Tonic (d.c.) intracellular depolarization of the previously identified phasic paracerebral feeding command interneurones (PCps) in the brain of the carnivorous gastropod Pleurobranchaea causes oscillatory neural activity in the brain, both before and after transecting the cerebrobuccal connectives. Therefore, cycle-by-cycle ascending input from the buccal ganglion is not essential to cyclic brain activity. Instead the brain contains an independent neural oscillator(s), in addition to the oscillator(s) demonstrated previously in the buccal ganglion (Davis et al. 1973).
Transection of the cerebrobuccal connectives immediately reduces the previously demonstrated (Kovac, Davis, Matera & Croll, 1983) long-latency polysynaptic excitation of the PCps by the polysynaptic excitors (PSEs) of the PCps. Therefore polysynaptic excitation of the PCps by the PSEs is mediated by an ascending neurone(s) from the buccal ganglion.
The capacity of feeding command interneurones to induce neural oscillation in the isolated brain declines to near zero within 1 h after transection of the cerebrobuccal connectives, suggesting that this capacity is normally maintained by ascending information from the buccal ganglion.
The results show that this motor system conforms to a widely applicable general model of the neural control of rhythmic behaviour, by which independent neural oscillators distributed widely in the central nervous system are coupled together to produce coordinated movement.
Rhythmic movements can in principle originate by one of two different neural mechanisms. First, the underlying rhythm may be generated by a single ‘master’ neural oscillator, the output of which is then imposed upon follower neurones. Second, the rhythm may be generated simultaneously by several independent neural oscillators, which are coupled together to produce coordinated movements. The second mechanism, i.e. multiple, coupled oscillators, has been documented for several motor systems in both invertebrates (for reviews see Delcomyn, 1980; Kennedy & Davis, 1977) and vertebrates (Grillner, 1975, 1981). Recently, however, it has been proposed that the buccal motor rhythm (s) of the mollusc Pleurobranchaea is generated by a single oscillator located only in the buccal ganglion (Cohan & Mpitsos, 1983). This interpretation stands in contrast to our earlier deduction that the buccal motor system of Pleurobranchaea contains independent neural oscillators in the brain and buccal ganglion (Davis, Siegler & Mpitsos, 1973).
The question raised by the interpretations of Cohan & Mpitsos require resolution for two reasons. First, if true, the buccal motor system of Pleurobranchaea would represent an exception to an apparently widespread rule of the central organization of motor systems. Second, our continuing studies of the neural mechanisms of associative learning in this motor system (Davis et al. 1974, 1983; Davis & Gillette, 1978; Davis, 1983) depend upon accurate background information on the underlying neural organization of the buccal motor system. Accordingly, we have tested the hypothesis proposed by Cohan & Mpitsos. Our results demonstrate the existence of an independent neural oscillator(s) in the brain, which is normally coupled to the buccal oscillator(s). Our study therefore shows that this motor system operates by the coupling of independent, distributed neural oscillators. The results also provide evidence regarding the neural mechanisms by which these independent neural oscillators are coupled.
MATERIALS AND METHODS
Experiments were performed on the carnivorous marine gastropod mollusc Pleurobranchaea californica. Specimens were obtained by. trawling in Monterey Bay at depths of 60–90 m using the UCSC research vessel Scammon. Prior to experiments animals were maintained in separate plastic containers in fresh, running sea water at ambient temperatures (11–17 °C) at UCSC’s Long Marine Laboratories and fed raw squid on a weekly schedule. Animals used for experiments were 70–300 ml in volume.
Neurophysiological experiments were performed on the isolated central nervous system (CNS), consisting of the brain (cerebropleural ganglion) and buccal ganglion, connected by the paired cerebrobuccal connectives (CBCs). All other ganglia were removed prior to experiments. Preparations were pinned and/or glued with Super-Glue (Duro Corp.) to the Sylgard (Dow Chemical) floor of an experimental chamber in cold (12 ± 1 °C) sea water. The brain was manually desheathed using fine forceps. Glass capillary microelectrodes (tip resistance, approximately 10 MΩ measured in sea water) were inserted under visual control into previously identified feeding command interneurones in the brain, including the phasic paracerebral command interneurones (PCps) and the polysynaptic excitors (PSEs) of the PCps (Kovac, Davis, Matera & Gillette, 1982; Kovac et al. 1983). These feeding command interneurones were identified using criteria developed previously (ibid.), including position of the soma in the brain, presence of a characteristic cyclic inhibition, presence of a descending axon in the ipsilateral CBC, and the capacity to elicit cyclic motor output from efferent nerves that innervate buccal muscles. Studies on whole animal preparations have shown that tonic depolarization of these command interneurones induces ingestion, rather than egestion, and that these neurones are active during ingestion but silent during egestion (Gillette, Kovac & Davis, 1978, 1982). The identification of the tor programmes as those that normally underlie ingestion (feeding) rather than egestion was accomplished using several qualitative and quantitative criteria developed previously (Croll & Davis, 1981, 1982; R. P. Croll, M. P. Kovac & W. J. Davis, in preparation; R. P. Croll, M. P. Kovac, W. J. Davis & E. M. Matera, in preparation).
Extracellular recordings were made with polyethylene or glass capillary suction electrodes from two nerves of the buccal ganglion, namely the radula retractor nerve or root 1 (rl) and the radula retractor nerve or root 3 (r3). Extracellular recordings were also routinely made from three brain nerves, the mouth nerve (MN), the tentacle nerve (TN) and the small oral veil nerve (SOVN) (Davis et al. 1973). Recordings were amplified and displayed on a Tektronix oscilloscope for filming with a 70 mm oscilloscope camera, and/or recorded on a Brush-Gould eight-channel pen recorder. In order to uncouple the brain and buccal ganglion, the CBCs were cut with fine dissecting scissors, usually near the buccal ganglion, but also in the middle and in some cases at their exit from the brain.
In several previous papers it has been reported that tonic depolarization of single feeding command interneurones (PCps or PSEs) in the isolated CNS preparation causes cyclic motor output (e.g. Gillette et al. 1978, 1982; Kovac et al. 1982,. 1983). This motor output has been shown to represent the ingestion (feeding) motor programme (R. P. Croll, M. P. Kovac, W. J. Davis & E. M. Matera, in preparation). The same result was obtained routinely in the present work on the isolated CNS preparation. Injection of tonic (d.c.) current into the soma of a PSE and/or a PCp caused cyclic bursts of action potentials in these command interneurones and simultaneous rhythmical, coordinated extracellular bursts in three efferent nerves of the brain and buccal ganglion (e.g. Fig. 1).
In order to test the hypothesis that the brain contains neural oscillator(s) independent of those demonstrated previously in the buccal ganglion, experiments similar to that shown in Fig. 1 were performed on the isolated brain. Fig. 2 shows an experiment in which four command interneurones were simultaneously penetrated and stimulated. Records from only three command interneurones are shown: the fourth was monitored with an audio system. Immediately (2 min) prior to the recording shown in Fig. 2A, both CBCs were cut. The sudden and large synaptic input to the brain caused by severing these central connectives induced sustained cyclic activity in the command interneurones and in the efferent nerves (Fig. 2A). The production of this cyclic activity does not alone constitute rigorous evidence for a separate brain oscillator(s), however, since activation of CBC oscillatory properties (Cohan & Mpit-sos, 1983) by injury currents could in principle underlie the rhythmic brain activity observed immediately after cutting the CBCs (Fig. 2A).
Over a period of minutes, the cyclic brain activity induced by CBC transection gradually declined. During this time the frequency of the cyclic rhythm decreased from the initial value of about 0·1 Hz to less than 0·003 Hz (Fig. 2B, C). When the activity induced by cutting the CBCs had fully subsided (Fig. 3A), tonic (d.c.) current was injected into the somata of all four impaled command interneurones simultaneously. Such stimulation of the command interneurones in the now isolated and quiescent brain immediately induced vigorous, cyclic efferent activity in the SOVN and MN, and coupled cyclic activity in the command interneurones themselves (Fig. 3B). The frequency of this induced oscillation was approximately 0·1 Hz.
This rhythmic neural activity persisted for as long as the stimulus was maintained and stopped immediately upon cessation of the stimulus (Fig. 3C, D). The frequency of the rhythm could be more than doubled (>0·2 Hz) by increasing the amount of current injected into the four impaled command interneurones (not shown). Therefore this cyclic neural activity in command neurones and efferent nerves was presumably causally related to the injection of current into the command inter-neurones. Induction of such rhythmic neural activity in the isolated brain occurred even when the CBCs were transected at their exit from the brain, indicating that the underlying neural oscillator was located within the brain.
A similar experiment on a different preparation is quantitatively illustrated in Fig. 4. Here two feeding command interneurones, a PSE and an ipsilateral PCp, were stimulated together and separately while motor activity was recorded from ipsilate brain nerves (MN, TN) and nerves of the buccal ganglion (rl and r3). Only the activity of the MN is presented in Fig. 4 as an index of rhythmic motor output. Tonic stimulation of either the PSE or the PCp alone or together prior to transecting the CBCs caused rhythmic efferent output in the MN (Fig. 4) and in the other three recorded nerves (not shown). As documented in earlier work (Kovacet al. 1983; R. P. Croll, M. P. Kovac, W. J. Davis & E. M. Matera, in preparation), the PSE was more efficacious than the PCp in eliciting rhythmic motor output (Fig. 4). Cutting the CBCs induced vigorous cyclic activity in the MN (Fig. 4) and also in the two impaled command interneurones and in the other three efferent nerves (not shown). As in the preceding experiment described (Fig. 3), this induced rhythmicity gradually subsided over the course of a few minutes (Fig. 4). Simultaneous intracellular stimulation of the two command interneurones in the now isolated and quiescient brain again induced cyclic motor output in thMN, although the frequency of the rhythm was lower than prior to cutting the CBCs (Fig. 4). As expected, cyclic activity in buccal nerves was not elicited by stimulation of command interneurones in the brain after cutting the CBCs.
The experiment illustrated in Figs 3 and 4 was performed on seven preparations. Without exception, tonic intracellular stimulation of one or more feeding command interneurones in the brain before cutting the CBCs induced vigorous rhythmic activ-B of brain nerves (SOVN and MN), and coupled bursting in the command neurones themselves, as exemplified in Fig. 1. Following transection of the two CBCs, tonic depolarization of one or more brain command interneurones caused cyclic bursting in the command interneurones themselves in all seven preparations. This central neural oscillation was accompanied by rhythmic discharge recorded extracellularly from peripheral brain nerves (MN, SOVN, TN) in three of the seven preparations, as exemplified in Figs 3 and 4. In the remaining four preparations, coupled cyclic extracellular activity in the brain nerves was absent or at least much less obvious immediately after the brain was isolated by transecting the CBCs (e.g. Fig. 5). These results demonstrate that the brain contains a neural oscillator(s) capable of producing cyclic activity in the feeding command interneurones, and that this oscillator(s) can operate without driving simultaneous cyclic activity in efferent brain nerves. It follows that the coupling of the brain oscillator with efferent neurones is variable, and may depend in part on input from the buccal ganglion. As discussed later, this variable coupling between the brain oscillator(s) and the motoneurones that it drives may partially explain the results of Cohan & Mpitsos (1983).
A pronounced effect of removing the buccal ganglion was observed in the polysynaptic pathways documented earlier (Kovac et al. 1983) from the PSEs to the PCps. Previously it was shown that the PSEs send descending axons to the buccal ganglion via the ipsilateral CBC, and that the PSEs produce a characteristic, long-latency chemical polysynaptic excitation of the PCps (ibid.). In Fig. 6A, for example, the amplitude of this polysynaptic excitation was 25 mV before cutting the CBCs. Immediately upon isolating the brain by transection of the CBCs, however, this component of the PCp response to PSE stimulation declined to approximately 8 mV (Fig. 6B ; note different voltage calibration from Fig. 6A). Even when the PSE was induced to fire at more than twice the discharge frequency than before CBC transection, this long-latency polysynaptic excitation was substantially reduced (50–75 %) in comparison with pretransection values (N = 4 preparations). These findings show that a neurone(s) that ascends in the CBC from the buccal ganglion, i.e. one or more of the previously identified corollary discharge (CD) interneurones (Davis et al. 1973, 1974), mediate the major part of the long-latency polysynaptic excitation of the PCps by the PSE.
In all of the experiments reported above, the oscillatory activity of the brain was elicited within a few minutes (<30 min) of transecting the CBCs. In a separate series of seven experiments on seven different preparations, the CBCs were transected before the brain was removed from the animal, and feeding command interneurones were stimulated intracellularly from 1 to 3 h later. In none of these cases did tonic intracellular depolarization of the feeding command interneurone(s) in the brain cause clear cyclic output in feeding nerves of the brain. In only two of the seven experiments did tonic intracellular depolarization of the command interneurones induce these neurones to fire in their normal pattern of cyclic bursts, and in both of these cases the rhythmicity was weak. In contrast, when the brain and buccal ganglion are left connected, rhythmic motor output can be driven for hours by intracellular depolarization of the command interneurones (not shown). Negative findings are by nature inconclusive, but these results are consistent with the hypothesis that ascending information from the buccal ganglion contributes to the capacity of the brain oscillator(s) to generate cyclic neuronal activity.
The main conclusion of this paper is that there exists within the brain of Pleurobranchaea a neuronal oscillator(s) that is independent of the separate oscillator(s) demonstrated previously in the buccal ganglion (Davis et al. 1973). The normal function of these coupled oscillators is presumably to produce the cyclic motor output that underlies feeding behaviour in this mollusc. The evidence for this identity is as follows. First, previous studies on behaving whole animal preparations have shown that intracellular stimulation of the same command interneurones as stimulated here, namely the phasic paracerebral neurones or PCps, induces only feeding movements (Gillette et al. 1978, 1982). Second, these same command neurones are excited by food stimuli and are active cyclically in association with feeding movements in whole animal preparations, but are silent during egestion movements (Gillette et al. 1982, Fig. 8). Third, it has recently been shown that intracellular stimulation of the PCps elicits the identified ingestion motor programme or its characteristic components, but never the egestion motor programme and seldom its characteristic components (R. P. Croll, M. P. Kovac, W. J. Davis &E. M. Matera, in preparation). It has been suggested that the same neural oscillators that subserve feeding behaviour also subserve egestion movements (McClellan, 1982a,b‘, Croll & Davis, 1981, 1982; R. P. Croll, M. P. Kovac, W. J. Davis & E. M. Matera, in preparation), although independent command pathways for the two motor programmes have been documented (R. P. Croll, M. P. Kovac & W. J. Davis, in preparation; R. P. Croll, M. P. Kovac, W. J. Davis & E. M. Matera, in preparation).
The present experiments on the isolated CNS preparation have shown that cyclic motor output can be induced by stimulation of feeding command interneurones even when the CBCs are cut at their exit from the brain. Therefore this oscillation is generated within the confines of the brain, and does not depend upon interactions with neurones in the CBCs or other ganglia. Beyond this general localization of the brain oscillator(s), its nature is unknown. It is possible that the oscillation is produced entirely by neurones whose processes are normally restricted to the brain. Alternatively, or in addition, it is possible that the oscillation involves interactions with the terminal processes of ascending neurones from the buccal ganglion. Further experiments would be required to distinguish between these possibilities.
Comparison with previous studies
We proposed earlier that the buccal motor system of Pleurobranchaea operates by the coupling of independent neural oscillators located in the brain and buccal ganglion (Davis et al. 1973). This conclusion was based on the observation that cyclic efferent activity occurs in nerves of the isolated brain and the isolated buccal ganglion. Such motor rhythms occur spontaneously or can be induced by tonic extracellular stimulation of the appropriate nerves (ibid.). Cohan & Mpitsos (1983) confirmed these observations, but also reported that tonic electrical stimulation of the isolated CBCs caused cyclic output from this nerve, implying that neurones within the CBCs have oscillatory properties. They suggested that such properties could account for the observed oscillation of the isolated brain. Cohan & Mpitsos (1983) also found that rhythmic brain activity induced by tonic extracellular stimulation of a buccal nerve the stomatogastric nerve, ceased when ascending information in the CBCs was eliminated by sucrose gap or anodal block. From these observations Cohan & Mpitsos concluded that ‘the rhythmic, coordinated activity in the brain and buccal ganglion of Pleurobranchaea arises from oscillatory circuits (or neurones) located only in the buccal ganglia’ (Cohan & Mpitsos, 1983, p. 40).
The cause of these different interpretations cannot lie in the nature of the preparation used, since the isolated CNS preparation was used by Cohan & Mpitsos (1983) and in the present experiments. Instead we suggest two possible reasons for the primary observation of Cohan & Mpitsos that short-term interruption of ascending CBC activity blocked bursts of motor output in brain nerves. First, in their experiments the brain oscillation was activated via the buccal ganglion, by means of stomatogastric nerve stimulation. Therefore, when they blocked ascending activity they eliminated the only input to the brain that was activated using their methods. Under these conditions the cessation of brain oscillation is not unexpected. Second, Cohan & Mpitsos utilized motoneurone discharge as their primary index of brain oscillation. We have found here that the brain oscillator(s) is often uncoupled from the brain motoneurones that it drives (e.g. Fig. 5), and hence motoneurone discharge is alone inadequate as an index of central oscillatory activity.
Cohan & Mpitsos (1983) concluded further that the paracerebral command neurones cannot drive cyclic activity in the isolated brain (ibid. p. 35). The present experiments, however, illustrate that tonic intracellular stimulation of these identified feeding command interneurones invariably drives cyclic bursting in the command neurones (Fig. 5), and frequently elicits coupled rhythmic discharge in efferent nerves of the brain (Figs 3,4). The negative result of Cohan & Mpitsos may have resulted from stimulating only one paracerebral command neurone (approximately 6% of the known cerebral command population) at a time. Cohan & Mpitsos (1983) also concluded that ascending CBC activity is necessary and sufficient to cause brain oscillation (ibid. p. 34). The present studies show that ascending activity is not necessary to cause brain oscillation, although the possibility that it is sufficient has not been tested. Finally Cohan & Mpitsos (1983) concluded that the corollary discharge neurones that ascend from the buccal ganglion supply the drive as well as the pattern for rhythmic brain activity (ibid. p. 35). The present experiments show that an independent neural oscillator(s) in the brain plays these roles, although a contribution from ascending activity has not been excluded. Published data do not permit rigorous assessment of the relative roles of brain and buccal oscillators, although previously we reported that the brain oscillator(s) is relatively labile in response to stomatogastric nerve stimulation in comparison with the buccal oscillator (Davis et al. 1973).
Coupling of independent neural oscillators
In combination with previous experiments, the present studies suggest two mechanisms by which coupling between brain and buccal oscillators of Pleuro-branchaea might be accomplished. First, we have observed here that cutting the CBCs eliminates most of the previously demonstrated (Kovac et al. 1983) long-latency chemical polysynaptic excitation from the PSEs to the PCps. This finding indicates that most of this excitation is mediated by ascending neurone(s) from the buccal ganglion, i.e. the corollary discharge (CD) neurones. The cyclic bursts of ascending action potentials that occur in the CD neurones (Davis et al. 1973) would by this mechanism deliver phasic entraining excitation to the command neurones m the brain. Second, we observed here that cyclic inhibition of the command neurones is weakened but not eliminated by isolating the brain (cf. Figs 2 and 3). Therefore, the ‘cyclic inhibitory network’ or CIN studied previously (Gillette et al. 1982) is at least partly located in the brain. Stimulation of single CD neurones causes polysynaptic inhibition of the brain command neurones (Gillette et al. 1978), perhaps via the CIN. Cyclic CD activity could by this arrangement deliver phasic entraining in-hibition to the brain command neurones.
In addition to these postulated ascending excitatory and inhibitory coupling mechanisms, the command interneurones themselves discharge cyclic bursts of action potentials that descend to the buccal ganglion. This cyclic discharge also contains information on the timing of brain oscillator(s) activity, and could be used to couple the brain and buccal oscillations. Consistent with this hypothesis, intercalated bursts of impulses in the PCps advance or retard the phase of buccal oscillation (Gillette et al. 1982).
Maintenance of oscillatory capacity by ascending information
The present experiments show that the capacity of the command interneurones to elicit cyclic motor output declines rapidly after the CBCs are cut. Therefore ascending information from the buccal ganglion may help maintain the oscillatory capacity of the brain oscillator(s). Similarly, in the stomatogastric motor system of the lobster, input from the stomatogastric nerve contributes to the oscillatory capacity of neurones controlling both the gastric (Russell, 1976) and the pyloric (Russell & Hartline, 1978, 1982; Selverston & Miller, 1980) rhythms.
In the buccal motor system of Pleurobranchaea, the ascending information that is hypothesized to help maintain the oscillatory capacity of the brain is presumably carried by the CD neurones, but its nature is unknown. In the lobster stomatogastric system, dopamine has been circumstantially implicated as playing such a role (e.g. Anderson, 1977; Raper, 1979; Selverston & Miller, 1980; Kushner & Maynard, 1975; Friend, 1976). Similarly, various pharmacological agents sustain or release oscillatory neural activity in vertebrates (Ayers, Carpenter, Currie & Cinch, 1983; Buchanan & Cohen, 1982; Cohen & Wallen, 1980; Jankowska, Jukes, Lund & Lundberg, 1967a, bPoon, 1980; see Grillner, 1981, for a review). Such a possibility requires further investigation in the buccal motor system of Pleurobranchaea.
Distributed neuronal oscillators controlling rhythmic behaviour
The present results show that rhythmic cerebro/buccal motor output in Pleuro-branchaea is controlled by a separate and independent neural oscillator(s) located in the brain, in addition to a neural oscillator(s) demonstrated previously in the buccal ganglion (Davis et al. 1973). Cohan & Mpitsos (1983) have described a possible third source of neural oscillation, namely, the intrinsic oscillatory capacity of neurones in the CBCs, resulting presumably from properties of axonal membranes. The nerves of the two pedal ganglia also discharge cyclically and in phase with rhythmic motor output from the brain and buccal ganglion (Davis et al. 1973; Lee & Ligeois, 1974; G. Stoner&W. J. Davis, unpublished data). Therefore the pedal ganglia may contain yet additional neural oscillators for the control of this motor output. These studies electively indicate that the rhythmic cerebro/buccal motor programme(s) of rleurobranchaea is generated by the coupling of multiple neuronal oscillators distributed widely throughout the nervous system.
The organization of this motor system therefore corresponds to the general principle that rhythmic patterns of behaviour are controlled by multiple coupled oscillators. This principle appears to apply to the leech swim system (Friesen, Poon & Stent, 1978; Kristan, 1974; Kristan & Calabrese, 1976; Kristan & Guthrie, 1977; Stent et al. 1978; Weeks, 1981), the leech heartbeat system (Peterson & Calabrese, 1982; Peterson, 1983a,b), the lobster stomatogastric ganglion (Miller & Selverston, 1982; Russell, 1976; Russell & Hartline, 1978, 1982; Selverston & Miller, 1980), the crustacean swimmeret system (Hughes & Wiersma, 1960; Wiersma & Ikeda, 1964; Stein, 1971, 1974), the swim system of Tritonia (e.g. Getting, 1983), and mammalian locomotor systems (Grillner, 1975, 1981). All of these motor systems share the common feature that coordinated, rhythmic movements are produced by independent neural oscillators distributed bilaterally and intersegmentally in the nervous system. Such decentralization of coupled neural oscillators presumably confers operational stability and control flexibility upon the corresponding motor systems, by enabling a degree of local central and reflex autonomy while nonetheless assuring coordination of widely separated body parts or appendages.
This paper was written in part during the tenure of an Alexander von Humboldt Senior Scientist Award to WJD, who thanks Professor H. Markl (Universitat Konstanz) and Professor F. Huber (Max-Planck Institut fur Verhaltensphysiologie, Seewiesen) for hospitality and excellent facilities. We thank also Professor A. I. Selverston for constructive criticism of the manuscript. Supported by NIH Research Grant NS-09050 and NSF Research Grant BNS-8110235 to WJD.