Identification and Characterization of a Multifunction Neuron Contributing to Defensive Arousal in ApZysia

Transcription

1 JOURNALOFNEUROPHYSIOLOGY Vol. 70, No. 5, November Printed in U.S.A. Identification and Characterization of a Multifunction Neuron Contributing to Defensive Arousal in ApZysia LEONARD J. CLEARY AND JOHN H. BYRNE Department of Neurobiology and Anatomy, University of Texas Medical School at Houston, Houston, Texas SUMMARY AND CONCLUSIONS 1. The tail withdrawal reflex is mediated by a monosynaptic circuit composed of tail sensory and motor neurons, but there appear to be additional neuronal elements that also contribute to the reflex. A newly identified interneuron, called, was located in the pleural ganglion. This neuron formed a parallel excitatory pathway between sensory and motor neurons. The distinguishing feature of LPI 17 was its ability to elicit a long-lasting ( s) excitatory postsynaptic potential (EPSP) in the motor neuron. 2. Intracellular labeling of LPI 17 revealed axons projecting to the cerebral and abdominal as well as the pedal ganglia. Simultaneous intracellular recordings confirmed the widely divergent output of LPI 17 to tentacle motor neurons in the cerebral ganglion, as well as to gill, siphon, and ink motor neurons in the abdominal ganglion. 3. Also receiving input were abdominal neurons L29, which excites LF, motor neurons and facilitates LE sensory neurons, and L25, which is part of the pattern-generating network underlying respiratory pumping. Thus LPI 17 appears to be a neural element important for the conduction of information about tail stimulation to ganglia that are not innervated by tail sensory neurons themselves. Moreover, the divergent outputs suggest that LPI 17 is an element of a neural circuit underlying defensive arousal. 4. LPI 17 produced slow EPSPs in several motor neurons. The long time course of the EPSP could prolong the burst in the motor neuron produced by LPl17 itself as well as increase the effectiveness of coincident synaptic input. This suggests that an important function of this interneuron is to extend the duration of the response to tail stimulation in the motor neuron. This could account for the relatively long time course of the motor neuron response to tail stimulation compared with that of the sensory neuron. 5. Sensitization is a form of nonassociative learning that produces changes in the amplitude and duration of reflex responses. It seems unlikely that all of these changes can be attributed to enhanced amplitude of the sensory-motor synapse, however. Therefore LP117 may itself be a site of plasticity for reflexes elicited by tail stimulation. INTRODUCTION Reflexes have proven to be particularly suitable for analysis of the cellular mechanisms of behavior because the reflex response is closely coupled to the stimulus and the underlying neuronal circuits tend to be composed of few elements (Baldissera et al ; Kandel 1979). Nevertheless, even simple reflexes may be integrated into larger behavioral responses. This concept is apparent when studying defensive withdrawal reflexes in the marine mollusk Aplysia californica. For example, weak mechanical stimulation of the tail elicits a coordinated contraction of the tail itself (Walters et al. 1983a) as well as the siphon and gill, which are located in the mantle cavity (Byrne et al ; Carew et al. 1983; Scholz and Byrne 1987; Walters and Erickson 1986 ). Escape locomotion may also be elicited (Hening et al. 1979; Mackey and Carew 1983). A more intense stimulus may elicit the release of ink and opaline from glands in the mantle cavity and mucus from glands in the skin (Carew and Kandel1977; Denny 1989; Rayport et al. 1983; Scholz and Byrne 1987; Tritt and Byrne 1980). Some elements of the circuitry underlying these behaviors have been described. The cell bodies of the primary mechanoafferents innervating the tail are located in a discrete, readily identified cluster in the pleural ganglion (Walters et al. 1983a). A tail sensory neuron (TSN) has one large axon that projects to the pedal ganglion, and fine neurites arise in both pleural and pedal ganglia (Clear-y and Byrne 1984; Nazif et al ). These neurons make monosynaptic connections onto tail motor neurons ( TMNs), located in a second cluster of cells in the pedal ganglion (Walters et al. 1983a). This monosynaptic circuit may be sufficient to mediate the tail withdrawal component of the tail-siphon withdrawal reflex (Walters et al. 1983a). Similarly, opaline motor neurons located in the right pleural ganglion (Tritt and Byrne 1980) may be activated directly by TSNs, although this has not been shown experimentally. A monosynaptic circuit may not be sufficient to mediate other components of the response to tail stimulation, however. Motor neurons mediating actions of organs in the mantle region are located in the abdominal ganglion (Carew and Kandel 1977; Frost et al. 1988; Kupfermann et al. 1974; Perlman 1979). Therefore the effects of tail stimulation on the siphon, gill, and ink gland require a polysynaptic pathway. Interneurons may also play a role in the tail withdrawal component of the reflex. For example, the activation of TMNs by brief mechanical stimulation of the tail outlasts the transient response of the TSNs, in some cases by > 1 min (Walters et al. 1983a). One mechanism to account for the duration of the motor neuron response is recruitment of interneurons that augment the excitation produced by sensory neurons (SNs) (Frost et al ; White et al. 1991). Interneurons that might contribute to either the tail or siphon components of the response to tail stimulation have not been identified. The pleural ganglion would be a likely location for neurons with this integrative function. Indeed, neurons in that ganglion have been studied for their effect on other cells within the pleural (Kehoe 1972, 1985a,b; Xu et al ), cerebral (Fredman and Jahan-Parwar 1979), pedal (Buonomano et al. 1992; Xu et al ), and abdominal ganglia (Walters et al. 1982). Consequently, neurons in the pleural ganglion were surveyed for their effects on TMNs. We identified an interneuron LP117 that forms a /93 $2.00 Copyright The American Physiological Society 1767

2 1768 L. J. CLEARY AND J. H. BYRNE parallel excitatory pathway from TSNs to TMNs mediating the tail withdrawal reflex, and whose function may be to enhance and lengthen the contraction elicited by mechanical stimulation of the tail. In addition, this neuron may contribute to the integration of tail withdrawal with behaviors mediated by neural circuits located in other ganglia. LPI 17 may be an element of, or have strong input to, a unitary neural circuit mediating defensive arousal in Aplvsia. w METHODS ApZ~7.~iu cal@nica ( g) were obtained from Marine Specimens Unlimited (Pacific Palisades, CA) and were maintained in artificial sea water ( ASW; Instant Ocean) at 15 C. Before dissection, animals were anesthetized by injection of a volume of isotonic MgCl, equal to approximately one-half of their body volume. Isolated ganglia were pinned in a small Plexiglas chamber and perfused with ASW at room temperature. In some preparations the left pleural-pedal ganglia were isolated. In others, the cerebral, left pleural-pedal, and abdominal ganglia were isolated together with connectives intact. The connective tissue sheath surrounding the ganglia was removed surgically. Neurons were impaled with glass microelectrodes (3-7 MQ) filled with 3 M potassium acetate. Neurons were identified by multiple criteria, including location, electrophysiological properties, and antidromic stimulation of peripheral nerves (Carew and Kandel 1977; Frost et al. 1988; Hawkins et al ; Koester and Kandel 1977; Kupfermann et al. 1974; Schwartz and Shkolnik 198 1; Walters et al. 1983a). In the initial experiments, recordings were made from isolated pleural-pedal ganglia. In a typical configuration, simultaneous recordings were made from a TSN, two TMNs, and two pleural interneurons. To manipulate the postsynaptic membrane potential, a second current-passing electrode was placed in a TMN. In later experiments, cerebral and/or abdominal ganglia were also included with their connectives intact. In those experiments, only one electrode was placed into the TMN, and only one pleural interneuron was recorded. Recordings were also made from identified neurons in the cerebral and abdominal ganglia. LPll7 was injected by pressure with 2% horseradish peroxidase (HRP) in distilled water (Bailey et al. 1979; Eisenstadt et al. 1973). The ganglia were then fixed in 2% glutaraldehyde, 2% formaldehyde in a 0.1-M cacodylate buffer (Cleary and Schwartz 1987). The location of HRP was then revealed with diaminobenzidine and H,O,! (Bailey et al. 1979). Ganglia were prepared as whole mounts by dehydrating through graded ethanols, clearing with methyl salicylate, and mounting on concavity slides with Permount (Fisher). RESULTS LPll7 produced slow excitatory postsynaptic potentials (EPSPs) in TMNs As a first step toward identifying interneurons important for the behavioral response to tail stimulation, an extensive survey was made of connections to TMNs by neurons located on the ventral surface of the pleural ganglion, excluding the well-characterized sensory cluster. Occasionally, displaced SNs appeared outside of the cluster, but these cells were easily recognized as such on the basis of their size, pigmentation, and electrophysiological properties. In this initial survey of 70 preparations, recordings were made from I43 neurons whose stimulation elicited postsynaptic potentials in TMNs. Of these. 78 were excitatorv. 48 were inhibitory, and 17 produced a mixed response. In several preparations, neurons that produced EPSPs in TMNs also elicited inhibitory postsynaptic potentials ( IPSPs) in TSNs. The EPSPs in TMNs elicited by pleural interneurons fell into three classes: fast EPSPs elicited by a single action potential, fast EPSPs elicited by a burst of action potentials but not by a single spike, and slow EPSPs elicited by a burst of action potentials. Slow EPSPs elicited by a single action potential were never observed. This paper will focus on the subset of interneurons that elicited slow EPSPs in follower TMNs. Recordings were made from 44 such neurons, and these were found to have several electrophysiological features in common. Resting membrane potentials were in the range of -50 to -60 mv, and spontaneous action potentials were never present. Spontaneous excitatory synaptic input was present at low frequency, but little or no spontaneous inhibitory input was seen. In addition, stimulation of the pleural-abdominal connective elicited an antidromic action potential, indicating the presence of an axon in the connective. An unusual feature of this action potential was its small size, suggesting that active propagation to the cell body was blocked in the axonal arbor. Although stimulation of peripheral pedal nerves P7-P9, in which the TSNs send their axons to innervate the tail and posterior part of the animal, produced excitatory input, it never elicited an antidromic action potential. Neurons were identified as LPI 17 if they produced a slow EPSP in TMNs and had these electrophysiological features. In general, a single LP117 was studied throughout the recording session. In nine preparations, however, recordings were made from two different LPI17 neurons in the same pleural ganglion. An individual LPI 17 could elicit an EPSP in a TMN that was composed of both fast and slow components (33 of 44; see Fig. 1). Ten of these 33 neurons, however, produced an EPSP in a second follower that was composed only of the slow component (see Fig. 2). Of the remaining 11 neurons, 8 produced EPSPs composed only of the slow component. Thus the fast component, although common, is not necessary to elicit the slow component. Slow EPSPs were characterized by a duration of s from the end of the burst in LPl17 (Fig. 1). The amplitude of the slow EPSP was roughly proportional to the frequency of the action-potential burst in the presynaptic neuron, with a relatively high threshold for recruitment of the slow component. In current-clamp preparations, slow EPSPs were accom- I 4mV LPl17-J 1 20 mv 1 2 set FIG. 1. Intracellular stimulation of LPl17 elicited a biphasic excitatory postsynaptic potential (EPSP) in a follower tail motor neuron (TMN). The fast component resulted in action potentials in the TMN (clipped by the pen recorder) that were followed by a slow depolarization lasting for -25 s.

3 MULTIFUNCTION INTERNEURON IN APLYSIA 1769 I 4mV FIG. 2. A: membrane input resistance in a TMN during a slow EPSP was estimated by passing constant hyperpolarizing current pulses through a bridge circuit. These pulses were 1 -s duration and delivered at intervals of 3 s. During the slow EPSP elicited by LPI 17 stimulation (indicated by the black bar), the amplitude of the voltage step produced by the current pulse was increased by -30%. B: manual depolarization of the same neuron did not produce a change in the amplitude of the voltage pulse, indicating that the increased resistance associated with the slow EPSP illustrated in A was not due solely to voltage-dependent properties of the postsynaptic membrane. panied by increased membrane input resistance (Fig. 2). Because the postsynaptic neurons were not voltage clamped, however, the possibility remained that the apparent increase in membrane input resistance associated with the slow EPSP was due to voltage-dependent properties of the TMN membrane. For example, of eight TMNs whose input resistance was recorded, three displayed anomalous rectification (e.g., Hille 1984). It is unlikely that anomalous rectification in the TMN contributed to the slow EPSP, however, because slow EPSPs occurred in TMNs lacking anomalous rectification (Fig. 2). Moreover, an inward current of similar time course could be elicited after voltage clamp of the postsynaptic neuron (unpublished observation). In four of four preparations, the amplitude of the slow EPSP was reduced by hyperpolarizing the postsynaptic neuron (Fig. 3). Functional role of slow potentials Slow synaptic potentials have been of interest because of their potential role in neuromodulation (Jones and Adams 1987; Kupfermann 1979; Weight et al. 1979). In the TMNs, they seem to serve at least two functions. First, the slow EPSP may prolong the burst of spikes in the motor neuron. For example, a slow EPSP of sufficient amplitude to trigger action potentials may lead to prolonged discharges (see Figs. 5 and 7). This suggests that the slow EPSP may convert the signaling mode from amplitude to both amplitude and duration (see White et al. 1993). Second, the slow EPSP may increase the effectiveness of parallel input pathways activated after induction of the slow EPSP. An example of such an effect is illustrated in Fig. 4. Subthreshold EPSPs were evoked in a TMN at 5-s intervals by brief electric shock to a peripheral nerve. Intracellular stimulation of the interneuron produced a brief burst of action potentials in the motor neuron, followed by z slow EPSP. After interneuron stimulation, however, the evoked EPSP that previously was subthreshold triggered an action potential in the motor neuron. This effect lasted for s. When examined at an expanded time base, the evoked EPSP was larger than that occurring before interneuron stimulation. The enhanced EPSP could be due to the increased input resistance, but it is also possible that presynaptic facilitation contributes to the enhanced effectiveness of evoked input. Similar enhancement of EPSPs was obtained in two additional preparations. LPl17 motor forms a neurons parallel pathway between tail sensory and The effects of excitatory pleural neurons on TMNs suggest that they are elements of the neural circuit mediating tail withdrawal. To be elements in a pathway parallel to the monosynaptic pathway, however, they must also be excited by SNs. A single spike in the TSN elicited an EPSP in both a follower TMN and LPl 17 ( Fig. 5A ). The same LPI 17 had powerful effects on the TMN, producing a prolonged burst of spikes in the motor neuron (Fig. 5 B). Moreover, there was a residual depolarization that lasted for s. Excitatory input to LP117 from SNs was observed in four additional preparations. Some SNs, but not all, were hyperpolarized by LP117 (data not shown). Interconnections of LPll7 with other neural circuits Because LPI 17 had effects on TMNs, it was likely that it sent an axon to the pedal ganglion. It did not have axons in pedal nerves P7-P9, however, as determined by electrical stimulation of those nerves. To explore this issue further, an interneuron was filled with HRP to examine the extent of its arborization (Fig. 6). LPI 17 was located on the ventral surface in the rostra1 region of the pleural ganglion, near the cluster of SNs. There were three main branches. As ex- mv 5 set FIG. 3. Slow component of the response to LPI 17 was diminished by hyperpolarization of the TMN. LP117 was stimulated with a 1 -s depolarizing current pulse (indicated by the black bar) once every 5-7 min. After the slow component of the response returned to baseline, the membrane potential of the TMN was systematically varied by passing current through a 2nd electrode in the TMN. Although the fast component of the EPSP was relatively unchanged, the amplitude of the slow component diminished, but did not reverse, as the membrane potential was changed from its resting level of -50 mv to mv. The slow component was restored to its original amplitude after returning to the resting, membrane potential. 4mV

4 L. J. CLEARY AND J. H. BYRNE B 1 2 LllO 3 10 set t L J -a w &llomv msec FIG. 4. LPI 17 stimulation modulated the effectiveness of synaptic input to TMNs. A: evoked synaptic potentials were produced in both a TMN and LPI 17 by electrical stimulation of peripheral nerve PS at 5-s intervals. A burst of spikes in the interneuron produced by a depolarizing current pulse elicited a brief burst of spikes in the motor neuron followed by a slow EPSP that lasted for -40 s. As a result, evoked EPSPs that were subthreshold for action-potential generation were converted to suprathreshold EPSPs for a period of at least 23 s. Evoked EPSPs numbered l-3 in fop trace are illustrated at a faster time base in B. B: evoked EPSP before interneuron stimulation was 4 mv in amplitude (trace I ). After stimulation the evoked EPSP reached 6 mv before triggering a spike (trace 2). One of the evoked EPSPs after stimulation failed to trigger a spike (trace 3). Amplitude of this EPSP was also 6 mv. petted, one process projected through the pleural-pedal connective to the TMN region of the pedal ganglion, consistent with its effects on those cells. It branched within the neuropil without exiting via pedal nerves. A second large process projected through the cerebral-pleural connective to the cerebral ganglion. A third small process projected down the pleural-abdominal connective. Numerous fine branches penetrated the neuropil of the pleural ganglion, including the region underlying the SN cluster. mv The projection of axons to neighboring ganglia suggested that may play a role in the integration of tail withdrawal with other behaviors. Indeed, stimulation of a single interneuron produced PSPs in motor neurons from four different ganglia ( Fig. 7). The interneuron was initially identified by its connection to a TMN in the pedal ganglion (Fig. 7, 1st trace). In addition, stimulation of the interneuron excited a cell in the B cluster of the cerebral ganglion (Fig. 7, 2nd trace). Cells receiving input from LPI 17 had narrow spikes, identifying them as Bn cells. These are motor neurons involved in withdrawal movements of the head (Fredman and Jahan-Parwar 1977; Teyke et al. 1989). Neurons in the A cluster, located just lateral to the B cluster, were inhibited (data not shown). In the abdominal ganglion, the ink motor neuron L14 (Carew and Kandel 1977) was simultaneously excited (Fig. 7, 3rd trace). The pleural giant neuron LPll also received input from the interneuron (Fig. 7, 4th trace). This cell is a visceromotor neuron that elicits the secretion of mucus from the skin (Rayport et al. 1983). The interneuron produced a mixed PSP in the pleural giant neuron. After an initial excitation that was sufficient to generate two action potentials, LPll was hyperpolarized by a prolonged IPSP. Although output from LP117 was extensive, not all identified neurons received input. For example, the giant cerebral neuron (C 1) was unaffected by LPI 17 stimulation (data not shown). The effects of pleural interneurons on L 14 suggested that these interneurons may also conduct information about tail stimulation to the motor neurons mediating gill and siphon contraction. An LPI17 that produced a slow EPSP in a TMN (Fig. 8, 1st trace) also excited motor neurons in the abdominal ganglion. Both the tail-sensitive siphon motor cluster LF, cells (Fig. 8,2nd trace) and the gill motor neuron LDG, (Fig. 8, 4th trace) were strongly excited by the pleural interneuron, as was the ink motor neuron L 14 (Fig. 8,3rd trace). Excitation of LDG, and L14 occurred in two bursts, an early burst coincident with stimulation of the interneuron and a delayed burst. The delayed burst in LDG, and the second component of the slow EPSP in L14 may have been due to recruitment of L25 and R25 pattern-generating neurons involved in respiratory pumping (Byrne 1983; Byrne and Koester 1978; Koester 1989; Koester and TMN,- *14mV TMN, - 10mV TSN i, 14OmV 250 msec 10 set FIG. 5. Simultaneous intracellular recordings were obtained from a TMN, LPI 17, and a tail sensory neuron (TSN). A: a single spike in the TSN elicited EPSPs in both the TMN and LPI 17. B: a high-frequency burst of spikes in the same LPI 17 illustrated in A elicited a long-lasting burst of spikes in the TMN. The burst was apparently prolonged by an underlying depolarization that lasted for -25 s. The action potentials in the motor neuron were clipped by the pen recorder. 40 mv

5 MULTIFUNCTION INTERNEURON IN APLYSIA 1771 F IG. 6. Whole-mount preparation of LPI17 labeled with horseradish peroxidase. The cell body of the interneuron is located in the pleural ganglion. One axonal branch ( 1) passes underneath the cluster of sensory neurons (SN) and projects through the connective to the pedal ganglion, consistent with the ability of the interneuron to elicit a short-latency EPSP in the motor neuron. In addition, 2 other branches exit the ganglion. One of these (2) projects through the connective to the cerebral ganglion and another (3) through the connective to the abdominal ganglion. The scale bar represents 200 pm. Kandel 1977). The source of the slow EPSP in LF, motor tion of LP117 produced biphasic responses in L 14 (Fig. 9A, neurons is probably not L25, however (Frost et al. 1988). 2nd truce) and LD,, (Fig. 9A, 3rd truce) as in Fig. 8, al- This potential may be produced by LP117 directly or from though the second component occurred more rapidly in the interposed interneurons (see below). latter case. This response was correlated with a burst of The conclusion that gill and ink motor neurons were indi- spikes in an L25 neuron (Fig. 9A, 4th trace). Because rectly excited by recruitment of the L25/R25 circuit was bursts in L25 neurons occur spontaneously, it is possible supported by simultaneous recordings (Fig. 9A). Stimula- that the burst after LPl17 stimulation was not evoked but TMN (Pedal) (CeZral) (*txkinai)~ II.,,,,I ul, (Pleural) Lp L-:: mm: --^--- :+ m FIG. 7. Divergent output of LPI17 to motor neurons in 4 different ganglia. A tail motor neuron (TMN) in the pedal ganglion and a Bn cell in the cerebral ganglion exhibited postsynaptic potentials with both fast and slow components. L14 in the abdominal ganglion exhibited only a slow component that outlasted the duration of the interneuron burst, but in this case returned to baseline faster than the slow EPSPs generated in the TMN and Bn neurons. The pleural giant neuron LPI1 exhibited a biphasic PSP in which a fast EPSP was followed by a slow hyperpolarization. Action potentials in the top 4 truces were clipped by the pen recorder. LPI17 H,40m 20 set

6 1772 L. J. CLEARY AND J. H. BYRNE Tail MN FIG. 8. Divergent output of LP117 to motor neurons in the abdominal ganglion. The EPSP in the tail motor neuron (top trace) exhibited a single slow component, but this was sufficient to identify LPI 17. The same burst of spikes elicited a slow EPSP in the siphon motor neuron (LF,). In this experiment the EPSP in L14 was biphasic; the early slow component was followed by a late slow component of longer duration. The response in the gill motor neuron LDG, was also biphasic. The interneuron stimulation elicited a short-latency burst of spikes in LD,, that was followed -2 s later by a 2nd high-frequency burst. simply occurred by coincidence. This is unlikely, however, because spontaneous bursts in L25 / R25 occurred at intervals of several minutes, but evoked bursts could be driven at much shorter intervals ( 1 O-20 s). In addition, stimulation of LPl17 elicited a small prepotential in L25 that quickly reached threshold (Fig. 9 B, top). This prepotential was not present in a spontaneous burst (Fig. 9B, bottom). Note that LPI 17 was silent during the spontaneous burst. An additional pathway for excitation of siphon motor neurons is through activation of L29 neurons in the abdominal ganglion (Fischer and Carew 1993; Frost et al. 1988, 199 1). These neurons produce a biphasic EPSP in LF, neurons much like that produced by LPI 17 in TMNs (Frost et al ). Stimulation of LPI 17 produced a fast EPSP in L29 that decremented rapidly (Fig. 10A). The strength of this connection, however, was not likely to be sufficient to account for the slow EPSP elicited in LF, neurons by LPI 17 (e.g., Fig. 8), because the pleural neuron was never able to elicit spikes in L29. Interestingly, the LPll7-L29 connection was reciprocal (Fig. 1OB). Stimulation of L29 elicited an EPSP composed of one-for-one synaptic potentials. Moreover, in some preparations, it was possible to elicit fast EPSPs in TSNs in the pleural ganglion and TMNs in the pedal ganglion by stimulation of L29. DISCUSSION Contribution of LPl17 to the tail withdrawal reflex In Apfysia, both the amplitude and duration of the withdrawal response appear to be shaped by the activity of motor neurons (Stopfer and Carew 1987). Although the relationship between motor neuron activity and behavioral response has not been studied systematically for tail withdrawal, it has been studied for contraction of the gill J20tnV 5 set and a buccal muscle (Byrne et al. 1978a,b; Cohen et al. 1978; Swann et al. 1982; Kupfermann et al. 1970). In general, there is, above some threshold value, a roughly linear relationship between spike frequency in the motor neuron and the behavioral response. This relationship reflects the amplitude of the excitatory junction potential elicited in the muscle fiber by the motor neuron. Other factors are important, however, because tonic low-frequency firing in siphon motor neurons, although not producing a contraction itself, can enhance the contraction elicited by a brief high-frequency burst of spikes (Frost et al. 1988). In addition, contractility of buccal muscle can be modulated in the absence of changes in synaptic strength (Weiss et al ). Despite its strength, the monosynaptic connection be- tween TSNs and TMNs cannot account fully for the activity pattern elicited by tactile stimulation in TMNs. Intracellular activation of a TSN generally elicits a burst of spikes in the TMN whose termination coincides with that of the SN burst (Walters et al. 1983a). Thus a TSN codes information primarily in terms of the amplitude of the summated EPSP in the motor neuron. This EPSP may generate spikes in the motor neuron. Tactile stimulation, however, elicits a burst of spikes in the motor neuron that greatly outlasts the burst elicited in the SN (Walters et al. 1983a). Additional circuit elements are presumably required to extend the motor neuron response in time. By eliciting a relatively longlasting ( s) EPSP in the TMN, LPI 17 appears to perform this function. This slow EPSP may prolong the burst produced by a fast component of the synaptic response, or it may increase the effectiveness of coincident synaptic input. Thus, although fast input from both SNs and interneurons may shape the amplitude of the response, LP117 may also shape the duration of the response (see White et al. 1993). Interneurons in the circuit for the analogous siphon-gill withdrawal reflex may also participate in shaping

7 MULTIFUNCTION INTERNEURON IN APL YSIA 1773 Evoked. Tar Ink MN (L14) k I SmV LPI17 y ~, 1 )20mV Spontaneous l25 1 I 5mV A a j20mv Hl20mV FIG. 9. Correlation between activity in L25 neurons and PSPs in abdominal motor neurons. A: in this experiment, LP117 produced a biphasic EPSP in the TMN (top trace) and produced short-latency PSPs in L14, LD,,, and L25. Activity in L25 was correlated with the late components of the EPSPs in both L14 and LD,,. In this preparation the 2nd component of the PSP in LDG, merged with the 1 st component (see Fig. 8). B: activation of L25 by the interneuron was synaptically driven (top trace), as evidenced by the prepotential that occurred during interneuron stimulation. This prepotential was absent in spontaneous bursts of the same L25 neuron (bottom trace). the duration of the response (Byrne 1983; Trudeau and opaline release (Rayport et al. 1983; Tritt and Byrne 1980). Castellucci 1992). A common feature of the responses elicited by tail stimulation is the transformation from a relatively quiescent state may participate in defensive arousal to a state of arousal in which defensive reflexes are elicited and locomotor behavior is enhanced (Walters et al ). Motor neurons controlling muscles involved in defensive Defensive arousal is also associated with inhibition of feedbehaviors are located throughout the nervous system. Many of these motor neurons can be activated by tail stimulation, including tail, siphon, and gill withdrawal, inking, and escape locomotion (Byrne et al ; Carew et al. 1983; Scholz and Byrne 1987; Walters et al. 1983a; Walters and Erickson 1986). Other defensive behaviors that are likely to be elicited by tail stimulation include head and tentacle withdrawal (Teyke et al. 1989) and mucus and ing, an appetitive behavior. The recruitment of individual components of the response to a given stimulus, however, depends on the intensity and location of the stimulus. For example, tail and siphon withdrawal can be elicited by a weak stimulus, whereas inking requires more intense stimulation (Carew and Kandel 1977; Scholz and Byrne 1987; Shapiro et al. 1979). A major function of the neural circuitry mediating the L2g k 15mV I 20 mv I I 20mV & UI 5mV I set 0.5 set FIG. 10. LPI 17 made reciprocal connections with L29. A: stimulation of LPI 17 produced a fast EPSP in L29 that decremented before the end of the burst. B: stimulation of L29 elicited EPSPs in LP117 that followed the spikes 1 for 1. I

8 1774 L. J. CLEARY AND J. H. BYRNE Mucous Glands Pleural I Cerebral Pedal Tail Siphon Gill Ink Gland FIG. 11. Schematic diagram illustrating the extent of the connectivity observed for LPl17. The interneuron makes excitatory connections (A) with motor neurons in 4 different ganglia. The connection with LPI I in the pleural ganglion is biphasic excitatory and inhibitojl (A). In addition, interneurons in the abdominal ganglion such as L29 and L25 are also excited by LPI 17. Although stimulation of a single LPI 17 is not sufficient to facilitate the synapse between sensory and motor neurons in the abdominal ganglion, there are potential connections to a facilitatory circuit (A) affecting sensory neurons in the abdominal ganglion. These include a potential connection to CB 1 in the cerebral ganglion (- - -) and a connection to L29, which has been shown to modulate the sensory-motor synapse in the abdominal ganglion. LPI 17 also inhibits some sensory neurons in the pleural ganglion (A). Triangles represent functional connections and do not necessarily signify monosynaptic connections. defensive arousal state is coordination of the individual motor acts that comprise the response to a particular stimulus by conducting information about tail stimulation to other ganglia, where the output motor neurons are located (Carew et al ). On the basis of their morphology alone, TSNs are unlikely candidates for this function, because TSNs do not project to the cerebral or abdominal ganglia (Cleary and Byrne 1984; Nazif et al )) where motor neurons controlling the tentacles or the mantle organs are located. This is in contrast to the siphon-gill withdrawal circuit in which SNs are located in the same ganglion as the motor neurons to mantle organs, and may, at least in principle, coordinate the response to mantle stimulation themselves. The interneuron identified in this study could be an element of the defensive arousal circuit. Stimulation of LP117 has widespread effects throughout the nervous system (Fig. 11). In addition to its effects on TMNs in the pedal ganglion, LP117 produces strong excitation of motor neurons in the cerebral ganglion mediating tentacle withdrawal, of motor neurons in the abdominal ganglion mediating siphon-gill withdrawal and inking, and of a motor neuron in the pleural ganglion mediating mucus release. In addition, it excites interneurons in the abdominal ganglion such as L25 and L29, which may contribute to siphon and gill withdrawal. Future studies may reveal additional connections. Therefore LPI 17 appears to be a command neuron in that it may elicit a stereotyped pattern of behaviors. Although most of these behaviors are not sustained rhythmic responses, it is of interest that LP117 does have strong input to the L25/R25 network that mediates respiratory pumping, a rhythmic contraction of mantle organs. Additional experiments will be necessary to examine the effects of LPI 17 on the locomotor and feeding pattern-generating systems. The effects of LPI 17 are similar in diversity but opposite in effect to those of CPR, an interneuron in the cerebral ganglion that appears to play a role in the circuit mediating food arousal (Teyke et al. 1990). This neuron excites elements of the circuit underlying the consummatory phase of feeding as well as elements of the cardiovascular and postural control systems but inhibits several defensive motor neurons. Although the effects of LP117 on its followers are diverse and powerful, it is not known if the actions of LPI 17 are monosynaptic or, like CPR, polysynaptic. In pilot experiments, application of bathing solutions containing high concentrations of divalent cations blocked the slow EPSP (data not shown), suggesting a polysynaptic connection. However, this traditional interpretation of the high divalent cation test may not be appropriate for testing the connections made by LP117 because of the long axonal distances and the possibility of conduction block (Waxman 197 1). Role of LPll7 in behavioral plasticity Plasticity in the tail-siphon and siphon-gill withdrawal reflexes has been analyzed both in terms of the amplitude of the contraction and its duration (Byrne et al ; Pinsker et al. 1970; Scholz and Byrne 1987; Stopfer and Carew 1987; Walters et al. 1983a). Enhancement of these reflexes as a result of sensitization has been correlated primarily with modulation of transmitter release from SNs (Castellucci et al. 1970; Walters et al. 1983b). Because SNs contribute primarily to the amplitude of the response, however, plasticity at other sites may be necessary to alter response duration. The hypothesis that LPI 17 contributes to the duration of withdrawal responses suggests that this neuron may be a site of plasticity during sensitization. This hypothesis is discussed in greater detail in the companion paper (White et al. 1993). Interneurons also appear to contribute to the early inhibition of the siphon withdrawal reflex that is produced by sensitizing stimuli (Wright et al ). LPI 17 may contribute to sensitization in other ways. For example, it is likely that sensitizing stimuli recruit a second class of modulatory interneurons that act on SNs. This circuit has not been identified for the tail-siphon withdrawal reflex. There was no evidence that stimulation of a single LP117 produced synaptic facilitation at monosynaptic synapses between TSNs and TMNs. However, the pleural interneurons could provide a link to interneurons in the abdominal ganglion that modulate the neural circuit mediat-

9 MULTIFUNCTION INTERNEURON IN APLYSIA 1775 ing the siphon-gill withdrawal reflex. The release of transmitter from siphon sensory neurons is enhanced by activation of three cells, L22, L28, and L29 (Hawkins et al ). At least one of these, L29, is excited by pleural interneurons. This is consistent with the observation that L29 itself is sensitive to tail stimulation (Frost et al. 1988). Moreover, the pleural interneuron to L29 connection is reciprocal. Thus a positive feedback loop exists that could prolong and enhance the effects of a sensitizing tail stimulus. Another neuron whose activity elicits facilitation in siphon sensory neurons is the serotonergic CB 1 located in the cerebral ganglion (Hawkins 1989; Mackey et al. 1989). Because neurons in the B cluster are strongly excited by LPl17, CBl cells are also likely to receive excitatory input. Preliminary evidence suggests that LPI 17 receives excitatory input from neurons in the sensory J cluster of the cerebral ganglion (J. L. Raymond and J. H. Byrne, personal communication). This interesting class of neurons elicits two correlates of sensitization, depolarization and antiaccomodation, in pleural SNs (Raymond and Byrne 199 1). Therefore LP117 may also contribute to the behavioral response to head stimulation. We thank D. Baxter, J. Raymond, and J. White for helpful discussions and comments on an earlier draft of the manuscript. T. Vicknair and J. Pastore provided assistance in the preparation of figures. This research was supported by National Institute of Mental Health award MH and National Institute of Neurological Disorders and Stroke Grant ROl NS to J. H. Byrne and F32 NS to L. J. Cleary. Address reprint requests to: L. J. Cleary. Received 10 March 1993; accepted in final form 28 June REFERENCES BAILEY, C. H., THOMPSON, E. B., CASTELLUCCI, V. F., AND KANDEL, E. R. Ultrastructure of the synapses of sensory neurons that mediate the gillwithdrawal reflex. J. Neuuocytol. 8: , BALDISSERA, F., HULTBORN, H., AND ILLERT, M. Integration in spinal neuronal systems. In: Handbook of Physiology. The Nervous System. Motor Control. Bethesda, MD: Am. Physiol. Sot., 1981, sect. 1, vol. II, part 1, p BUONOMANO, D. V., CLEARY, L. J., AND BYRNE, J. H. Inhibitory neuron produces heterosynaptic inhibition of the sensory-to-motor neuron synapse in Aplysia. Brain Res. 577: , BYRNE, J. H. Identification and initial characterization of a cluster of com- mand and pattern-generating neurons underlying respiratory pumping in Aplysia californica. J. Neurophysiol. 49: 49 l-508, BYRNE, J. H., BAXTER, D. A., BUONOMANO, D. V., CLEARY, L. J., ESKIN, A., GOLDSMITH, J. R., MCCLENDON, E., NAZIF, F., NOEL, F., AND SCHOLZ, K. P. Neural and molecular bases of nonassociative and associative learning in Aplysia. Ann. NY Acad. Sci. 627: , BYRNE, J. H., CASTELLUCCI, V. F., CAREW, T. J., AND KANDEL, E. R. Stimulus-response relations and stability of mechanoreceptor and motor neurons mediating defensive gill-withdrawal reflex in Aplysia. J. Neurophysiol. 4 1: , 1978a. BYRNE, J. H., CASTELLUCCI, V., AND KANDEL, E. R. Contribution of individual mechanoreceptor sensory neurons to defensive gill-withdrawal reflex in Aplysia. J. Neurophysiol. 4 1: , 1978b. BYRNE, J. H. AND KOESTER, J. Respiratory pumping: neuronal control of a centrally commanded behavior in Aplysia. Brain. Res. 143: , CAREW, T. J., HAWKINS, R. D., AND KANDEL, E. R. Differential classical conditioning of a defensive withdrawal reflex in Aplysia calzfornica. Science Wash. DC 2 19: , CAREW, T. J. AND KANDEL, E. R. Inking in Aplysia caltfornica. I. Neural circuit of an all-or-none behavioral response. J. Neurophysiol. 40: , CAREW, T. J., WALTERS, E. T., AND KANDEL, E. R. Associative learning in Aplysia: cellular correlates supporting a conditioned fear hypothesis. Science Wash. DC 2 11: 50 l-504, CASTELLUCCI, V. F., PINSKER, H., KUPFERMANN, I., AND KANDEL, E. R. Neuronal mechanisms of habituation and dishabituation of the gillwithdrawal reflex in Aplysia. Science Wash. DC 167: , CLEARY, L. J. AND BYRNE, J. H. Light and electron microscopic examination of sensory neurons and motor neurons mediating the tail withdrawal reflex in Aplysia. Sot. Neurosci. Abstr. 10: 9 16, CLEARY, L. J. AND SCHWARTZ, J. H. Movement of newly synthesized membrane by fast transport along the axon of an identified Aplysia neuron. J. Comp. Neurol. 263: , COHEN, J. L., WEISS, K. R., AND KUPFERMANN, I. Motor control ofbuccal muscles in Aplysia. J. Neurophysiol. 4 1: 157-l 80, DENNY, M. W. Invertebrate mucous secretion: functional alternatives to vertebrate paradigms. Symp. Sot. Exp. Biol. 43: , EISENSTADT, M., GOLDMAN, J. E., KANDEL, E. R., KOIKE, H., KOESTER, J., AND SCHWARTZ, J. H. Intrasomatic injection of radioactive precursors for studying transmitter synthesis in identified neurons of Aplysia caltfornica. Proc. Natl. Acad. Sci. USA 70: , FISCHER, T. M. AND CAREW, T. J. Activity-dependent potentiation of re- current inhibition: a mechanism for dynamic gain control in the siphon withdrawal reflex of Aplysia. J. Neurosci. 13: , FREDMAN, S. M. AND JAHAN-PARWAR, B. Identifiable cerebral motoneurons mediating an anterior tentacular withdrawal reflex in Aplysia. J. Neurophysiol. 40: , FREDMAN, S. M. AND JAHAN-PARWAR, B. Intra- and interganglionic synaptic connections in the CNS of Aplysia. Brain Res. Bull. 4: , FROST, W. N., CLARK, G. A., AND KANDEL, E. R. Parallel processing of short-term memory for sensitization in Aplysia. J. Neurobiol. 19: , FROST, W. N., WV, L. G., AND LIEB, J. Simulation of the Aplysia siphon withdrawal reflex circuit: slow components of interneuronal synapses contribute to the mediation of reflex duration. Sot. Neurosci. Abstr. 17: 1390, HAWKINS, R. D. Localization of potential serotonergic facilitator neurons in Aplysia by glyoxylic acid histofluorescence combined with retrograde fluorescent labeling. J. Neurosci. 9: , HAWKINS, R. D., CASTELLUCCI, V. F., AND KANDEL, E. R. Interneurons involved in mediation and modulation of gill withdrawal reflex in ApZysia. I. Identification and characterization. J. Neurophysiol. 45: , HENING, W. A., WALTERS, E. T., CAREW, T. J., AND KANDEL, E. R. Moto- neuronal control of locomotion in Aplysia. Brain Res. 179: 23 l-253, HILLE, B. Ionic Channels ofexcitable A4embranes. Sunderland, MA: Sinauer, JONES, S. W. AND ADAMS, P. R. The M-current and other potassium currents of vertebrate neurons. In: Neuromodulation, edited by L. K. Kaczmarek and I. B. Levitan. New York: Oxford Univ. Press, 1987, p KANDEL, E. R. Behavioral Biology of Aplysia, San Francisco, CA: Freeman, KEHOE, J. Synaptic block of a calcium-activated potassium conductance in Aplysia neurones. J. Physiol. Lond. 369: , 1985a. KEHOE, J. Synaptic block of a transmitter-induced potassium conductance in Aplysia neurones. J. Physiol. Land. 369: , 1985b. KEHOE, J. S. Three acetylcholine receptors in Aplysia neurones. J. Physiol. Land. 225: , KOESTER, J. Chemically and electrically coupled interneurons mediate respiratory pumping in Aplysia. J. Neurophysiol. 62: 1113-l 126, KOESTER, J. AND KANDEL, E. R. Further identification of neurons in the abdominal ganglion of Aplysia using behavioral criteria. Brain Res. 12 1: l-20, KUPFERMANN, I. Modulatory actions of neurotransmitters. Annu. Rev. Neurosci. 2: , KUPFERMANN, I., CAREW, T. J., AND KANDEL, E. R. Local, reflex, and central commands controlling gill and siphon movements in Aplysia. J. Neurophysiol. 37: , KUPFERMANN, I., CASTELLUCCI, V., PINSKER, H., AND KANDEL, E. Neuronal correlates of habituation and dishabituation of the gill-withdrawal reflex in Aplysia. Science Wash. DC 167: , MACKEY, S.. AND CAREW. T. J. Locomotion in Aplvsia: triggering bv sero-

10 1776 L. J. CLEARY AND J. H. BYRNE tonin and modulation by bag cell extract. J. Neurosci. 3: , MACKEY, S. L., KANDEL, E. R., AND HAWKINS, R. D. Identified serotonergic neurons LCB 1 and RCB 1 in the cerebral ganglia of ApZysia produce presynaptic facilitation of siphon sensory neurons. J. Neurosci. 9: , NAZIF, F., BYRNE, J. H., AND CLEARY, L. J. CAMP induces long-term morphological changes in sensory neurons of Aplysia. Brain Res. 539: , PERLMAN, A. J. Central and peripheral control of siphon-withdrawal reflex in Aplysia cah$ornica. J. Neurophysiol. 42: , PINSKER, H., KUPFERMANN, I., CASTELLUCCI, V. F., AND KANDEL, E. R. Habituation and dishabituation of the gill-withdrawal reflex in Aplysia. Science Wash. DC 167: , RAYMOND, J. L. AND BYRNE, J. H. Neurons in the J cluster of the cerebral ganglion produce modulatory effects on sensory neurons in the pleural ganglia of Aplysia. Sot. Neurosci. Abstr. 17: 548, RAYPORT, S., AMBRON, R. A., AND BABIARZ, J. Identified cholinergic neurons R2 and LPl, control mucus release in Aplysia: a model system for studying the development ofbehavior on the cellular level. J. Neurophysiol. 49: , SCHOLZ, K. P. AND BYRNE, J. H. Long-term sensitization in Aplysia: biophysical correlates in tail sensory neurons. Science Wash. DC 235: , SCHWARTZ, J. H. AND SHKOLNIK, L. J. The giant serotonergic neuron of Aplysia: a multi-targeted nerve cell. J. Neurosci. 1: , SHAPIRO, E., KOESTER, J., AND BYRNE, J. H. Aplysia ink release: central locus for selective sensitivity to long-duration stimuli. J. Neurophysiol. 42: , STOPFER, M. AND CAREW, T. J. Quantitative analysis of the relation between gill amplitude and siphon duration in the defensive withdrawal reflex of Aplysia. Behav. Neurosci. 101: , SWANN, J. W., SINBACK, C. N., KEBABIAN, P. R., AND CARPENTER, D. 0. Motoneurons which may utilize dopamine as their neurotransmitter. Cell. Mol. Neurobiol. 2: , TEYKE, T., WEISS, K. R., AND KUPFERMANN, I. A subpopulation of cerebral B cluster neurones of Aplysia cahjornica is involved in defensive head withdrawal but not appetitive head movements. J. Exp. Biol. 147: l-20, TEYKE, T., WEISS, K. R., AND KUPFERMANN, I. An identified neuron (CPR) evokes neuronal responses reflecting food arousal in Aplysia. Science Wash. DC 247: 85-87, TRITT, S. H., AND BYRNE, J. H. Motor controls of opaline secretion in ApIysia californica. J. Neurophysiol. 43: 58 l-594, TRUDEAU, L.-E., AND CASTELLUCCI, V. F. Contribution of polysynaptic pathways in the mediation and plasticity of Aplysia gill and siphon withdrawal reflex: evidence for differential modulation. J. Neurosci. 12: , WALTERS, E. T., BYRNE, J. H., CAREW, T. J., AND KANDEL, E. R. Mechanoafferent neurons innervating tail of Aplysia. I. Response properties and synaptic connections. J. Neurophysiol. 50: , 1983a. WALTERS, E. T., BYRNE, J. H., CAREW, T. J., AND KANDEL, E. R. Mechanoafferent neurons innervating tail of Aplysia. II. Modulation by sensitizing stimulation. J. Neurophysiol. 50: , 1983b. WALTERS, E. T., CAREW, T. J., HAWKINS, R. D., AND KANDEL, E. R. Classical conditioning in Aplysia: neuronal circuits involved in associative conditioning. In: Advances in Behavioral Biology, edited by C. D. Woody. New York: Plenum, 1982, vol. 26, p WALTERS, E. T., CAREW, T. J., AND KANDEL, E. R. Associative learning in Aplysia: evidence for conditioned fear in an invertebrate. Science Wash. DC21 1: , WALTERS, E. T., AND ERICKSON, M. T. Directional control and the functional organization of defensive responses in Aplysia. J. Comp. Physiol. A 159: , WAXMAN, S. G. Regional differentiation of the axon: a review with special reference to the concept of the multiplex neuron. Brain Res. 47: , WEIGHT, F. F., SCHULMAN, J. A., SMITH, P. A., AND BUSIS, N. A. Long- lasting synaptic potentials and the modulation of synaptic transmission. Federation Proc. 38: , WEISS, K. R., COHEN, J., AND KUPFERMANN, I. Potentiation of muscle contraction: a possible modulatory function of an identified serotonergic cell in Aplysia. Brain Res. 99: 38 l-386, WHITE, J. A., CLEARY, L. J., ZIV, I., AND BYRNE, J. H. A network model of the tail-withdrawal circuit in Aplysia. Sot. Neurosci. Abstr. 17: 1590, WHITE, J. A., CLEARY, L. J., ZIV, I., AND BYRNE, J. H. The role of interneurons in controlling the tail-withdrawal reflex in Aplysia: a network model. J. Neuroph-vsiol. 70: 1777-l 786, WRIGHT, W. G., MARCUS, E. A., AND CAREW, T. J. A cellular analysis of inhibition in the siphon withdrawal reflex of Aplysia. J. Neurosci. 11: , Xv, Y., CLEARY, L. J., AND BYRNE, J. H. Identification of pleural neurons that inhibit tail sensory neurons of Aplysia: correlation with FMRFamide immunoreactivity. Sot. Neurosci. Abstr. 17: 1590,