). However, it is not clear if these receptors are mediating the actions of neurons intrinsic to the CPG or descending from higher brain centers. G protein-coupled receptors mediate the neuromodulatory actions of amines and peptides that provide extrinsic neuromodulatory input to CPGs in many systems (Feldman and Smith 1989; Harris-Warrick and Marder 1991; Marder and Calabrese 1996; Straub and Benjamin 2001). In those cases, the role of G proteins would be to alter cellular and synaptic properties of CPG neurons and thereby alter the motor pattern produced by the CPG.

In the Tritonia swim CPG, G protein signaling mediates at least one of the neuromodulatory actions of the serotonergic DSIs. Because the DSIs themselves are constituent members of the CPG, these G protein-dependent pathways would be activated as part of the normal motor pattern-generating mechanism rather than as a means of only modulating it. Although these results indicate that G protein signaling is essential for the production of the motor pattern, its precise role and mechanisms have not been determined yet.

Role for biochemical signaling in pattern generation?

The Tritonia swim CPG has been described as a network oscillator that depends on the synaptic interactions of its component neurons (Getting 1989a). Yet hyperpolarization or voltage clamp of one of the bilaterally symmetric C2 neurons did not perturb the rhythmic network activity. This lack of effect of voltage contrasts strongly with the effect of intracellular iontophoresis of the guanine nucleotide analogues into a single C2 that were capable of shutting off rhythmic bursting.

Why should iontophoresis of biochemical agents be more effective than voltage at blocking bursting activity? Our results do not suggest that the guanine nucleotides are able to pass through the gap junctions and affect the contralateral, un-injected C2. The two C2s are only weakly electrically coupled with a coupling coefficient of ~0.02 (Getting et al. 1980), and our preliminary results suggest that they are not dye-coupled to each other. Moreover, injection of one C2 with a guanine nucleotide did not affect the properties of its contralateral homologue.

Our results suggest that biochemical manipulations are more successful than soma current injection or voltage clamp at halting the motor program not because they have better access to the un-injected cell but because they are more effective at interfering with the rhythm-generating mechanism within the injected C2. It might be that motor pattern generation is dependent on the production of nonsynaptic messengers such as nitric oxide or arachidonic acid that could be formed by activation of G-protein-coupled receptors but not depend on membrane depolarization (Gelperin 1994; Moroz et al. 1996; Volterra and Siegelbaum 1988). Or, alternatively, it might be that the gap junctions between the two C2s play a crucial role in the communication of second-messenger signals (Kiehn and Tresch 2002) and that altering G protein signaling in one C2 is sufficient to alter gap junctional coupling between the two neurons.

It is possible that second-messenger pathways are intimately and directly involved in the actual motor pattern-generating process, perhaps via cyclic nucleotide-gated channels; cyclic nucleotide-gated channels can form the basis for bursting properties (Cooper et al. 1995, 1998; Green and Gillette 1983; Huang and Gillette 1993; Sudlow and Gillette 1995). Membrane potential oscillations generated as a result of these channels could be insensitive to current injection. Del Negro et al. (2002) found recently that voltage-dependent bursting in pacemaker neurons of the preBötzinger complex is not necessary for rhythmogenesis in these neurons, and they speculated that the generation of the rhythmic motor pattern might depend on a biochemical oscillator.

In summary, we have presented evidence that G protein signaling, intrinsic to the Tritonia escape swim CPG, can be an important factor in the generation of rhythmic motor patterns and that information signaled through G proteins might play a role in the normal operation of this neural circuit.