Action-Potential Broadening and Endogenously Sustained Bursting Are Substrates of Command

JOURNALOFNEUROPHYSIOLOGY Vol. 43, No. 3, March 1980. Printed in U.S.A.

Ability in a Feeding Neuron of Pleurobranchaea

RHANOR GILLETTE, MARTHA U. GILLETTE, AND WILLIAM J. DAVIS

The Thimann Laboratories, University of California, Santa Cruz, Califarnia 95064

SUMMARY AND CONCLUSIONS longed burst episodes, triggered by de- I. The ventral white cells (VWCs) of the polarizing current pulses, and progressive buccal ganglion of Pleurobranchaea, so spike broadening can be demonstrated in named for their position and color, are a the surgically isolated VWC soma. bilateral pair of neuron somata. Each sends 5. The paired VWCs are strongly elec- a single axon out its contralateral stomato- trically coupled and display highly synchronous activity. They receive synaptic inputs gastric nerve and has a dendritic field origi- from many previously identified interneurons nating close to the soma. 2. The VWCs exhibit spontaneous epi- of the feeding network and are thus recipro- cally coupled within the network. sodes of prolonged depolarization (duration 6. These results demonstrate that the l-4 min) accompanied by repetitive action- potential activity and separated by regular capacity of this neuron to generate broadened action potentials during repetitive activity intervals of 3 - 30 min. Such prolonged burst confers the ability to command coordinated episodes can be triggered by short pulses of motor-network output. The appropriate depolarizing current. During the repetitive repetitive activity can be produced endog- activity of the spontaneous bursts or that driven by imposed depolarization, the action enously in the form of prolonged bursts of potentials progressively broaden to 5- 16 spikes. times their initial duration. INTRODUCTION 3. During spontaneous bursting or activity driven by imposed depolarization, the cyclic Feeding behavior in the carnivorous motor output of the feeding network is marine gastropod Pleurobranchaea consists initiated or accelerated with a latency cor- of coordinated rhythmic biting and swallow- responding with the development of ap- ing movements of the feeding apparatus, preciable VWC spike broadening. When including the radular and lateral teeth and broadening of antidromic VWC spikes is the muscular esophagus (4). These move- suppressed by imposed hyperpolarization of ments are controlled by a neural network the soma, the frequency of feeding cycles whose elements and interconnections are is significantly lower than when broadened partially known (e.g., Refs. 5, 8, 10, 17, 18). spikes are allowed to develop. When trains Among the central elements of the feeding of spikes are driven by depolarizing current, network we have identified several popula- the motor output of the feeding network is tions of neurons, which, when stimulated, not initiated until the VWC spikes have elicit the central motor output pattern broadened to a repeatable “threshold” underlying feeding (9, 10). duration, regardless of the intensity of the This paper describes a newly identified depolarizing current. pair of cells, the ventral white cells (V WC), 4. The endogenous production of pro- capable of driving feeding output in the

0022-3077/80/0000-0000$01.25 Copyright 0 1980 The American Physiological Society 669 670 GILLETTE, GILLETTE, AND DAVIS

isolated buccal ganglion. In this paper we Types of preparation demonstrate that the command ability of the After dissection from the animal, ganglia VWCs depends on broadening of action po- were pinned to Sylgard in a cooled water-jacketed tentials consequent to repetitive firing. We preparation dish. Temperature was maintained demonstrate further that repetitive firing oc- at 12- 14OC.Ganglia were bathed in either filtered curs in isolated nervous systems in the form seawater, the filtered blood of the donor animal, of prolonged, recurrent bursts that are sus- or an artificial saline (mM) (420 NaCl, 25 MgCI,, 25 MgSO4, 10 KCl, 10 CaCl,, and 5 Tris-HCl tained endogenously within the VWC. buffer, pH 7.5). In some experiments the con- Forthcoming papers will examine the ionic nective tissue was softened to facilitate electrode basis of spike broadening and bursting and penetration by a 6- to 12-minute incubation in the possible roles of cyclic nucleotides and 0.2% pronase (grade B, Calbiochem) in seawater intracellular calcium in modulating these at room temperature. Three types of dissected phenomena. preparation were used: the isolated buccal ganglion, the isolated brain and buccal ganglia con- MATERIALS AND METHODS nected by the cerebrobuccal connectives, and the semi-intact preparation. The semi-intact prepa- Neuron morphology ration included the entire buccal mass with a 5- The morphology of the VWCs in the buccal cm length of esophagus and the attached buccal ganglion and brain. The buccal mass was pinned ganglion was studied using the intracellular cobalt-staining procedure ( 14). The neurons were loosely to Sylgard so as to allow the rostro- filled with 1 M CoCl, either intrasomatically by caudal movements of feeding to occur, while the buccal ganglion was stabilized on a small pressure injection (n = 1) or backfilled through the axon in the contralateral stomatogastric micromanipulated platform above the buccal mass for intracellular stimulation and recording nerve (n = 3). Cobalt filling and subsequent of VWC activity. The outer, muscular connec- development of the stain and histological pro- cedures were performed as described previously tive tissue sheath overlying the ventral side of the (8). Cleared whole mounts were examined and lobes of the buccal ganglia was routinely removed. photographed in the light microscope. Recon- Occasionally the inner transparent sheath was also removed. struction of neuron morphology was done by tracing photographs and adding finer details from Isolation of the VWC soma was performed direct microscopic observation. The axon path in the desheathed buccal ganglion. A cluster of was confirmed by electrophysiological mapping 3- 10 somata including the VWC was cut away (described below), and similar methods were from the neuropil using fine scissors. In three of used to study the branching of the axon over four cases, all adhering somata were gently the esophagus. teased off the VWC with fine tweezers; direct microscopic observation showed VWC axon Recording methods stubs of less than 50 pm in length. In the fourth Multiple extracellular recordings were made case adhering somata were not removed; how- using suction electrodes on the cut stumps of ever, postexperimental cobalt injection and nerve roots and connectives or en passant from development showed an axon stub of less than 50 intact nerves. Switching circuits allowed elec- pm and no evidence of dendritic processes. Thus, trical stimulation through the suction electrodes. isolation was virtually complete in this case. Single or dual intracellular recordings were made The isolated somata were stabilized against two with 3 M KCl-filled glass microelectrodes and insect pins for intracellular recording. displayed and photographed on a storage oscilloscope or recorded on a Brush-Gould eight- RESULTS channel chart recorder. The response capability of the chart-recorder pens (100 Hz for 1 cm and The results are presented in three sec- 40 Hz for full-scale 4-cm sine waves) caused slight tions. In the first, we report data on the clipping of the amplitude of the fastest action morphology of the VWCs, as determined by potentials, but comparison with photographic cobalt staining. The second section shows records showed that it was adequate for re- producing the waveforms of prolonged spikes, that the command ability of the VWC derives their undershoots, and synaptic potentials. from broadening of action potentials that Movements of the buccal mass in the semi- occurs during endogenously sustained VWC intact preparation (see below) were registered bursts. The third section summarizes what by a force transducer attached by a small hook to is known about the synaptic physiology of the muscle of the radular sac. the VWCs. BROADENED SPIKES AND COMMAND ABILITY 671

Morphology and peripheral innervation The detailed morphology of the VWC has been examined in three ganglia by back- The bilaterally paired ventral white cells filling its stomatogastric axon with cobalt are named for their somata locations on or chloride, and in one ganglion by intracellular just beneath the ventral surface of each cobalt injection into the soma (Fig. 1). The buccal ganglion (Fig. 1) and their whitish drawing of Fig. 1 was traced from photo- color. The whitish color contrasts with the graphs of a whole mount, and fine details orange pigmentation of surrounding somata were added during direct microscopic ob- and allows ready visual identification. servation. The single axon emerges from the Electrophysiological mapping (intra- large (200 pm) soma and narrows to a cellular soma recordings combined with diameter of lo-15 ,um in the neuropil. extracellular recordings and orthodromic The majority of the fine, varicose dendritic and antidromic stimulation) showed that processes branch off close to the soma; each VWC sends an axon out on the contra- most of them ramify from two large primary lateral stomatogastric nerve to ramify over dendrites near the soma. The VWC axon the surface of the contralateral esophagus. traverses the commissure to the contra- Spontaneous VWC action potentials or those lateral side of the buccal ganglion and ex- driven by intracellular stimulation are pands abruptly to a diameter of 25-30 pm. followed by longitudinal contracture of the Some additional fine processes sprout from contralateral side of the esophagus, sug- the expanded regions. The axon then con- gesting a motoneuronal role for the VWC tracts to about 10 pm before entering the in addition to the interneuron function contralateral stomatogastric nerve and documented below. traveling to the periphery. ,cbc

FIG. 1. Tracing of a photograph of a cobalt-stained ventral white cell in the buccal ganglion. Note dendritic fields near the soma, axon that extends contralaterally to exit via the stomatogastric nerve, and expanded diameter of contralateral axon segment. The position of the soma of the contralateral homologous VWC is indicated. cbc, cerebrobuccal connective; sgn, stomatogastric nerve; rl, r2, and r3, nerve roots innervating musculature of the buccal mass; comm, commissure between the paired lobes of the ganglion. Calibration bar is 500 pm. See text for further details. 672 GILLETTE, GILLETTE, AND DAVIS

withdr. mn --.4”w-

Lvwc

FIG. 2. Intracellular stimulation of the left ventral white cell (LVWC) drives the feeding rhythm, recorded intracellularly from a withdrawal motoneuron (withdr mn) and extracellularly from roots 2 and 3 of the buccal ganglion (Rr 2, Lr 3).

Command ability of VWC stimulation. All of the 178 VWCs examined When feeding, the hungry intact Pleuro- showed this capacity to elicit feeding output branchaea commonly shows cyclic biting during intracellular stimulation. and swallowing movements at rates of 0. l- In spontaneously active buccal ganglia, 0.3 Hz. Previous papers have demonstrated activity recorded intracellularly from the that the feeding rhythm of the animal can be VWCs usually shifted between two distinct elicited in an isolated nervous system (e.g., modes: brief bursts (measured in seconds) Refs. 5, 10). The rhythm consists of alter- that were phase-locked to the withdrawal nating bursts of activity in two antagonistic portion of the feeding cycle, and longer populations of feeding neurons of the brain bursts (measured in minutes) that recurred and buccal ganglion: ‘ ‘eversion” neurons, at regular long intervals. Figure 3 illustrates which fire before and during radular pro- the transition from the first to the second traction, and “withdrawal” neurons, which type of burst activity. Nearly all of the VWCs fire during the retraction phase of the cycle. examined showed the first type of bursting In the present work we found that a similar (Fig. 3); 136 (76%) of the VWCs examined rhythmic motor output pattern can be in- showed the second type of bursting. In duced by intracellular stimulation of a single these cases large and prolonged depolariza- VWC at 3-5 Hz (Fig. 2). The rhythm typi- tions (15- 20 mV, l-4 min) recurred spon- cally commences with a latency of lo-20 s taneously at intervals of 3-30 min. Within and terminates shortly after cessation of individual preparations these parameters

Rt I . ,” ‘I ‘I ‘” I Rt 3

40 mv I 10 set

FIG. 3. Spontaneous feeding activity recorded from an isolated buccal ganglion preparation illustrating the two types of burst activity typically seen in VWC. Upper two records are extracellular recordings from roots 1 and 3 of the buccal ganglion, while lower record is an intracellular recording from the ipsilateral VWC. Typical transition from short bursts phase-locked to feeding to longer burst accompanied by intensified feeding activity is illustrated. BROADENED SPIKES AND COMMAND ABILITY 673 were typically quite constant, though sub- during such episodes was invariably followed stantial variation was seen from one prepa- by a doubling or tripling in the frequency ration to the next. The episodes of pro- of the feeding rhythm, as well as an increase longed bursting usually appeared to be in the intensity of motor activity during each triggered off of the excitatory synaptic activity burst (Fig. 3, upper traces). These sustained occurring during the withdrawal phase of depolarizations and bursts could also be the spontaneous feeding rhythm (Fig. 3). triggered by l- to 10-s depolarizing current During these longer bursts the cyclic VWC pulses injected into the VWC late during the activity described above was overshadowed interburst interval (Fig. 4A) and also by while the neuron discharged a prolonged spontaneous or stimulated EPSPs or by train of action potentials (3 -5 Hz) (Fig. 3, rebound from imposed hyperpolarization (Fig. middle of record). The VWC discharge 4B). For the greater length of the observed

A 2

50 mv

B 10 S vwc

FIG. 4. Initiation and suppression of prolonged VWC bursts by current injection. A 1: a short injection of depolarizing current initiates spikes and is followed by a depolarizing afterpotential, which supports several further spikes. A 2: a longer current phase triggers a full prolonged burst. B: a burst is initiated on rebound from imposed hyperpolarization and is terminated by a short hyperpolarizing pulse. Evidence that the current pulse terminates the burst is the lack of a characteristic slow repolarization during which spikes regain amplitude, as seen in the latter portion of the burst shown in Fig. 3. 674 GILLETTE, GILLETTE, AND DAVIS normal interburst interval, the triggerability train. Thus spike durations increased by 5- showed refractoriness; that is, a pulse of to 16-fold. Associated with the increase in depolarizing current-induced spikes and a spike duration was a net decrease in the decaying depolarizing afterpotential with- amplitude of the undershoot, or afterhy- out triggering a full sustained burst. Bursts perpolarization, of the spike. The progres- triggered prematurely in the cycle exhibited sive decrement in the undershoot accom- shorter durations. Those VWCs that did not panying spike broadening contributed sig- show spontaneous episodes of prolonged nificantly to the average depolarization of bursting also did not show the triggerability, the neuron during repetitive discharge. For or at best fired only l-20 spikes post- instance, the undershoot decrement between stimulation. the first and last spike shown in Fig. 5 is During trains of action potentials stim- 12 mV. Broadened action potentials fre- ulated by injected current or occurring quently developed a depolarizing inflection spontaneously, the waveform of the VWC or “hump” on the repolarizing phase spike changed dramatically. Specifically, (Fig. 6, lower trace). The occurrence of both individual action potentials progressively single and multiple humps is the reason broadened during repetitive discharge (Fig. for the irregular appearance of the spikes 5). Spike durations typically increased from in the temporally compressed burst of Fig. 3 15 rns (measured from takeoff) at the begin- (lower trace). The hump in the somatic ning of such trains to 75-250 ms during the spikes was associated with a second spike

FIG. 5. Superimposed action potentials from a depolarization-induced train showing progressive spike broadening. Action potentials were driven at about 3 Hz by a separate current-passing electrode and samples, respectively, at 0, 10, 20, 30, 45, 60, and 90 s. Thus, the 1st and approximately 30th, 60th, 90th, 135th, Moth, and 270th spikes of the train are depicted. BROADENED SPIKES AND COMMAND ABILITY 675

30 mv L50 ms

FIG. 6. Intracellular recordings of VWC action potentials before a repetitive train (upper trace) and in the latter portion of a prolonged spontaneous burst like that shown in Fig. 3. Note the distinct hump on the repolarizing phase, which occasionally develops with action-potential expansion. recorded extracellularly in the peripheral gastric nerve, the frequency of discharge axon (not shown). Peripheral axon spikes could be held constant while the somatic did not show appreciable broadening. Pre- spike duration was controlled by injected sumably, the hump induces the extra axon current. By simultaneously monitoring the spike since broadened action potentials in feeding rhythm, the effect of spike broad- the axonless isolated soma may also exhibit ening on the frequency of feeding rhythm the hump (see further). The hump acts to could be directly ascertained. prolong the somatic action potential further. Figure 7 illustrates the results of one such While acquiring such records, we noticed experiment. During a 2-min period of stim- that the initiation or acceleration of the ulation at 3 Hz in which spike broadening feeding rhythm coincided with the develop- was permitted by imposed depolarization, ment of appreciable spike broadening, 21 cycles of feeding activity occurred. together with net membrane depolarization When spike broadening was blocked by associated with decay of the spike under- hyperpolarization, in an adjacent trial 5 min shoot. This correlation suggested that spike later, only 12 cycles occurred in the same broadening and/or the accompanying slow time interval. For the experiment shown in depolarization increased the functional ef- Fig. 7, the average difference in the num- ficacy of the VWCs. We tested this hypothe- ber of feeding cycles for trials of broadened sis in two ways: first by assessing the ef- spikes versus those of unbroadened spikes fects of blockade of spike broadening and over 18 alternating trials was 4.22. The dif- slow depolarization on feeding output, and ference is significant at the 0.001 level (one- second by ascertaining the existence of a tailed t test for matched pairs of adjacent threshold value of spike duration at which trials). Table 1 summarizes the similar the feeding network was activated by results of three such experiments on three the VWC. different preparations. In the first type of test we took advantage of Assessment of the efficacy of the VWC the fact that spike broadening is voltage de- in this experimentis complicated by the fact pendent. That is, it is enhanced by tonic that stimulation of the stomatogastric nerve depolarization of the soma and it is sup- causes feeding output by itself, even when pressed by tonically hyperpolarizing the cell. the stimulus intensity is below threshold for By stimulating the VWC antidromically VWC activation. Assuming a nonlinear with a suction electrode on the stomato- response curve (see Fig. 2 in Ref. 5), 676 GILLETTE, GILLETTE, AND DAVIS

0.25s L50 mv R3 20s

vwc

0.2%

FIG. 7. Records from two adjacent trials comparing the efficacy of broadened VWC action potentials (upper records) versus unbroadened action potentials (lower records) in driving the cyclic motor output of feeding. VWC spikes were antidromically driven at 3 Hz for 2 min by stimulation of the stomatogastric nerve. Spike broadening was enhanced by depolarizing current injection in the VWC soma (upper VWC record) and suppressed by hyperpolarizing current injection (lower VWC record), while cycles of motoneuron activity were recorded extracellularly in the nerve R3. The chart record speed was increased transiently by 10x to show the relative waveforms of the action potentials during the latter portion of the trials. In these trials broadened spikes elicited 21 cycles of motor activity during the stimulus train, whereas unbroadened spikes were accompanied by only 12 cycles. See text.

activation of the feeding rhythm by stomato- network was recorded extracellularly from gastric nerve stimulation may partially mask a nerve (root 3) of the ganglion. VWC spikes the additional excitatory contribution of were allowed to broaden until a burst of broadened VWC spikes. motorneuron action potentials was initiated The second type of experiment testing the in the nerve. In this manner we could ob- efficacy of spike broadening was performed tain an approximate threshold value of the on three preparations in which spontaneous spike duration necessary to initiate feeding feeding activity was low enough to permit activity that was repeatable and similar for reliable measurements (about 1 cycle/3- 10 all strengths of depolarizing currents. The min). For two of these preparations the results are summarized in Table 2. In each spontaneous feeding output was reduced to experiment, over a wide range of depolarizing an acceptable level by raising the Mg2+ con- current intensities, the VWC spike dura- centration of the bath saline from 50 to 70 tion at which motoneuron activity was mM. A VWC was intracellularly stimulated initiated in the nerve invariably measured at various levels of depolarizing current within a few milliseconds of the overall while the extent of spike broadening was mean (2 1.6 ms). The rate of spike broadening recorded from a storage oscilloscope screen during a train is largely dependent on the to the nearest millisecond. Final spike dura- level of depolarizing stimulus current (un- tion was measured at the half-amplitude of published observations). Thus, average the initial spike of the train stimulated at spike frequencies measured between 2.5 the lowest current. Initial spikes were 8- 10 and 2.9 Hz at the lower stimulus currents and ms in duration. The activity of the feeding between 4.1 and 6.0 Hz at the higher cur- BROADENED SPIKES AND COMMAND ABILITY 677

TABLE 1. Efficacy of broadened vs. unbroadened spikes in driving feeding output

Avg No. of Feeding No. of Trials Cycles Stimulated Broadened Unbroadened Broadened Unbroadened Expt spikes spikes spikes spikes Significance Level, P

I 8 8 19.13 14.63 co.005 2 6 6 21.33 16.83 co.025 3 9 9 19.75 15.38

rents, while the spike train durations ranged current strengths, this contribution is likely between 30 and 150 s and 9 and 22 s for to be small compared to that of the broad- lowest and highest currents, respectively. ened spike itself. Figure 8 shows records from these experi- In order to separate endogenous expres- ments. The records of Fig. 8A and B com- sion from presynaptic influences, 19 attempts pare feeding burst initiation at stimulus- were made to dissect the VWC soma from current intensities of 2.4 and 5.3 nA, respec- the ganglion and study it in isolation. Of tively. In the first test shown in Fig. 8A, these, only four cell bodies retained high the stimulus current was stopped when resting potentials (60-65 mV) and the spike duration reached 19 ms, a duration capacity to generate action potentials. Over- that did not elicit activity in the feeding shooting action potentials occurred in the nerve (lower trace). However, when the isolated somata only when the membrane spike duration was allowed to expand to was depolarized by 25-30 mV. As in the 20 ms (second spike train, middle of the intact neuron, action potentials recorded record) motoneuron activity was initiated. from the isolated VWC somata showed Similarly in Fig. 8B, feeding activity oc- marked broadening on repetition and also curred when spike duration was recorded at developed a hump on the repolarizing phase 23 ms in the three trials shown. These latter (Fig. 9). Unlike the intact VWC, action values are likely to be high due to the high potentials stimulated in the isolated somata spike frequency and rapid broadening. emerged gradually out of regular oscillations The experiments described above show of the membrane potential (Fig. 9, record l), both that broadened spikes are more ef- a result that possibly reflects differences in fective than unbroadened spikes in driving the somatic and axonal membrane. the feeding network and that in a given All four isolated somata demonstrated preparation, a certain minimum VWC spike sustained depolarization and repetitive duration may be necessary for its effective- spiking that could be triggered by a depolar- ness in driving the network to be attained, izing stimulus. Figure 10 shows a typical regardless of the depolarizing stimulus. prolonged burst in an isolated soma triggered While these experiments do not directly by a 40-s depolarizing stimulus. The burst assess the possible contribution of the de- outlasted the triggering stimulus by 2 min polarizations from which spikes are gener- and 32 s. The record of Fig. 10 may be ated due to the uncontrolled spike fre- compared with that from an intact VWC quencies, in view of the rather exact thresh- in Fig. 4A 2, which is similar except for the olds of effective spike duration at all gradual emergence of overshooting spikes in 678 GILLETTE, GILLETTE, AND DAVIS

TABLE 2. V WC broadened spike duration unsuccessful search include the powerful initiating feeding burst interneurons, the anterior-ventral (AV) cells (10) (30 penetrations in 12 preparations); Stimulation VWC Spike Duration at another identified bilateral pair of com- Expt Current, nA Feeding Burst, ms mandlike neurons in the buccal ganglion I 2.4 22.29 + 1.35 (7) termed the B3 interneurons (1 penetration in 2.7 19.50 + 0.71 (2) 1 preparation); withdrawal motoneurons 3.0 22.00 2 1.73 (3) (10 penetrations in 5 preparations); eversion 3.4 24.10 + 1.95 (5) motoneurons (6 penetrations in 5 prepara- 3.6 24.50 + 2.12 (2) 23.20 + 1.79 (5) tions); putative sensory neurons (17) (10 5.3 penetrations in 3 preparations); corollary 2 4.0 21.16 + 1.47 (6) discharge neurons (2 penetrations in 2 prep- 4.75 20.67 + 0.82 (6) arations); and the efference copy neurons 5.25 20.00 + 0.63 (6) 6.0 20.00 t 0.63 (6) (5) (2 penetrations in 2 preparations). It would appear that either a major class of 3 2.0 22.33 + 0.82 (6) 2.5 21.50 2 1.22 (6) feeding neurons remains undiscovered or 3.0 20.83 + 0.75 (6) the VWC exerts its effect in some way other 3.5 20.67 + 0.52 (6) than conventional postsynaptic potentials. 4.0 21.33 + 0.52 (6) In contrast to the elusive postsynaptic effects of the VWC, one electrically coupled Values are means ? SD. Figures in parentheses and various chemically presynaptic neurons are n. Results of three experiments on different preparations ascertaining the VWC spike duration at which were readily found. Each VWC is strongly feeding network activity was initiated. Trains of coupled to its contralateral homolog by progressively broadening VWC spikes were driven at presumed electrical synapses (Fig. 11). different strengths of depolarizing current while motoneuron activity was recorded in a feeding nerve. Oscillatory potentials imposed on the soma The first experiment was performed with a ganglion of one member of the pair by current injec- that showed a relatively low frequency of spontaneous tion are recorded in the contralateral partner motoneuron bursts (about 0.3/min). In the second as attenuated oscillations; action potentials and third experiments, spontaneous activity was stimulated in one caused biphasic inflections reduced by raising bath Mg*+ by 20 mM. This procedure also seemed to reduce the variance in the of 4- to 5-mV amplitude in the partner with VWC spike duration necessary to induce motoneuron negligible latency (Fig. 11). As would be activity. See text for additional detail. The repeat- expected from such coupling, the paired ability of the value of the effective VWC spike VWCs show a high degree of synchrony in duration at all strengths of depolarizing current stimuli confirms the effectiveness of broadened action spontaneous fluctuations of their membrane potentials and suggests that the contributory effects potentials (Fig. 12A), in the times of onset of the depolarizations from which the action potentials and termination of spontaneous prolonged arise are relatively small. bursting episodes (Fig. 12B), and in the occurrence of spikes during the burst the isolated soma and the duration of the (Fig. 12C). triggering stimulus. One of the isolated Numerous synaptic inputs to the VWC somata showed the capacity for sustaining have been found from other identified mem- poststimulus activity only weakly, at best bers of the feeding network. In fact, practi- firing spikes of decreasing amplitude at 0.75 cally every known feeding interneuron Hz for 20 s poststimulation. As in intact tested exhibited a synaptic effect on the VWCs, the triggerability of prolonged bursts VWC. Of 22 anterior-ventral neurons tested in the isolated somata showed refractoriness in 12 preparations, 16 caused EPSPs (0.5- for a period (20-30 min) following a burst. 10 mV) and 6 caused IPSPs (0.5-5 mV) in the VWC. The paracerebral neurons pro- Synaptic physiology of VWC vide an example w%hoserole in behavior has Despite an intensive exploration of the been characterized. These cells, whose identified classes of neurons within the somata are located in the brain and send feeding network, we have been unable to descending axons to the buccal ganglion, find definite postsynaptic targets of the were previously shown to serve as command VWC. Classes of neurons sampled in this neurons for the feeding behavior (10). The BROADENED SPIKES AND COMMAND ABILITY 679

I .l mvL 30 set

FIG. 8. Current-stimulated trains of progressively broadening action potentials in a VWC initiate motoneuron bursts in a nerve (r3) only on reaching action-potential durations of about 20 ms. A: depolarizing stimulus currents 2.4 nA. The first stimulus train is terminated when spike duration attains 19 ms. Motoneuron activity is only initiated when, in the second train, spike durations attain 20 ms. B: depolarizing stimulus currents 5.3 nA; l-2 cycles of motoneuron bursts are initiated at spike durations of 23 ms in the three trials shown. While the appearance of the motoneuron bursts changed somewhat during this prolonged experiment, the VWC spike-duration threshold did not (see Table 2, expt I). paracerebral neurons normally fire bursts of B3 interneurons fire normally during the action potentials during radular eversion, withdrawal phase of feeding; therefore, the and they supply long-latency (400 ms), excitation they provide the VWCs may also relatively slow (Fig. 13A) and summating contribute to the cyclic discharge pattern (Fig. 13B) IPSPs to both VWCs. Since the of the VWCs during feeding output. VWCs fire during the withdrawal phase of feeding, the phasic inhibition from the para- DISCUSSION cerebrals during eversion is in accord with the typical cyclic discharge pattern of the Morphology of ventral white cell vwcs. The geometry of the VWC, as visualized An example of excitatory inputs to the by intracellular cobalt staining, exhibits two VWCs is from another pair of neurons of structural features that are potentially the buccal ganglion with the ability to drive relevant to its function. First, the threefold feeding (unpublished observations), the B3 expansion of axon diameter as it courses interneurons (so named for their ventral through the contralateral buccal ganglion position in rectangular coordinates (17)). in the region of the major dendritic field These inputs consist of discrete and facili- of its bilateral homolog (Fig. 1) furnishes tating, short-latency EPSPs (Fig. 13C). The a possible anatomical site for the presumed 680 GILLETTE, GILLETTE, AND DAVIS

30 mv L 100 ms 125

FIG. 9. Broadening of action potentials in a train stimulated by intracellular injection of depolarizing current in the surgically isolated VWC soma. Record 1 shows the gradual development of full-sized action potentials from regular oscillations in the potential of the depolarized membrane. Records 2-4 illustrate the progressive broadening of the action-potential duration during repetitive activity. Record 4 shows the development of a depolarizing hump on the repolarizing phase. Small numbers refer to the serial-order number of the action potentials in the train. electrical coupling between the two cells. this case the expanded axonal segment may Gap junctions, the morphological sub- act as a significant current source to the strates of electrical coupling, have been contralateral homolog, contributing to the localized at axodendritic sites in other gas- observed functional synchrony. Synchronous tropod central neurons (7, 11) and may form discharge of the two ventral white cells may the basis of the functional coupling between in turn contribute to bilateral coordination the VWCs demonstrated here (Fig. 11). In of the esophagus during feeding behavior.

soma

I J 20 mvl- IO set

FIG. 10. A sustained-burst episode triggered in the isolated VWC cell body by depolarizing current. BROADENED SPIKES AND COMMAND ABILITY 681 L vwccc

I R VWC

40 mv 5 na

FIG. 1 Electrical coupling of the V WCs. Regular oscillation s of the soma potential of the right VWC induced bY injected current (I) cau se attenuated oscillations in the partner n the I eft ganglion lobe.

A second morphological feature with pos- molluscan neurons (6, 20), but a functional sible functional implications is the close role has not been previously demonstrated. physical proximity to the VWC soma to its In the present work we have shown that dendritic field. Cobalt injection has shown spike broadening in the VWC is invariably that the major dendritic fields of the VWC correlated with acceleration of the feeding branch off near the soma and extend to a few rhythm (Fig. 3). Direct manipulation of the hundred micrometers of the soma. Com- degree of spike broadening by controlling bined physiological and morphological the membrane voltage (Fig. 7) suggests studies on other gastropod neurons suggest further that spike broadening is directly space constants of several millimeters (12). causal to augmented feeding output. The Therefore, the dendrites of the VWC may finding that the initiation of activity in the be located within a space constant of the feeding network is associated with a rather soma. By analogy with earlier studies on precise threshold duration of the VWC molluscan neurons, the dendrites in turn spike regardless of the level of the depolar- presumably represent the major synaptic izing stimulus current (Table 2, Fig. 8), input-output sites of the VWC (7,11,12,19). both confirms the effectiveness of the broad- Therefore, the broadened action potentials ened spike and provides evidence that the occurring in the VWC soma would be ex- contribution of the tonic membrane depolar- pected to exert a controlling influence on the ization driving repetitive spiking is itself synaptic output of the VWCs. negligible. Functional role of b broadened Mechanism of effects of broadened action pote n tials action potentials Progressive spike broadening with repeti- Despite an intensive search, we have not tion has been demonstrated previously in been able to identify sites of direct post- 682 GILLETTE, GILLETTE, AND DAVIS

R VWC’

5s 25 s 250ms

FIG. 12. Electrical coupling and synchrony of activity in the two VWCs. A: synchrony of spontaneous fluctuations of membrane potentials. B : synchrony of onset and termination of spontaneous burst episodes in the neurons. C: synchrony of action potentials occurring during spontaneous bursts in neurons.

synaptic action of the VWCs. Such negative excite the feeding network via potassium results suggest that the VWCs exert their depolarization; that is, the broadened action effects by indirect or unconventional syn- potentials may cause significant local release aptic mechanisms, such as heterosynaptic of potassium within the neuropil, resulting facilitation (13) or the induction of special in depolarization and consequent activation properties of excitability in other network of many membersof the feeding network. members (2, 15). In this regard it is notable that white neurons in the related mollusk Recurrent prolonged bursting in VWC Aplysia have been implicated in the release The broadened action potentials that im- of neuromodulatory peptides (2). Another part the ability of the VWC to drive the possibility, as yet untested, is that the VWCs feeding network normally occur spon- BROADENED SPIKES AND COMMAND ABILITY 683 A vwcc

PCNl -1 1 B vwc (A+B) 5 mv 125 mv

PCN

12.5 mv 125 mv

FIG. 13. Synaptic inputs to the VWC from major feeding interneurons of the network. A: single current- driven action potentials in a paracerebral cell (PCN), which has its cell body in the brain, cause slow IPSPs in the VWC. B: summation of IPSPs in the VWC driven by the PCN. C: one of the bilaterally paired B3 interneurons in the buccal ganglion mediates facilitating EPSPs to the VWC. Signs of the synaptic potentials are consistent with the phases of spike activity of neurons during the feeding cycle (see text). taneously during characteristic prolonged tive spiking can be triggered by injected and recurrent membrane depolarizations. current in both the intact and the isolated The prolonged depolarizations and repeti- VWC cell body, suggesting that they are 684 GILLETTE, GILLETTE, AND DAVIS endogenous to the neuron. These potentials would act to initiate feeding or to intensify are superficially similar to those that under- ongoing feeding activity. Evidence that the lie endogenous bursting in other molluscan VWCs normally do exert a command func- and arthropod neurons (1, 15, 16), with the tion in behavior has so far proved difficult distinction that the burst durations and to obtain as the VWCs are located on the interburst intervals are exceptionally pro- ventral side of the buccal ganglion, closely longed. In the intact cell spontaneous attached to the buccal mass that undergoes synaptic inputs are apparently adequate large movements during feeding. stimuli for triggering a burst episode in the In substance, the available data show nonrefractory VWC. Whether the prolonged that the VWC is capable of commanding depolarizations may actually occur as spon- the motor output underlying feeding. The taneous “pacemaker” potentials (1) or if evidence reported here shows that this com- they always require an external triggering mand ability arises from progressive broad- stimulus is not yet certain due to the limited ening of spikes during repetitive activity, number of observations on the isolated that in turn results from an endogenous cell body. capacity for producing prolonged bursts of spikes. Role of ventral white cell in feeding Episodes of spontaneous feeding behavior ACKNOWLEDGMENTS in intact animals in the absence of ap- This study was supported by a National Institutes propriate stimuli have been observed but of Health postdoctoral fellowship and a research grant rarely. The recurrent prolonged burst from the Graduate Research Board of the University episodes observed in the majority of the of Illinois to R. Gillette, a research grant from the isolated nervous systems may reflect some American Philosophical Society to M. U. Gillette, and National Institutes of Health Research Grants NS facet of arousal or disinhibition of the ventral 09050 and MH 23254 to W. J. Davis. white cell resulting from dissection and/or Present address of R. Gillette and M. U. Gillette: isolation of the CNS. Our results suggest Dept. of Physiology and Biophysics, 524 Burrill Hall, that during feeding in the intact animal, University of Illinois, Urbana IL 61801. burst episodes triggered in the VWCs by Received 6 November 1978; accepted in final form synaptic inputs from the motor network 12 September 1979.

REFERENCES

1. ALVING, B. Difference between pacemaker and metacerebral giant neuron in the feeding behavior nonpacemaker neurons of neurons of Aplysia on of Pleurobranchaea. J. Comp. Physiol. 116: 129- voltage-clamping. J. Gen. Physiol. 54: 512-531, 159, 1977. 1969. 9. GILLETTE, R., GILLETTE, M. U., AND DAVIS, 2. BRANTON, W. D., MAYERI, E., BROWNELL, P., W. J. Prolonged endogenous bursting and spike AND SIMON, S. B. Evidence for local hormonal broadening are substrates of command function communication between neurones in Aplysia. Nature in the feeding network of Pleurobranchaea London 274: 70-72, 1978. californica. Sot. Neurosci. Abstr. 4: 1210, 1978. 3. DAVIS, W. J. AND GILLETTE, R. Neural cor- 10. GILLETTE, R., KOVAC, M. P., AND DAVIS, W. J. relate of behavioral plasticity in command neurons Command neurons in Pleurobranchaea receive of Pleurohranchaea. Science 199: 801-804, 1978. synaptic feedback from the motor network they 4. DAVIS, W. J. AND MPITSOS, G. J. Behavioral excite. Science 199: 798-801, 1978. choice and habituation in the marine mollusc 11. GILLETTE, R. AND POMERANZ, B. Ultrastructural Pleurobranchaea californicu. 2. Vergl. Physiol. correlates of interneuronal function in the ab- 75: 207-232, 1971. dominal ganglion of Aplysia californica. J. Neuro- 5. DAVIS, W. J., MPITSOS, G. J., SIEGLER, M. V. S., biol. 6: 463-474, 1975. PINNEO, J. M., AND DAVIS, K. B. Neuronal 12. GRAUBARD, K. Voltage attenuation with Aplysia substrates of behavioral hierarchies and associ- neurons: the effect of branching pattern. Bruin ative learning in Pleurobranchaea. Am. Zool. 14: Res. 88: 325-332, 1975. 1037- 1050, 1974. 13. KANDEL, E. AND TAUC, L. Mechanism of 6. ECKERT, R. AND Lux, H. D. Calcium-dependent heterosynaptic facilitation in the giant cell of the depression of a late outward current in snail abdominal ganglion ofAplysia depilans. J. Physiol. neurons. Science 197: 472-475, 1977. London 181: 28-47, 1965. 7. GILLETTE, R. Microstructural and Ultrastructural 14. PITMAN, R. N., TWEEDLE, C. W., AND COHEN, Studies on Identified Neurons of the Abdominal M. J. Branching of central neurons: intracellular Ganglion of Aplysia californica (Thesis). Toronto: cobalt injection for light and electron microscopy. University of Toronto, 1974. Science 176: 384-385, 1972. 8. GILLETTE, R. AND DAVIS, W. J. The role of the 15. RUSSEL, D. F. AND HARTLINE, D. K. Bursting BROADENED SPIKES AND COMMAND ABILITY 685

neural networks: a reexamination. Science 200: rhythmic feeding output in the buccal ganglion of 453-456, 1978. Pleurobranchaea. J. Neurophysiol. 37: 1173- 16. SELVERSTON, A. I. Structural and functional 1196, 1974. basis of motor pattern generation in the stomato- 19. THOMPSON, E. B., SCHWARTZ, J. H., AND gastric ganglion of the lobster. Am. Zoo/. 14: KANDEL, E. R. A radioautographic analysis in 957-972, 1974. the light and electron microscope of identified 17. SIEGLER, M. V. S. Motor neurone coordination Aplysia neurons and their processes after intra- and sensory modulation in the feeding system of somatic injection of L-(3H) fucose. Brain Res. the mollusc Pleurobranchaea californica. J. Exp. 112: 251-281, 1976. Biol. 71: 27-48, 1977. 20. THOMPSON, S. H. AND GETTING, P. A. Calcium 18. SIEGLER, M. V. S., MPITSOS, G. J., AND DAVIS, currents and K-inactivation modulate molluscan W. J. Motor organization and generation of soma spikes. Sot. Neurosci. Abstr. 3: 189, 1977.