INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In addition to generating specific patterns of behaviorally relevant neural outputs, a major task of the nervous system is to coordinate the functions of different neural circuits so as to optimize behaviors. Neurons that project from one region of the nervous system to distant regions provide unique vantage points from which the coordination of neuronal function can be studied. The projection neurons typically constitute a small subset of the total number of neurons within a region (Coleman et al. 1992; Fredman and Jahan-Parwar 1979; Gillette et al. 1982; Jing and Gillette 1999; McCrohan and Kyriakides 1989; Rosen et al. 1991). Such neurons may be important either in coordinating different components of a single class of behavior or in behavioral selection of unrelated or opposing behaviors. The responses controlled by interganglionic neurons may be specific components of a complex behavior or may be components of general responses that represent the operation of a central arousal system. Previous findings have indicated that feeding behavior in Aplysia involves specific responses of the feeding apparatus (e.g., buccal mass) but also involves considerable coordinated activity of the visceromotor effectors that are controlled by the abdominal (Dieringer et al. 1978; Koch et al. 1984) and pedal-pleural ganglia (Xin et al. 1996b). We sought to identify cerebral neurons that might be involved in coordinating aspects of feeding behavior with behaviors regulated by the abdominal ganglia. We used dye backfills of the abdominal connectives to define candidate neurons. In the present research we report on the characteristics of three identified neurons that appear to be involved in both buccal-mass behaviors and visceral responses.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Experiments were done using wild-type Aplysia californica weighing 200-300 g (Marinus, Long Beach, CA). A total of approximately 200 animals were used. The animals were fed ad lib and were maintained at 14-16°C in holding tanks containing aerated, filtered artificial seawater (ASW) for at least 3-6 days before being used for experiments. Two types of preparations were used in the experiments: isolated ganglia, which included all the head ganglia (buccal, cerebral, pedal, pleural) as well as the abdominal ganglion and pleural-abdominal connectives, and semi-intact preparations that included head ganglia and the head, including the lips, anterior tentacle region, and cephalic artery, and buccal mass. In the semi-intact preparations, the tentacle nerves and buccal nerves remained intact. In another type of semi-intact preparation, the organs associated with the innervation of the abdominal ganglion were retained together with the circumesophageal and abdominal ganglia and the main nerves of the abdominal ganglion (siphon, genital-pericardial, branchial) and the pleural-abdominal connectives.

For dissection, animals were immobilized by injection of isotonic magnesium chloride (25% of body weight). Isolated ganglia were pinned to a clear silicone elastomer (Sylgard) floor of a recording chamber containing fresh ASW. The cerebral and pedal-pleural ganglia were pinned dorsal side up, and the abdominal ganglion was pinned dorsal or ventral surface up, depending on the neurons to be investigated in a particular experiment. In some experiments, the abdominal artery remained attached to the abdominal ganglion via the pericardial nerve. The semi-intact preparations were set in a clear acrylic plastic (Lucite) recording chamber consisting of two compartments containing ASW. The head ganglia were pinned in one compartment. The lips, anterior tentacles, buccal mass, and cephalic artery that supplies the lips and buccal mass were set in the second compartment. The second compartment was deeper than the first so that the tissue could be completely immersed in the ASW. The cephalic artery was cannulated, and fresh ASW was pumped into the vascular system at a rate of about 0.5 ml/min to perfuse the tissue and to simulate the hydroskeleton of the animal. A suction tube for the outflow was set in the compartment to control the fluid level. The isolated head could be presented with mechanical or chemical stimuli. Mechanical stimuli were provided by the tip of a heat-sealed glass Pasteur pipette that was applied manually to a given receptive surface. Combined chemo-mechano stimuli consisted of pieces of moistened dried-seaweed (laver, Vega Trading, New York, NY) that were applied to the lips or tentacles with a fine, blunt forceps. A pure chemical stimulus consisted of a seaweed extract solution that was applied by a 1-ml syringe and slowly injected into the ASW 1 cm from one side of the lip and tentacle region (Susswein et al. 1978). The partition between the two compartments contained fine grooves that allowed the peripheral nerves to pass through. The grooves were then filled with petroleum jelly (Vaseline) to maintain a watertight seal between the two compartments.

Electrophysiology

All in vitro experiments were carried out at room temperature (19-21°C). For the intracellular recording and stimulation, neurons were impaled with double-barreled microelectrodes that were made of thin-walled glass (World Precision Instruments, FL) and contained 2 M potassium acetate. The electrodes were flow beveled so that their impedance ranged from 10 to 15 M. Neurons were identified based on multiple criteria including size, morphology, coloration, position within a ganglion, and synaptic interconnection to other cells. To test for monosynapticity of connections, the ganglia were bathed in a high divalent cation solution. Unless otherwise specified, this solution contained 5× normal Ca²⁺ and 2× normal Mg²⁺ [(in mM) 334 NaCl, 10 KCl, 50 CaCl₂, and 100 MgCl₂]. High divalent cations raise the threshold for action potential generation although at the concentrations that still permit intracellular firing of neurons, some firing of interposed interneurons may occur (see, for example, Fig. 20B).

For extracellular recording or stimulation the cut ends of nerves were drawn into small-diameter polyethylene suction electrodes. Nerve recordings were made with an AC amplifier (A-M Systems), and electrical stimulation of the nerves was provided by a Grass S88 stimulator. For monitoring contractions of blood vessels, an arm of an isotonic tension transducer (Harvard Bioscience) was attached to the distal end of the vessel (Xin et al. 1996a).

Morphology

To determine the sizes, shapes, and destinations of axons of identified neurons, cells were filled with dye. For these experiments, the potassium acetate in the stimulating electrode was replaced by a solution of 3% 5(6)-carboxyfluorescein dye (Kodak) in 0.1 M potassium citrate, titrated to pH 8.0 with KOH (Rao et al. 1986). The electrodes were beveled so that the impedance of the electrode containing the dye was 15-20 M, and the impedance of the potassium acetate electrode was 10-15 M. Successful intracellular labeling was achieved by iontophoretic injection of the dye for 15-60 min, followed by a 48-h incubation at 4°C to allow the dye to fully fill the processes. To reduce active transport of the dye from the cells during incubation (Rosen et al. 1991; Steinberg et al. 1987), the bathing ASW solution included 10 mM (final concentration) probenecid (Sigma). The living ganglia were cleared in 50% glycerol in ASW. Fluorescence was visualized with a Nikon fluorescence microscope. Backfills of the pleural-abdominal connectives were done using biocytin (Sigma), as previously described (Xin et al. 1999).

Immunocytology for serotonin combined with intracellular dye labeling

To label identified cells with dye, 5% Lucifer yellow was injected into cells by hyperpolarizing current. The ganglia were fixed in 4% paraformaldehyde, 0.1 M phosphate buffer (PBS; pH 7.4), 30% sucrose at room temperature for 2 h, then washed overnight with 0.1 M PBS, 30% sucrose at 4°C. The immunocytology was carried out using a modification of previously described techniques (Lloyd et al. 1985; Longley and Longley 1986). The fixed and washed ganglia were preincubated at room temperature for 2 h with 1% normal goat serum (NGS; Miles Science, Naperville, IL) in 0.1 M PBS to reduce nonspecific binding of primary antibody. The ganglia were then incubated in rabbit serotonin antiserum (Sigma) diluted 1/500 in PBS containing 1% NGS for 2 days at 4°C; washed in PBS at 4°C for 1 day, and incubated at 4°C for 1 day in goat anti-rabbit IgG rhodamine-conjugated Fab fragment diluted 1/50 (Cappel, Malvern, PA). Finally, the ganglia were washed in PBS again at 4°C for 1 day, then mounted on slides and coverslipped with Aqua-Poly Mount (Polyscience, Warrington, PA). The ganglia were viewed under a Nikon microscope equipped for epifluorescence and were photographed with Tri-X film. In one series of immunocytological studies, a rat antiserotonin antibody obtained from F. S. Vilim (Mt. Sinai School of Medicine) was used and the results were identical to those obtained with the rabbit antibody.

Controls for specificity of staining

Six ganglia were used for controls. Replacement of primary or secondary antibody did not result in staining. Dopaminergic cells that are revealed by formaldehyde-glutaraldehyde-induced fluorescence (Goldstein and Schwartz 1989) were not observed in the C cluster following fixation in paraformaldehyde.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

There are three CAIs in the C cluster of the cerebral ganglion

Backfills of the pleural-abdominal connectives revealed up to 50 cerebral neurons that send an axon into one or both the connectives. The number of cells filled was highly variable, but there appeared to be at least three relatively large neurons (50-100 µm) that typically were seen in the medial region of the cerebral C cluster. We designate these cells as CC neurons (CC2, CC3, and CC7) to refer to the fact that they are in the cerebral ganglion C cluster. The C cluster (Jahan-Parwar and Fredman 1976) consists of a compact group of cells (see Fig. 2) located in the caudal region of the dorsal surface of the cerebral ganglion, just rostral and slightly medial to the group of large neurons referred to as the B cluster (Jahan-Parwar and Fredman 1976; Teyke et al. 1989). As explained later, the cell we refer to as CC3 appears to be the same cell that was previously reported to be located in the cerebral B cluster (Hawkins 1989; Mackey et al. 1989). CC2 and CC3 are bilateral cells, whereas CC7 has only been found unilaterally in the right hemiganglion.

CC2 is a monopolar neuron that sends its axon to the ipsilateral cerebral-pleural connective. A main branch continues down into the abdominal ganglion via the pleural-abdominal connective. Another branch goes to the ipsilateral pedal ganglion and then travels to the contralateral pedal ganglion via the pedal-pedal commissure (Fig. 1A). CC3 is a bipolar neuron that sends one axon to the ipsilateral cerebral-pleural connective and the other to the contralateral cerebral-pleural connective (Fig. 1B). In the pleural ganglia, the main axons continue bilaterally to the abdominal ganglion. The axon sends branches to the pedal ganglia. CC7 is located in the right cerebral C cluster. The cell is presumed to be unilateral since it has never been encountered in the left hemiganglion despite extensive searches. CC7 sends its axon across to the contralateral pleural ganglion via the contralateral cerebral-pleural connective (Fig. 1C). In the pleural ganglion, the axon divides and sends one branch to the abdominal ganglion via the pleural-abdominal connective. The other branch travels to the ipsilateral pedal ganglion and continues to the contralateral pedal-pleural ganglion via the pedal-pedal commissure.

View larger version (39K): [in this window] [in a new window]   Fig. 1. Schematic drawings illustrating the position and axon distribution of 3 cerebral-abdominal interneurons located in the cerebral C cluster. The drawings are based on visualizations of neurons after fills with 5(6)-carboxyfluorescein dye. A: example of CC2 and its processes (n = 6). A symmetrical cell is present in each hemi-ganglion. CC2 sends its axon down to the ipsilateral pleural ganglion. The axon bifurcates and one branch enters the pleural-abdominal connective; the other branch enters the pleural-pedal commissure, continues through the ipsilateral pedal ganglion and then continues into the pedal-pedal commissure. The axon branch appears to terminate in the contralateral pedal ganglion. B: example of CC3(CB-1) and its processes (n = 8). A symmetrical cell is present in each hemi-ganglion. Shortly after leaving the cell body, the axon of CC3 bifurcates (see Fig. 2A). One axon enters the ipsilateral cerebral-pleural connective; the other axon enters the contralateral cerebral-pleural connective. In the pleural ganglion, each axon bifurcates and sends 1 branch to the pedal ganglion, and the other branch into the pleural-abdominal connective. C: example of CC7 and its processes (n = 8). CC7 has only been encountered in the right cerebral hemiganglion and is presumed to be a unilateral neuron. The axon of CC7 crosses to the contralateral cerebral ganglion and continues down the contralateral cerebral-pleural connective into the pleural ganglion and then, via the pleural-pedal commissure and the pedal-pedal commissure it continues into the ipsilateral and contralateral pedal ganglia. A branch of the axon then continues into the contralateral pleural ganglion where it appears to terminate. ATn, anterior tentacular nerve; C-BC, cerebral buccal commissure; C-PC, cerebral-pedal commissure; C-PlC, cerebral-pleural commissure; LLABn, lower labial nerve; P, pedal nerve; Pl-AbC, pleural-abdominal connective; P-P comm., pedal-pedal commissure; ULABn, upper labial nerve.

Neuron CC3 is immuno-positive for serotonin

The morphology of CC3 was studied by means of carboxyfluorescein or Lucifer yellow dye injection Fig. 2A shows Lucifer yellow labeling of CC3, which is shown at its typical location within the C cluster. The label clearly shows CC3 has a short initial segment which bifurcates into two axons. At higher magnifications, the axons could be traced into the ipsilateral and contralateral cerebral-pleural and pleural-abdominal connectives.

View larger version (36K): [in this window] [in a new window]   Fig. 2. CC3 is immunopositive for serotonin. The figure shows a dorsal view of the right anterior and middle region of the cerebral ganglion. CC3 was intracellularly labeled with Lucifer yellow and the preparation was then stained for serotonin. A: visualization with a Lucifer yellow filter set revealed CC3 () and its characteristic bifurcating axon. B: visualization with a rhodamine filter set revealed cells that are immunopositive for serotonin. The large immunopositive cell in the anterior region of the ganglion (top) is likely the metacerebral cell (MCC). A small number of cells, up to five, in each C cluster were immunopositive. In this example, 2 serotonin-positive cells were stained in each C cluster. One of the cells in the right cluster () coincides with the Lucifer-filled CC3 neuron. Low-level background staining reveals the other neurons that comprise the right C cluster (for orientation within the ganglion see the position of CC3 indicated in Fig. 1B). The left border of B shows 2 immunopositive cells located in the left cerebral C cluster. Based on its position, the lower, stained cell in the left hemiganglion is likely to be the left homologue of CC3 (n = 5, calibration bar: 100 µm).

Intracellular injection of CC3, CC2, or CC7 with Lucifer yellow, followed by immunostaining for serotonin, revealed that CC3 was the only cerebral-abdominal interneuron in the C cluster that was immunopositive for serotonin. Figure 2B shows an example in which CC3 was first identified by its characteristic morphology, position, and synaptic connections. The cell was then labeled with Lucifer yellow and subsequently shown to be stained with a serotonin antibody (Fig. 2B). As many as five cells immunopositive for serotonin (5-HT) were stained in the C cluster (cf. Wright et al. 1995), but neither CC2 nor CC7 was immunopositive. CC3 is the largest, most posterior cell of the five serotongeric cells in the C cluster. The 5-HT staining pattern (up to 5 immunopositive cells) was similar to those observed in two other gastropod mollusks, Pleurobranchaea and Tritonia (Sudlow et al. 1998). In one experiment, we labeled both the right and left CC3 with Lucifer yellow and found that they both subsequently stained for serotonin.

In a previous study, it was reported that there is a serotonergic cerebral-abdominal interneuron that is located in the B cluster of the cerebral ganglion and consequently was named CB-1 (Hawkins 1989; Mackey et al. 1989). Since CC3 appeared to have the same distinctive axonal morphology as CB-1 (Mackey et al. 1989; Wright et al. 1995), we sought to determine if there might be two similar serotonergic cells, one located in the C cluster and a second in the B cluster. The B cluster is composed of relatively large neurons spaced relatively far apart, whereas the C cluster is a compact group of small and medium sized cells located just anterior to the B cluster cells. Provided that there is no significant damage to the cells during desheathing, the C cluster forms a clearly demarcated distinct group. To determine if there is a serotonergic B cluster cell that might have properties similar to CC3, we stained for serotonin in a series of eight ganglia. We found, however, that in none of these ganglia were any serotonin positive cells found in the B cluster. In all cases, serotonergic neurons that were located on the dorsal surface of the posterior region of the cerebral ganglion were found to be restricted to the C cluster. Consequently we conclude that CC3 is likely to be identical to the previously described CB-1 neuron. We suggest (see DISCUSSION) that the full designation of CC3 be CC3(CB-1).

All three CAI cells in the cerebral C cluster get inputs from stimuli applied to the head and lips

Feeding behavior in Aplysia can be evoked by tactile and/or chemical stimuli applied to the anterior regions of the animal (Kupfermann 1974). We found that stimulation of the lips or anterior tentacles with a piece of moist seaweed or a glass rod evoked excitatory synaptic input into CC2 (Fig. 3, A and B). There was no significant difference between touching the lip or tentacle with a piece of moist seaweed compared with a polished glass probe. The responses evoked by inputs to CC2 occurred with a relatively short latency (less than 100 ms), well before any sign of a rhythmic buccal ganglion responses, which typically have latencies of 3-10 s. CC3 also receives excitatory input when the lip is stimulated (Fig. 4, A and B), whereas CC7 is inhibited (Fig. 5A). CC2 and CC3 were similar not only with regard to their responses to stimuli applied to the head, but they also received similar input during evoked buccal-mass movements that were initiated by applying brief tactile stimuli to the buccal mass. During evoked cycles of forward and backward movement, both CC2 and CC3 became tonically active (Figs. 3C and 4C). Electrical stimulation of buccal nerve 2, which can elicit rhythmic buccal nerve activity or buccal mass movements (Nargeot et al. 1997), also produced tonic activity in CC2 and CC3 during the time that rhythmic buccal activity occurred as judged by buccal nerve recordings (Figs. 3D and 4D). In contrast to CC2 and CC3, rhythmic buccal activity was associated with a different pattern of input to CC7. CC7 showed phasic activity. Spiking in CC7 was absent during the phase of each retraction movement of the buccal mass or radula (Fig. 5B). We also determined if CC7 fired phasically during a nonfeeding rhythmic program, namely a locomotor program. To examine the firing of CC7 during rhythmic locomotor programs, fictive locomotion was elicited by electrical stimulation of pedal nerve 9 (P9). Fictive locomotion was monitored by extracellular recording from P11 (also called the pedal artery shortener nerve). This nerve contains the axon of the pedal artery shortener neuron, which fires in phase with locomotor programs (Skelton and Koester 1992; Xin et al. 1996b). In contrast to the strong phasic activity seen during buccal mass movements, CC7 showed tonic firing following the stimulation of P9, and it continued to fire during the fictive locomotor program (Fig. 5C). The firing of CC7 showed some variation, but this was not clearly phase locked to the locomotor rhythm recorded in P11.

View larger version (22K): [in this window] [in a new window]   Fig. 3. CC2 receives sensory inputs from the lips and buccal ganglion. A: in a simplified preparation consisting of the lips and the buccal mass together with the cerebral and buccal ganglia, touching the lips with a moistened piece of seaweed () produced excitation in CC2 (n = 4). B: when CC2 was hyperpolarized, touching the lips evoked numerous fast excitatory postsynaptic potentials (EPSPs, n = 4). C: touching the buccal mass () evoked buccal mass movements and an increase in the tonic firing of CC2 (n = 3). D: electrical stimulation of buccal nerve 2 (BN2) evoked a buccal motor program as evidenced by rhythmic activity recorded with an extracellular electrode on buccal nerve 2. CC2 showed a burst of activity immediately following the shock, and then showed irregular spiking not distinctly in phase with the BN2 activity (n = 3).

View larger version (28K): [in this window] [in a new window]   Fig. 4. CC3 receives inputs from the lips and buccal ganglion. A: touching the lips with a piece of moist seaweed evoked excitation in CC3 (n = 4). B: when CC3 was hyperpolarized, touching the lip evoked numerous fast EPSPs (n = 4). C: during buccal mass forward and backward movements evoked by direct touch of the buccal mass, CC3 fired tonically (n = 3). D: electrical stimulation of buccal nerve 2 (BN2) evoked rhythmic activity in buccal nerve 2 and irregular firing of CC3 not distinctly in phase with BN2 activity (n = 4).

View larger version (43K): [in this window] [in a new window]   Fig. 5. CC7 receives inhibitory inputs from the lips and buccal ganglion. A: touching the lips with a piece of moist seaweed () evoked brief inhibition of CC7 spiking (n = 4). B: in preparations of head ganglia with the buccal mass attached, during buccal mass forward and backward movements, CC7 exhibited activity that was in phase with buccal mass movements. CC7 spiking strongly decreased during the retraction phase of the radula and buccal mass movement (observed visually and marked with horizontal lines; n = 3). C: electrical stimulation of pedal nerve 9, which produces a locomotor program, increased CC7 firing rate. P11, which contains the axon of the pedal artery shortener neuron, reflects the rhythmic activity of a fictive locomotor program. The firing of CC7 was not distinctively in phase with the fictive locomotor program (n = 3).

Table 1 summarizes the basic morphological and input features of the cerebral-abdominal interneurons in the cerebral C cluster.

Firing of cerebral C cluster CAI cells influences cells that appear to be involved in buccal mass activity

To explore the possible involvement of the C-cluster CAIs in feeding, we fired individual cells and recorded possible spike activity of units that project to the buccal ganglion via the cerebral-buccal (c-b) connectives. Firing of the CAIs evoked activity of several cells that were recorded extracellularly in the ipsilateral and contralateral c-b connectives (Figs. 6-8). The evoked activity may be mediated by interneurons in the pleural-pedal or abdominal ganglia since it was eliminated by cutting the cerebral-pleural and -pedal connectives. Additional evidence for a possible role of the CAIs in feeding was the finding that CC7 inhibited C-PR and in turn was inhibited by firing of cerebral-pedal regulator neuron (C-PR) (Fig. 9). C-PR is involved in generating aspects of appetitive feeding responses such as the head-up posture and lengthening of the neck (Teyke et al. 1990).

View larger version (26K): [in this window] [in a new window]   Fig. 6. Firing of CC2 evokes neural activity in the cerebral-buccal connectives (n = 4). In this example, the left CC2 (LCC2) was fired.

View larger version (21K): [in this window] [in a new window]   Fig. 7. Firing of CC3 evokes neural activity in the cerebral-buccal connectives (n = 4). In this example, the right CC3 (RCC2) was fired.

View larger version (25K): [in this window] [in a new window]   Fig. 8. Firing of CC7 evokes neural activity in the cerebral-buccal connectives (n = 3). Since CC7 is located in the right hemiganglion, the RCBC represents the ipsilateral connective. R, right; L, left; CBC, cerebral-buccal connective.

View larger version (25K): [in this window] [in a new window]   Fig. 9. CC7 and cerebral-pedal regulator neuron (C-PR) mutually inhibit one another. A: effect of firing CC7 on C-PR. B: effect of firing C-PR on CC7. Discrete synaptic potentials one for one with CC7 or C-PR spikes were not observed in these experiments (n = 3).

Effects of CC2 on left upper quadrant cells and other neurons in the abdominal ganglion

Firing of CC2 was found to excite or inhibit numerous cells in the abdominal ganglion. Some of the cells that are excited by CC2 apparently send axons into the genital nerve since brief firing of CC2 evoked a prominent burst of unit activity that was recorded from the genital nerve (Fig. 10). Among cells that send their axons into the genital nerve, we found that CC2 produced one-for-one excitatory postsynaptic potentials (EPSPs, Fig. 11A) in the left upper quadrant cells (LUQs), which appear to be motor neurons for the renal pore and cause it to close (Koester and Alevizos 1989). The EPSPs persisted in a high divalent cation solution (Fig. 11B), suggesting that CC2 is monosynaptically connected to LUQ cells.

View larger version (32K): [in this window] [in a new window]   Fig. 10. Firing of CC2 evokes a burst of neural activity recorded in the genital nerve of the abdominal ganglion (n = 4).

View larger version (12K): [in this window] [in a new window]   Fig. 11. CC2 monosynaptically excites abdominal ganglion left upper quadrant (LUQ) cells. A: firing of CC2 produced 1-for-1 EPSPs in an LUQ cell. The LUQ cell was slightly hyperpolarized to prevent spontaneous firing and enhance the size of the EPSP. Note that because of the long distance between the cerebral and abdominal ganglia, the EPSPs had a long latency. B: the EPSPs persisted in a high divalent cation solution (3× Mg2+, 3× Ca2+; n = 6).

CC2 strongly excited the L9 gill motor neurons (Fig. 12). The excitation was associated with the presence of EPSPs that were not one-for-one with the CC2 spikes and outlasted the firing of CC2. Thus the excitation that CC2 produced was likely due to the activity of one or more interneurons. Since CC2 appeared to be the only C cluster cerebral-abdominal interneuron that strongly excited L9 cells, the presence of excitatory input to L9 could be used as one of the identifying features of CC2.

View larger version (22K): [in this window] [in a new window]   Fig. 12. CC2 has strong excitatory effects on gill motor neuron L9 cells. Firing of CC2 strongly excited both L91 and L92 (n = 5). The excitation was associated with the occurrence of fast EPSPs that were not 1-for-1 with CC2 spikes and outlasted the firing of CC2, indicating that the excitation is very likely polysynaptic (n = 3). Because CC2 appears to be the only neuron in the C cluster of the cerebral ganglion found to strongly excite L9 cells, L9 was often used to identify CC2.

Since L9 is excited as part of a defensive reflex (Kupfermann and Kandel 1969), we recorded from L9 together with cerebral Bn (B cluster, narrow spike) cells, which also appear to be involved in defensive responses (Teyke et al. 1989). We found that while L9 was excited by CC2, Bn cells exhibited a polysynaptic inhibition (Fig. 13A). Similarly, simultaneous recording from L9 and the gill motor neuron LDG revealed that when L9 was excited, LDG was inhibited (Fig. 13B). Thus both the excitation of L9 and the inhibition of LDG and Bn outlast CC2's firing for the same duration, consistent with the possibility that there is an interposed interneuron presynaptic to all of these cells.

View larger version (24K): [in this window] [in a new window]   Fig. 13. CC2 inhibits cells involved in withdrawal responses. A: firing of CC2 inhibits Bn cells, cerebral ganglion neurons that are involved in head withdrawal responses (n = 3). B: CC2 inhibits LDG cells, which are abdominal ganglion neurons that are involved in gill withdrawal. In these experiments an L9 cell was also recorded and showed its characteristic excitation in response to firing of CC2 (n = 4).

Firing of CC2 appeared to modestly increase the rate of firing of the "spontaneous" EPSP (Fig. 14A) that can be recorded in bursting neuron R15 (Koester et al. 1993; Kupfermann 1970). The EPSPs were not one for one with the CC2 spikes, suggesting that the effect was polysynaptic. It was previously shown that one source of rhythmic EPSPs to R15 is interneuron XIII, which is located in the right pedal ganglion (Koester et al. 1993). To test whether the excitatory effect of CC2 on R15 might be mediated by interneuron XIII, we cut the right pleural-abdominal connective that is the pathway by which interneuron XIII extends its axon to the abdominal ganglion. When the connective was cut, as expected, spontaneous EPSPs in R15 were no longer present. Nevertheless firing of CC2 still excited R15 even in the absence of any prominent fast EPSPs (Fig. 14B). Another possible neuron that could mediate an excitatory effect of CC2 onto R15 is L40, an abdominal ganglion interneuron that excites interneuron XIII as well as L9 cells (Koester et al. 1993). We found that firing of CC2 excited L40 (Fig. 15A). As revealed by hyperpolarization of L40, the excitation evoked by CC2 did not appear to be monosynaptic but was associated with a burst of EPSPs that were not one for one with CC2 spikes (Fig. 15B). The excitatory effect on the presumptive interneuron that excites L40 was not invariably observed. For example in one series of runs, CC2 affected L40 in only approximately 50% of the trials.

View larger version (13K): [in this window] [in a new window]   Fig. 14. Firing of CC2 excites the abdominal ganglion R15 neuron. A: R15 usually exhibits regular rhythmic bursting in the isolated abdominal ganglion; firing of CC2 increased the firing rate of a spontaneously firing excitatory interneuron that is often seen in recordings from R15 (n = 3). One type of EPSP seen in R15 is known to be produced by activity of interneuron XIII, which is a unilateral cell, located in the right pedal ganglion and sends its axon to the abdominal ganglion only via the right pleural-abdominal connective. B: cutting the right connective eliminated the EPSP in R15 neuron as seen in the figure, but firing of CC2 still excited R15, suggesting that the effect of CC2 on R15 may not only involve interneuron XIII (n = 3).

View larger version (10K): [in this window] [in a new window]   Fig. 15. The effect of CC2 on L9 may involve L40, an abdominal ganglion interneuron that excites L9. A: firing of CC2 produced a strong repeatable excitatory input onto L40. B: when L40 was strongly hyperpolarized, it was revealed that firing of CC2 increased the rate of a fast EPSP in L40, suggesting that CC2 excites an interneuron which in turn excites L40 (n = 2).

Effects of CC3 on L10, LUQ cells and other abdominal ganglion cells

Firing of CC3 was found to produce effects on neurons that project out to various nerves including pedal-ganglion nerves such as pedal nerve 3 and abdominal ganglion nerves, most prominently the genital nerve (Fig. 16). Examination of identified neurons in the abdominal ganglion revealed that one neuron that receives prominent excitation when CC3 is fired, is L11 (Fig. 17), a cell that sends a large axon into the genital nerve (Koester and Kandel 1977). The EPSP evoked in L11 has a slow onset and decay. Fast trace recordings showed it was not associated with any obvious fast potentials (Fig. 17D). When CC3 was fired in bursts every 2-3 s, the evoked slow EPSP in L11 was little changed from burst to burst (Fig. 17A). CC3 evoked a similar slow excitation of abdominal ganglion interneuron L10 (Fig. 17, B and C).

View larger version (35K): [in this window] [in a new window]   Fig. 16. Firing of CC3 evokes activity in peripheral nerves of the abdominal and pedal ganglia. This figure shows a simultaneous recording from CC3 with pedal ganglion nerve 3 and abdominal ganglion genital nerve (n = 4).

View larger version (28K): [in this window] [in a new window]   Fig. 17. CC3 excites identified neurons in the abdominal ganglion. A: excitatory effect of firing of CC3 on L11. B: excitatory effect of firing of CC3 on L10. C and D: the recordings at faster sweep speeds; the excitatory input evoked in L10 and L11 is not associated with any visible fast EPSPs, suggesting that the cells are excited by a slow potential (n = 6).

L10 is known to monosynaptically inhibit LUQ cells (L2-L5) (Frazier et al. 1967) in which it evokes a characteristic dual-phase (fast/slow) inhibitory postsynaptic potential (IPSP). We found that firing of CC3 evoked a similar IPSP in LUQ cells (Fig. 18). Typically after a variable delay of approximately 100-400 ms, small discrete IPSPs, not one-for-one with CC3 spikes, were observed together with a large, slow hyperpolarization.

View larger version (24K): [in this window] [in a new window]   Fig. 18. CC3 inhibits abdominal ganglion LUQ cells. As shown in the figure, in some instances CC3 produced fast IPSPs without a prominent slow phase. In other LUQ cells, CC3 evoked IPSPs that exhibited a prominent summating slow component in the absence of fast components (see Fig. 21A; n = 5). We have not attempted to analyze the source of these different effects, which could be characteristic of the specific LUQ cell (Evans et al. 1991) or could be related to some experiment-specific property of the LUQ cell, such as its membrane potential.

CC3 is present bilaterally in the cerebral ganglion. Because the left and right abdominal hemiganglia are highly asymmetrical, we examined whether there were any major differences between the ipsilateral and contralateral connections of CC3. For these experiments, both the left and right CC3 cells were impaled in the same ganglion, and their effects on selected abdominal ganglion cells were examined. We found that both the left and right CC3 inhibited LUQ cells (Fig. 19A), and both the left and right CC3 cells excited L10 (Fig. 19B).

View larger version (17K): [in this window] [in a new window]   Fig. 19. CC3 neurons have effects on abdominal ganglion cells ipsilateral and contralateral to the CC3 cell. A: LUQ cells receive inhibitory input from both the ipsilateral and contralateral CC3 cells. In this example, the LUQ cell exhibited the rhythmic bursting that is often seen in LUQ cells. B: L10 cells receive excitatory input from both the ipsilateral and contralateral CC3 (n = 3).

To test for the possible monosynapticity of the actions of CC3 on L10, L11, and LUQ cells, we examined it actions in a high divalent cation solution. In this solution, the slow excitation to L10 persisted (Fig. 20, A and B), but the excitation to L11 and inhibition of LUQ cells did not persist (data not shown). These results are consistent with the conclusion that the CC3 connection to L10 may be monosynaptic, whereas the effects of CC3 on the LUQ cells are polysynaptic, arising, at least in part, from the firing of L10.

View larger version (15K): [in this window] [in a new window]   Fig. 20. The synaptic connection of CC3 to L10 persists in the presence of a high divalent cation solution. A: in normal artificial seawater (ASW), firing of CC3 excited L10 strongly. B: when the abdominal ganglion was bathed in a high divalent cation solution (3× normal Mg2+ and 3× normal Ca2+), firing of CC3 still depolarized the membrane potential of the L10 neuron, suggesting that CC3 produces a slow monosynaptic excitatory input to L10. The decreased firing of L10 in the high divalent cation solution, compared with ASW, may be a consequence of an increased threshold for spiking in the presence of the high divalent cations (n = 3).

Another abdominal ganglion neuron that receives synaptic input from L10 is R15 (Kandel et al. 1967; Koester and Kandel 1977), and as expected, R15 was excited by firing of CC3. When R15 was in its burst mode, firing of CC3 produced a shortening of the interburst interval (Fig. 21A). In fast sweep speed recordings in which the R15 neuron was slightly hyperpolarized to suppress spontaneous spiking, firing of CC3 resulted in spiking of R15 and increased the frequency of a spontaneous EPSP in the cell (Fig. 21B).

View larger version (14K): [in this window] [in a new window]   Fig. 21. CC3 produced weak excitation of R15. A: firing of CC3 decreased the interburst interval of R15. An LUQ cell was simultaneously recorded to aid in identifying CC3. B: when R15 was hyperpolarized, firing of CC3 revealed an increase in the firing of an excitatory interneuron (n = 5). C: when interneuron XIII (which fires rhythmically and directly excites R15) was hyperpolarized, firing of CC3 still increased the firing rate of R15 even in the absence of any visible fast EPSPs (n = 2).

In addition to excitatory input from L10, R15 is excited by interneuron XIII, a cell located in the right pedal ganglion (Koester et al. 1993). When interneuron XIII and R15 neurons were recorded from simultaneously, we found that although firing of CC3 excited R15, it had no effect on interneuron XIII (Fig. 21C).

Another neuron excited by L10 is the heart motor neuron RB_HE (Mayeri et al. 1974). As expected, firing of CC3 resulted in excitatory input to RB_HE (Fig. 22A). In addition, we found that CC3 may weakly excite an R20 cell (Fig. 22A) and LBvc (Fig. 22B); and it inhibited L9 (Fig. 22B). None of these effects were associated with one-for-one synaptic potentials.

View larger version (20K): [in this window] [in a new window]   Fig. 22. Firing of CC3 had synaptic effects on a diverse set of neurons in the abdominal ganglion. A: CC3 excited the heart exciter cell, RBHE, and produced weak excitation of a cell identified as R20. B: CC3 evoked excitatory input to an LBvc cell, while it inhibited an L9 cell (n = 3).

Effects of CC7 on LBvc cells and other abdominal cells

In a semi-intact preparation that included the organs innervated by the abdominal ganglion, firing of CC7 was observed to evoke a large contraction of the abdominal artery (Fig. 23). The contraction appeared to involve the circular muscles of the artery since the diameter of the artery was reduced markedly and there was a virtual shut-down of the flow of fluid that was perfused through the artery. Because the abdominal artery is innervated by the LBvc motor neurons in the abdominal ganglion (Mayeri et al. 1974), we tested for a connection between CC7 and the LBvc cells. We found that CC7 excites all three LBvc cells. Figure 24 shows an example of the EPSP that CC7 produced in a LBvc cell. The EPSPs appeared to be monosynaptic. They were one-for-one with the CC7 spikes and had a constant latency. Furthermore the EPSPs persisted in a high divalent cation solution (Fig. 24B).

View larger version (17K): [in this window] [in a new window]   Fig. 23. Firing of CC7 evoked strong contraction of the abdominal artery. CC7 was intracellularly stimulated at 15 Hz. The contraction of the abdominal artery was monitored by an isotonic displacement transducer (n = 6).

View larger version (14K): [in this window] [in a new window]   Fig. 24. CC7 excites LBvc cells, which are motor neurons for the abdominal artery. A: firing of CC7 produced 1-for-1 EPSPs in an LBvc cell in normal seawater. B: the EPSPs persisted in a high divalent cation solution (n = 5).

In addition to evoking contractions of the abdominal artery, firing of CC7 resulted in contraction of the heart, and we found that CC7 produced one for one EPSPs in the heart motor neuron, RB_HE (Fig. 25A). The EPSPs in the RB_HE cell persisted in a high divalent cation solution (Fig. 25B), suggesting that CC7 monosynaptically excites this neuron.

View larger version (15K): [in this window] [in a new window]   Fig. 25. CC7 excited RBHE, a heart excitatory neuron. A: in normal seawater, firing of CC7 produced 1-for-1 EPSPs in RBHE. B: the EPSPs persisted in a high divalent cation solution (n = 3).

Although CC7 appears to exert its main actions on elements of the cardiovascular system, it also has effects on other abdominal neurons that do not appear to be directly involved in cardiovascular function. It produced a weak slow excitation of LUQ cells (Fig. 26). It also evoked a multiphasic synaptic input (excitatory-inhibitory-excitatory) to R20. In addition to affecting abdominal ganglion neurons it has diverse actions on numerous neurons located in the pedal (Fig. 27, A and B) and pleural ganglia (Fig. 28, A and B).

View larger version (10K): [in this window] [in a new window]   Fig. 26. Firing of CC7 excited LUQ cells and produced signs of a weak complex input to neuron R20 (n = 3). The traces from the 3 cells were obtained in different experiments, and a train of typical frequency and duration of CC7 spikes is shown in the bottom trace.

View larger version (20K): [in this window] [in a new window]   Fig. 27. Firing of CC7 excites or inhibits pedal ganglion neurons. A: example of a pedal neuron (Pn1) that was excited by firing of CC7 (n = 7). B: example of a pedal neuron (Pn2) that was inhibited by firing of CC7 (n = 7).

View larger version (21K): [in this window] [in a new window]   Fig. 28. The observed effects of CC7 on pleural neurons were all excitatory. A: some actions of CC7 were associated with a slow EPSP in pleural neurons. B: some unidentified pleural ganglion cells exhibited a short-latency excitation, and hyperpolarization of these pleural neurons revealed that CC7 could evoke short-latency fast EPSPs (n = 6).

The synaptic outputs of the cerebral-abdominal interneurons (CAIs) to different functional systems and identified cells in the abdominal ganglion are summarized in Table 2.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The overall data (Tables 1 and 2) suggest that although each of the CAIs in the cerebral C cluster has distinctive properties, they also have certain shared features. Thus the CAIs may each have individual distinctive behavioral functions but may also be active conjointly in certain similar behaviors.

The properties of CC2 and CC3 suggest that they are more closely related to each other than to CC7. All three CAIs receive synaptic inputs when the lips and tentacles are contacted with relatively gentle tactile stimulation such as that provided by sea weed. The input, however is excitatory for CC2 and CC3 but is inhibitory for CC7. The firing of the three CAIs produces activation, albeit weak, of neurons that send an axon into the cerebral-buccal connective. It is likely that the activated cells are cerebral-buccal interneurons (CBIs), and it remains to be determined if the same set of cells are affected by each of the three CAIs.

The CAIs all alter the activity of abdominal ganglion neurons, but examination of the specific neurons affected indicates that each CAI influences a distinctive group of cells. Each of the three cells has a prime site of action, involving monosynaptic connections (indicated in bold in Table 2) and in addition affects overlapping groups of cells by means of polysynaptic connections.

CC2 and CC3 may affect kidney function

CC2 has monosynaptic excitatory connections to the LUQ cells, whereas CC3 has a monosynaptic connection to interneuron L10 (Frazier et al. 1967) and indirectly inhibits LUQ cells, at least in part by means of the inhibitory connections that L10 makes onto LUQ cells (Kandel et al. 1967). CC2 and CC3 may be an example of an "antipodal" pair of interneurons, that is, pairs of interneurons that exert actions that are largely opposite in sign, as is the case for example of the CC6/CC5 pair of cerebral neurons (Xin et al. 1996b, 2000). The group of LUQ cells and L10 also exert opposing actions on the renal pore. LUQ cells promote closing of the kidney pore and L10 promotes opening of the kidney pore (Koester and Alevizos 1989). Thus firing of CC2 would be expected to promote closing of the kidney pore, whereas firing of CC3 should promote opening of the kidney pore. CC2 and CC3 also have opposing polysynaptic actions on gill motor neurons. CC2 evokes excitatory inputs into L9 cells, while CC3 evokes inhibitory inputs to L9 cells.

In these experiments, we did not specifically attempt to examine the significance of serotonin as a possible neurotransmitter of CC3. In Aplysia it is now clear that serotonin is involved in a variety of functions, including roles in both aversive (Clark and Kandel 1993; Sun and Schacher 1998) as well as appetitive behaviors (Lloyd et al. 1984). Given the suggestion that CC3 may be involved in some aspects of feeding behavior, it would be interesting to explore the role of serotonin on neurons that may be involved in ingestive aspects of feeding or perhaps in defensive behaviors that might compete with feeding (Jing and Gillette 2000).

In previous work, a serotonergic cell termed CB-1 was suggested to be involved in heterosynaptic facilitation of the EPSPs of siphon sensory neurons in the abdominal ganglion and to be involved in dishabituation and sensitization of gill and siphon withdrawal responses (Hawkins 1989; Mackey et al. 1989; Wright et al. 1995). Because in our studies, CC3 was always found in the C cluster and never in the B cluster, we assumed that this was not CB-1 and was therefore a previously unidentified neuron. Our attempts to locate a serotonergic neuron in the B cluster, however, were unsuccessful, and based on its overall physiological and morphological characteristics, we have concluded that CC3 is very likely identical to CB-1. In fact, CC3 is located in the posterior region of the C cluster, adjacent to the anterior region of the B cluster and might have been visualized as part of the B cluster. Because the prefix "CB" implies CB-1 is in the B cluster, to avoid possible confusion about the location of the cell, we suggest that the full designation for the cell be CC3(CB-1).

Because CC3(CB-1) may be involved in defensive responses elicited by noxious stimuli, it is possible that its actions on cerebral-buccal interneurons reflects a role in egestive feeding responses that occur when the animal encounters noxious or inedible food stimuli. It is not possible, however, to exclude multiple roles for this cell, such as a general arousal function as suggested for serotonergic neurons in Pleurobranchaea (Jing and Gillette 2000). For example, it might be involved in aspects of appetitive arousal in response to relatively mild stimuli applied to the lips but might be involved in defensive arousal in response to strong or noxious stimuli. The putative cerebral-buccal interneurons that are excited by the CAIs have not been identified so that it is not known what the likely buccal program or programs are associated with activity of the CAIs. Indeed cerebral-buccal interneurons appear to be involved in both ingestive (Perrins and Weiss 1998; Rosen et al. 1991) as well as egestive (Rosen et al. 1991, 1998) buccal motor programs.

The fact that the serotonergic neuron CC3 in turn excites another serotonergic neuron RB_HE as well as neurons involved in diverse behaviors is reminiscent of an organization of a serotonergic arousal network that may be present in Pleurobranchaea (Jing and Gillette 2000) and Clione (Satterlie and Norekian 1996). In gastropod mollusks, a limited number of serotonergic cerebral neurons such as As4 in Pleurobranchaea excite other serotonergic neurons and may provide a form of general arousal (Jing and Gillette 2000). CC3 in Aplysia may be analogous or perhaps even homologous to cells such as As4 of Pleurobranchaea (Jing and Gillette 1999), dorsal swim interneurons of Tritonia (Sudlow et al. 1998), cerebral ganglion Cd3 cell in Lymnaea (Croll and Chiasson 1989), and CPA1, CPB1 in Clione (Panchin et al. 1995).

CC7 affects the vasoconstrictor/heart system

The overall pattern of inputs and outputs of CC7 suggests that it plays a functional role quite different from that of CC3 and CC2. CC7 has a strong monosynaptic excitatory connection to vasoconstrictor neurons, and its firing can evoke a substantial constriction of the abdominal aorta. Furthermore CC7 excites the serotonergic heart motor neuron RB_HE. CC3 can also excite RB_HE, but its actions are polysynaptic, mediated most likely by its excitation of L10. CC7, by contrast, appears to monosynaptically excite RB_HE. An important functional difference between CC7 and CC2/CC3 is the nature of the synaptic inputs they receive. Tactile stimuli to the head evoke inhibition in CC7 instead of excitation as seen in CC2 and CC3. Furthermore the synaptic inputs CC7 receives in association with buccal programs are strongly phasic, whereas the inputs to CC2 and CC3 are tonic and not distinctly correlated with the phase of the buccal program. Since the firing of CC7 produces a substantial constriction of the abdominal aorta, CC7 has the properties appropriate for a cell that can mediate or at least contribute to the previously described correlation between rhythmic variations of blood flow and movements of the buccal mass that occur during feeding (Koch et al. 1984). It has been suggested that this gating of blood flow during buccal mass movements may aid in the generation of functional biting movements (Koch and Koester 1982; Koch et al. 1984). Thus CC7 may provide a means by which a somatic neural control system is coordinated with visceral responses that provide support for the somatic behaviors [see also data on neuron CLE1 in Clione (Arshavsky et al. 1992)]. In this sense, CC7 may function to optimize feeding movements and therefore can be considered an important component of the neural basis of food-induced arousal. It is significant that although CC7 fires in a strongly rhythmic mode during a rhythmic feeding motor program, it fires in a tonic mode during a rhythmic locomotor program. Thus CC7 may provide another example of a cell such as the previously described cells CC5 and CC6 (Xin et al. 1996b, 2000), that are active in multiple behaviors, but play different roles in the various behaviors.

CC7 may be identical to a previously described neuron in Aplysia termed CB2 (Wright et al. 1995). CB2 is located near CC3(CB-1). Like CC7, CB2 is not serotonergic and sends its axon into the contralateral pleural-abdominal connective. It is not known whether CB2 is present only unilaterally, and because its synaptic connections have not been described, it is not possible to definitively identify it as identical to CC7.