Proprioceptive Control of the Bilaterally Organized Rhythmic Activity of the Oesophageal Neuronal Network in the Cape Lobster Jasus Lalandii

Rhythmic motor behaviour is controlled by central pattern generators that can sometimes function without sensory feedback (Wilson, 1968; Grillner, 1975). This is apparent for cases in which the motor output can be recorded in an isolated nervous centre in vitro (i.e. after total deafferentation). Such preparations, which allow analysis of motor patterns in terms of the activity of single cells, have been very useful for studying the organization of central pattern generators (see Fentress, 1976).

Nevertheless, in intact animals, the expression of such generators is controlled by sensory activity, notably by proprioceptive feedback (Pearson et al., 1973; Carlson, 1977; Stent et al. 1978). Moreover the central nervous system can in turn modulate the sensory pathways, and enhance or decrease the efficacy of sensory inputs on the motor activity during its execution (Stein, 1978). Despite the advantages of isolated preparations for studying the mechanism by which this is achieved, they have seldom been used. This may be due to the technical difficulties involved in isolating the motor and integrative centres, as well as identified and controllable sensory input.

The stomatogastric nervous system of decapod Crustacea, particularly its oesophageal components, is a favourable preparation for such studies. The preceding paper (Moulins & Nagy, 1980) has shown how the oesophageal motor behaviour is organized by a neuronal network distributed in three different centres, and how the activity of this network can be monitored through a single motor neurone functioning with several spike-initiating sites. This preparation has been recently developed by isolating in association with the oesophageal nervous centres, a pair of mechanoreceptors innervating the foregut (the posterior stomach receptors, PSRs) and an integrative centre connecting them with the oesophageal centres (the suboesophageal ganglion) (Nagy, 1977).

In the present work we have analysed the action of the PSRs’ inputs on the oesophageal network activity to answer two main questions:

• (a) How does a single receptor project on a distributed neuronal network, and what are the mechanisms by which the central nervous system controls the efficacy of these projections ? The preparation used is favourable because the integrative premotor step is anatomically distinct from the motor step, and the motor network is distributed in three anatomically distinct centres. It is possible to isolate reversibly the different elements. It is shown that a single receptor (each of the two PSRs) projects on to the three parts of the distributed oesophageal network, but that only one of these projections is effective at a given time. This control of efficacy of the PSRs’ different projections takes place at the level of the oesophageal motor network itself, although the PSRs’ inputs are integrated in the suboesophageal ganglion before reaching the oesophageal network.

• (b) How does a rhythm pattern generator react to a sensory input rhythmically activated by other motor systems ? Here again the preparation is favourable, as the PSRs respond to rhythmical stomacal movements and not to oesophageal ones, although the stomacal motor activity and the oesophageal motor activity are interrelated. It is shown that the oesophageal rhythm can be entrained by rhythmical PSRs’ inputs over a range of periods, which can include that of the spontaneous gastric rhythm. It is thus suggested that the PSRs allow these two rhythms to be coordinated through a peripheral loop. The participation of PSRs in the coordination of different motor sequences of the foregut is also discussed.

Male and female cape rock lobsters, Jasus lalandii, were used. Most of the experiments were performed on isolated nervous systems (in vitro preparations), but some were done using minimally dissected animals (semi-intact preparations).

Animals were restrained in a tank filled with refrigerated and oxygenated artificial sea water. The cephalothorax was opened dorsally to reach the foregut, and nervous activity was recorded with polyethylene suction electrodes. Identification of motor and sensory units was based upon previous anatomical and physiological works (Maynard & Dando, 1974; Selverston et al. Vedel & Moulins, 1977).

We enlarged the ‘combined preparation’ of Selverston et al. (1976) (stomagastric ganglion, oesophageal ganglion, commissural ganglia and the nerves connecting them), by the addition of the suboesophageal ganglion and the two posterior stomach nerves (psns) (Fig. 1; stomatogastric ganglion not shown). Isolated with each psn was a group of sensory cell bodies and 1 cm length of their dendrites, which comprise the posterior stomach receptor (PSR). After dorsal opening of the cephalothorax under artificial sea water, the distal part of the PSR dendrites was rapidly severed from the stomach. The stomatogastric ganglion and its nerves were dissected out of the gut and pulled forward, then the stomach was severed from the oesophagus and removed with the hepatopancreas, care being taken to preserve the stomatogastric nerve. The oesophageal and commissural ganglia and their connecting nerves were dissected around the oesophagus. The ventral endophragmal skeleton was removed to dissect out the perioesophageal connectives, suboesophageal ganglion and proximal part of the psns. The isolated nervous system was then pinned out in a Sylgard-lined petri dish (Fig. 1B). Haematopoietic and connective tissues were removed from the nerves. The oesophageal ganglion was desheathed to expose the cell bodies of oesophageal neurones. This dissection takes from 7 to 8 h, but if the saline is kept cold and suitably oxygenated, the oesophageal nervous system can operate for another 6-8 h, with a rhythm similar to that recorded from freshly dissected, semi-intact preparations. For description of recording techniques, see Moulins & Nagy (1980).

Impulse conduction was reversibly blocked along the connectives to determine central pathways of neurones. A small pool of petroleum jelly was made around a previously desheathed segment of the connective. When the pool was perfused with isotonic sucrose solution, conduction was blocked after 3 or 4 min; normal conduction was re-established 3-4 min after saline again flowed through the pool.

The oesophageal rhythm, which consists of regularly alternating bursting activity of constrictor and dilator motor neurones, is organized by a neuronal network distributed in three different centres: two commissural ganglia and the unpaired oesophageal ganglion (Spirito, 1975; Selverston et al. 1976; Moulins & Vedel, 1977). The rhythm is generated by an oscillator located in each commissural ganglion. In vitro, the activity of each oscillator consists of series of bursts separated by silent periods. The series of busts of one oscillator regularly alternate with those of the contralateral oscillator (Moulins & Nagy, 1980).

The activity of this distributed network can be monitored by that of the main oesophageal dilator motor neurone (OD1), whose anatomy and functioning were described in the preceding paper (Moulins & Nagy, 1980). OD1’s cell body lies in the unpaired oesophageal ganglion, and each of the two main branches of its axo reaches one commissural ganglion (Fig. 1). OD1 possesses three spike initiating sites, one in each of the three ganglia. Each commissural spike initiating site is driven by the oesophageal oscillator that lies in the corresponding ganglion (Moulins & Nagy, 1980). Thus, by considering the activity of OD1, it is easy to determine which of the two commissural oscillators is active at any time.

Each posterior stomach nerve (psn) arises as a branch of the outer mandibular nerve, which enters the suboesophageal ganglion (Fig. 1 A). The distal part of each psn comprises a group of sensory cell bodies and their 2 cm long dendrites, which form the PSR. The dendrites of the PSRs ramify in the soft tissue surrounding the posterior arch of the gastric mill. This arch supports the medial cuticular tooth of the stomach. Because the soft tissue is easily deformed, the PSRs can be stimulated by anything modifying the stomach volume. The records in Fig. 2 show that the PSRs are rhythmically stimulated by the motor activity of the gastric mill (trituration) and by the motor activity of the cardiac sac (temporary food storage). Activity of a psn is recorded on a semi-intact preparation in response to spontaneous movements of the foregut, which are driven by stomatogastric motor neurons. Rhythmical discharge of some of these motor neurones is recorded on the anterior lateral nerve (Fig. 2, aln). It is possible to recognize: (1) regular bursts of the four electrically coupled motor neurones GM, which drive the power-stroke muscles of the medial tooth of the gastric mill: (2) rhythmical bursts with a longer period. These are comprised of a small unit, CD1, a motor neurone which innervates dilator muscles of the cardiac sac, and fires with the cardiac sac rhythm (Moulins & Vedel, 1977). It appears that the PSR firing is driven by the rhythmical activity of the gastric neuronal network There in phase opposition to the GM neurons) as well as that of the cardiac sac network (here in phase with CD1). Thus the PSRs have a periodic activity pattern. They provide a rhythmical input to the stomatogastric nervous system, showing several bursts frequencies depending on which neuromuscular system calls them into play.

In isolated preparations, electrical stimulation of the dendrites of the PSRs modifies the oesophageal activity (Fig. 3). The stimuli we used consisted of short trains (0·15−0·20 s) of 1 ms square pulses at a frequency of 50 Hz. These stimuli were always delivered during the second half of the oesophageal period (see p. 242).

Such a stimulus elicits a burst of OD1 spikes, and all the neurones involved in a normal oesophageal sequence, as recorded extracellularly on the sons, are stimulated.

Stimulation of a PSR activates the oesophageal neuronal network. Because this network is distributed in different centres, we can ask which part of it is stimulated, and which of the two symmetrical oscillators is induced to fire. This question can be answered by looking at the activity of OD1 after PSR stimulation. If the left PSR is stimulated (Si), the elicited OD1 burst originates from spike initiating zone b (Fig. 4 A; b) in the left commissural ganglion (in Fig. 4 B2, the sequence of arrival of OD1 spikes at the extracellular recording sites is 1, 2, 3, 4). Conversely, if the right PSR is stimulated (S2), the elicited OD1 burst originates from the spike-initiating zone in the right commissural ganglion (Fig. 4A; c) and the sequence of extracellular recordings is reversed (Fig. 4C). The two kinds of OD1 burst are similar in duration, number of spikes and latency after stimulation. The terminal part of each record shows that not only OD1, but also most of the oesophageal units recorded in sons come from the left ganglion after S1 stimulation, and from the right one after S2 stimulation. Therefore, in isolated preparations, the electrical stimulation of a given PSR evokes the response of only one of the two oesophageal oscillators that located in the ipsilateral commissural ganglion.

There are at least two possible explanations for the observation that each PSR activates only the ispilateral oesophageal oscillator: either each PSR projects only on to the ipsilateral part of the oesophageal network; or each PSR projects bilaterally on to the different parts of the oesophageal network, but the contralateral projection ineffective.

It has not been possible to record the activity of the PSR sensory pathways in the connectives between suboesophageal ganglion and the commissural ganglia. It was, however, possible to unmask the contralateral effects of either PSR by reversibly interrupting its ipsilateral pathway. This was done by placing a sucrose gap on one of the connectives between the suboesophageal ganglion and the commissure (Fig. 5 A). When conduction in the left connective tract is so blocked, electrical stimulation of the left PSR elicits an OD1 burst which originates at the spike initiating zone in the right (contralateral) commissural ganglion (Fig. 5 A, c) (In Fig. 5 C the spikes of OD1 are recorded extracellularly in the order 4, 3, 2, 1, i.e. the inverse of that in the contract experiment of Fig. 5 B). The response latency is also 25-40% longer. The original response returns when normal conduction is restored (Fig. 5D). In each case the other oesophageal dilator units, which fire after OD1, originate in the same commissural ganglion as do the spikes of OD1. Therefore, even though each PSR excites only the ispilateral part of the oesophageal network, its inputs project on the oesophageal oscillators in both commissural ganglia via a direct and a crossed pathway; but the contralateral projection is normally ineffective.

It is possible to specify the pathways of the PSR inputs by checking for the existence of a possible crossing through the commissure (Fig. 6A, com). The right peri-oesophageal connective was cut between the commissure and the commissural ganglion (Fig. 6A). Under these conditions electrical stimulation of either PSR elicits an oesophageal burst originating in the left commissural ganglion. The burst resulting from stimulation of the right PSR, via the crossed pathway (Fig. 6D), shows a latency 25 % longer than that resulting from stimulation of the left PSR, which travels via the direct pathway (Fig. 6C). If impulse conduction is then blocked along the left connective tract by a sucrose gap (Fig. 6B), stimulation of neither the left psn (Fig. 6E) nor the right pm (Fig. 6F) produces an oesophageal response, although the commissure remains intact. When impulse conduction is restored, the response returns (Fig. 6G, H). It must be concluded that the only site of crossing over of PSR information is in the suboesophageal ganglion.

The preceding results have shown that although each PSR projects bilaterally on to both oesophageal oscillators, only one of these projections is effective (i.e. the ispilateral one when both connectives are intact). Therefore it appears that the CNS controls the sensory pathways to ensure the unilaterality of their effects. Because the branching point between the direct and the crossed sensory pathways lies in the suboesophageal ganglion, at least two possibilities for the central control of the sensory inputs may be considered: firstly, this control may take place in the suboesophageal ganglion, using the samples of oesophageal activity which go to this ganglion (see preceding paper, Fig. 13); secondly, this control may be effected by the oesophageal motor network itself. To test these two possibilities, all of the direct connexions between the two parts of the oesophageal motor network were severed by cutting the sons, ions and the commissure (Fig. 7 A). Each ganglion then produces oesophageal bursts with its own rhythm (Fig. 7C). Because the soma of OD1 has been separated from its two commissural spike initiating zones, no activity is recorded in the cell body. In such conditions, electrical stimulation of the left psn (Fig. 7 A, S1) simultaneously elicits an oesophageal burst from both commissural ganglia (Fig. 7D). Thus the ipsi- and contralateral projections of a PSR on the oesophageal network in the commissural ganglia may both be expressed freely only when the two parts of the network are independent. This means that the central control of the sensory pathways from the PSR occurs at the level of the motor network, and not in a premotor centre. This control requires the integrity of the connexions between the elements of the distributed oesophageal network.

Cutting some of these connexions can also unmask a PSR excitation of OD1 in the oesophageal ganglion, which involves its somatofugal spike initiating site. If the nerves connecting the two commissural ganglia are cut except the ions (Fig. 7 B), each oesophageal oscillator, bursting independently, stimulates the oesophageal spike-initiating zone a of OD1 through the ions (Fig. 7E). A small burst of OD1 somatofugal spikes is produced with each commissural burst. When the two oesophageal oscillators are induced to fire together by a PSR stimulation (Fig. 7E, Sr), OD1 gives a stronger somatofugal burst. These somatofugal bursts of OD1 are usually not elicited by the stimulation of PSRs when the sons are intact and allow control of OD1 activity by the commissural ganglia.

Since the central control of the sensory pathways from PSRs does not occur in the suboesophageal ganglion, it might be assumed that the PSR primary afferent fibres project directly in the commissural ganglia. However, the following evidence suggests’ that the PSR projections on the oesophageal network are not direct. First, there is a difference of 50−80 ms between activation, by a psn stimulation, of the oesophageal network in the ipsilateral commissural ganglion compared with activation of the network in the contralateral ganglion (Figs. 5, 6C, D). Because the same commissural elements are involved in such activations, the difference must be due to the crossing in the suboesophageal ganglion. The mean rate of impulse conduction in primary sensory fibres, as measured on a psn, is about 1·35 m s^-1. If these fibres passed through the suboesophageal ganglion without synapsing, impulses from the cell bodies of a PSR would reach a commissural ganglion after 70−80 ms. But a PSR stimulation elicits an oesophageal burst after a minimum delay of 200 ms (Figs. 4B, C; 5B, D; 6 C, D). This delay may be due in part to synapses in the commissural ganglion, but it also suggests the existence of interneurones intercalated in the pathway at the level of the suboesophageal ganglion. It is thus highly probable that the inputs from the PSRs are integrated in the suboesophageal ganglion before projecting on to the different parts of the oesophageal network in the commissural ganglia. Nevertheless, as shown in the preceding paragraphs, this pre-motor integration step is not involved in the central control of the efficacy of the sensory pathways which occurs at the motor network itself.

The oesophageal activity is phasically influenced by stimulation of a PSR. But in intact animal, PSRs are rhythmically induced to fire by movements of the foregut (see Fig. 2). We can now ask to what extent the oesophageal rhythm is influenced by such rhythmical inputs.

For studying the central control of the sensory pathways from PSRs, we used stimulations which are always delivered during the second half of the oesophageal period. In these conditions an oesophageal burst (as monitored by OD1 activity) is elicited after a delay of about 200 ms, and occurs before the expected time of occurence of the next spontaneous burst. In other words, such a PSR stimulation induces a positive phase shift of the oesophageal dilator burst.

Because a natural PSR stimulation can occur randomly in the oesophageal rhythm period, we examined the effects of electrically stimulating the PSRs at different times in the period of the oesophageal cycle. The PSR stimulated was always that ipsilateral to the active oesophageal oscillator. The stimulus used in these studies was a single pulse of current. Under these conditions, depending on the time of stimulus delivery within the oesophageal cycle, an oesophageal burst can be either positively or negatively phase-shifted by the stimulus (Fig. 8). If psn stimulation is given in the second half of the OD1 interburst interval, the next OD1 burst occurs sooner than it would have without stimulation. The oesophageal period is shortened and a positive phase shift of the oesophageal burst occurs (Fig. 8B). If stimulation is given near the end of an OD1 burst, the next burst is delayed and the oesophageal period is. lengthened; a negative phase-shift occurs (Fig. 8C). A. psn stimulation occurring the beginning of the OD1 interburst does not obviously modify the oesophageal period (Fig. 8D).

We have quantified these effects by plotting a phase-response curve (PRC) (Fig. 9B) in which the phase shift of the OD1 burst after a stimulation is expressed as a function of the stimulus phase. The phases here are defined relative to the oesophageal rhythm. As the free run period of this rhythm is somewhat variable, a reference period is defined as the average of the three unstimulated periods just prior to the stimulation (Fig. 9 A, Pm) (each period is measured from the midpoint of a burst to midpoint of the next burst). The stimulus phase (Φ stim) is the ratio of the stimulus latency (Fig. 9 A,L) to the reference period (Φ stim = L/P_m); the phase shift (ΔΦ) is difference between the reference period (P_m) and the stimulated period (P_g) as a function of the reference period (▲Φ=(P_m−P_t)/P_m= 1−(P_aIP_m)). If the burst of OD1 following a stimulus is delayed, P₈> P_m and ΔΦI< 0. If the burst following a stimulus is advanced, P” < P_m and ΔΦ > 0. As shown in Fig. 9B, the phase shift is negative for a stimulus phase of about 0·2 (stimulus occurring at the end of an OD1 burst), and the most positive for a stimulus phase of about 0-55 (stimulus occurring just after the middle of OD1 interburst). Thus, although the general effect of a PSR stimulation is to advance the occurrence of a subsequent OD1 burst (Fig. 9B, mean AΦD=≠+0.11), the occurrence of such a burst can also be delayed, depending upon the stimulus phase in the oesophageal cycle. The other oesophageal units recorded on a son (Fig. 8B, C, D, 1) follow the period variations of OD1. Therefore it may be concluded that each commissural network of oesophageal neurones shows a periodically varying sensitivity to PSR input.

It is known that if an oscillating system presents a periodically varying sensitivity to a rhythmical input, it can, between fixed limits, modify its own frequency to follow that of the input (Enright, 1965). The oscillator is thus entrained by the rhythmical input. Our results suggest that each of the two oesophageal oscillators has this necessary property for entrainment by the rhythmical PSR input. To demonstrate such an entrainment of each network, all the connexions between the two commissural ganglia are cut (see Fig. 7 A), so that each commissural ganglion generates oesophageal bursts at its own frequency (Fig. 7C). With the sons cut the free-run period of each oscillator is rather variable (mean values are 5·81 ±1·50 s for the left oscillator and 6·20 ± 1·148 for the right one). A PSR is then stimulated with repetitive electrical pulses, with a given period near the free-run period of each oesophageal network. The effects of such stimulation on the two independent oesophageal rhythms are illustrated in the curves of Fig. 10A and B. A curve showing the phase of oesophageal bursts in the stimulus period, and plotted in real time, is drawn for each stimulus period. The theoretical example of Fig. 10 C shows the measurements and calculations used in forming these curves (see legend). Each curve includes regions of absolute coordination (horizontal parts of the curves) where each electrical pulse is followed by an oesophageal burst at a fixed phase, separated by. regions of phase drift, where the oesophageal oscillator temporarily escapes from PSR influence and follows its own rhythm. If the stimulus period is longer than the free-run oesophageal period, phase drifts are negative and the curve is falling. If the stimulus period is shorter, the phase drifts are positive and the curve is rising. The starting position of each curve is arbitrary. The free-run period of the right oscillator indicated by curves of Fig. 10 B (between 5 and 6-6 s) fits well with the free-run period observed before stimulation (6·20 ± 1·14 s). For the left oscillator, the free-run period indicated by curves of Fig. 10A (between 4 and 5 s) is somewhat different from the free-run period observed before stimulation (5·81 ±1·50 s). This difference is most probably due to the variability of the free-run period when both oesophageal oscillators are independent.

It appears that with PSR stimulus periods near the free-run period of the oesophageal rhythm, the oesophageal oscillator in each commissural ganglion exhibits relative coordination, which includes periods of phase drift, and absolute coordination with the stimulus cycle. Oesophageal rhythm entrainment by rhythmical PSR inputs can also be illustrated by phase histograms of oesophageal bursts in the stimulus period (Fig. 10 D, E). For each stimulus period, the oesophageal bursts occur at a preferred phase in this stimulus period (0·22 in D, 0·12 in E). In Fig. 10D the preferred phase (0·22) in the stimulus period (4 s) corresponds to a phase shift of the oesophageal burst of+0·31, relative to the free-run period (5·81 s). This value is around the maximum possible phase advance predicted by the PRC of Fig. 9 and is produced by a stimulus whose phase in the oesophageal rhythm is 0·54. This fits well with the stimulus phase (0·53) measured for the entrainment corresponding to curve 4 3 in Fig. 10D. In Fig. 10E the preferred phase (0·12) in the stimulus period s) corresponds to a phase shift of the oesophageal burst of +0·19, relative to the ree-run period (6·2 s). As seen in the PRC of Fig. 9 such a phase shift is produced by a stimulus whose phase in the oesophageal rhythm is 0·67. This again fits well with the stimulus phase (0·71) measured for the entrainment corresponding to curve 5 s in Fig. 10E. The smaller values of each histogram represent the phase drifts seen between the horizontal portions of the curves in Fig. 10A and B. The stimulation period which produces the best entrainment is similar for the two ganglia. Therefore the oesophageal network in both commissural ganglia can be entrained by rhythmical inputs from the PSRs and then can be phase and time-locked by other central neuronal networks whose activities drive PSRs discharges.

The anatomy of the proximal parts of the posterior stomach nerves has until now been unclear (Dando, Chanussot & Nagy, 1974; Maynard & Dando, 1974). The psn’s were said to arise as a branch of the outer mandibular nerve in Homarus (Dando & Laverack, 1969; Wales, MacMillan & Laverack, 1976) and of the inner mandibular nerve in Panulirus and Panulirus (Dando & Maynard, 1974). Their sensory fibres were supposed to extend directly to the commissural ganglia in Cancer (Hermann & Dando, 1977). Our work on Jasus lalandii using a new isolated preparation including the stomatogastric nervous system and both psns (Nagy, 1977), has shown the following features :

• (1) each psn joints with an outer mandibular nerve which enters the suboesophageal ganglion, and not directly the commissural ganglion;

• (2) unlike what was reported to exist in Cancer (Hermann & Dando, 1977), a population of PSR primary sensory fibres synapses on suboesophageal interneurones. In other words, a pre-motor step in the sensory pathways of the PSRs exists in the suboesophageal ganglion;

• (3) these suboesophageal interneurones project on to the oesophageal network in the commissural ganglia.

It is not possible to completely exclude the possibility of PSR primary sensory fibres, which pass through the suboesophageal ganglion without synapsing, and thus project directly on a commissural ganglion. However, the delay between PSR stimulation and the response of the ispilatral oesophageal network, as well as the differences in delay between responses of the ispi- and contralateral oesophageal networks, strongly suggests that PSR inputs are integrated in the suboesophageal ganglion before projecting on to the oesophageal network. The integration of PSR inputs in several centres fits well with the proposed action of the PSRs in coordinating the activities of several central neuronal networks, as will be discussed later.

The oesophageal neuronal network is distributed in three nervous centers (Selver-ston et al. s.g’jb’, Moulins & Nagy, 1980). The data presented here show that the inputs from each PSR project simultaneously on to the different parts of the oesophageal network, (i.e. on the ipsi- and contralateral commissural ganglia and the oesophageal ganglion). A similar indirect bilateral projection occurs for sensory inputs from a superior laryngeal nerve to the two symmetrical pontic hemicentres which control swallowing in mammals (Jean, 1978). In this case the bilateral projection strengthens the synchrony of the discharge from the two hemicentres. This differs, however, from the situation in the lobster, in which only one of the multiple projections of the PSRs on to the oesophageal network is effective at a given time.

We have shown that the inputs from a given PSR are usually effective only on the insilateral part of the oesophageal network. The unilaterality of the PSRs’ effects is controlled by the oesophageal motor network itself, and requires the integrity of the interconnexions between the different parts of this distributed network. Such a center control of sensory pathways has been described in several other sensory-motor systems. This control generally functions to prevent maladaptive self-stimulation during execution of a programmed motor task (for a review, see Stein, 1978). In our case the motor program involves at least two oesophageal oscillators (Selverston et al. 1976; Moulins, Vedel & Nagy, 1979). Complex central interactions, which can induce the two oscillators to alternate, allow the oesophageal network to produce a unique and well determined motor output (Moulins & Nagy, 1980). Central control of the variable efficacy of the different PSRs projections probably allows the PSRs to influence preferentially the part of the network which is actually determining the oesophageal activity at a given moment, without eliciting an inappropriate discharge from the contralateral oscillator. This observation might be of general interest, since numerous behaviours (such as locomotion and breathing) involve bilateral nervous control centres which are influenced by bilateral proprioceptors through direct and crossed pathways. These different pathways must be considered in terms of the responsiveness of their respective targets at a given time.

It should also be noted that PSR inputs are controlled not in the suboesophageal premotor step of the sensory pathway, but in the motor network itself. This is different from the central control of tactile receptors in the crayfish abdomen during the tailflip, whose mechanism is one of the few which are well understood. In the crayfish abdomen, the control is exerted on the synapses between primary afferent fibres and first-order interneurons (Bryan & Krasne, 1977).

Rhythmical sensory inputs may have several functions in the execution of a central motor program: (1) to adapt expression of this program to environmental variations (Grillner, 1975; Bowerman, 1977; Clarac & Ayers, 1977); (2) to increase and stabilize the discharge frequency of an oscillatory system and to coordinate outputs of its follower neurones (Wendler, 1974; Kristan & Calabrese, 1976; Carlson, 1977); (3) to co-ordinate the activities of several oscillators (Pearson & Iles, 1973; Kristan & Stent, 1976; Bowerman, 1977). As oesophageal movements do not stimulate the PSRs it is unlikely that these last serve the first function mentioned above. This is probably done by mouthparts receptors, which monitor deformations of peribuccal areas (Moulins et al. 1969). The relationship between the PSRs and the oesophageal nervous system is, however, appropriate to the 2nd and 3rd functions.

For an input to modify the rhythm of an oscillator, the oscillator must show a periodically varying sensitivity to the input, i.e. a significant phase response curve (PRC) to this input (Enright, 1965; Pavlidis, 1973). This is true for endogenous oscillators (Ayers & Selverston, 1979; Benson, 1980; Pinsker, 1977 a)as well as network oscillators (Stein, 1976). If the input is itself driven by another oscillator, that last may govern the activity of the first (‘magnet effect’ of Von Holst, 1973). We have plotted a PRC of oesophageal rhythm, as reflected in the activity of OD1, to PSR’s input.

Theoretically such a curve may be used to predict the limits of stable entrainment (i.e. the range of input frequencies which elicit an oesophageal burst for each PSR stimulation), the lower and upper limits being respectively determined by maxim possible phase delay and maximum possible phase advance (Pinsker, 1977 b;,Benson, 1980). The PRC shows that, in our conditions of simulation, a PSR rhythm between 9% slower and 27% faster than the oesophageal free-run rhythm would produce such an absolute co-ordination. But the oesophageal free-run period is variable, and when the oscillator rhythm is not rigorously constant, a fixed stimulus period induces only a relative co-ordination (Ayers & Selverston, 1979). Indeed in our experiments PSRs produce such relative co-ordination of oesophageal rhythm with intervals of absolute co-ordination separated by phase drifts.

In intact and semi-intact animals the PSRs are stimulated by gastric mill movements and discharge with a gastric rhythm (Dando et al. 1974; present report). Are the PSRs able to entrain the oesophageal oscillators to the gastric rhythm ? We have shown that, in vitro, each PSR can entrain the oesophageal rhythm when it is induced to fire rhythmically with a period near the oesophageal period. Therefore, for the PSRs to entrain the oesophageal rhythm to the gastric one, the gastric and oesophageal periods must not differ greatly. Indeed, it appears that, even in an isolated preparation, the oesophageal and the gastric periods are sometimes the same (Selverston et al. 1976). Thus, the PSRs may be able to entrain the oesophageal rhythm at the gastric frequency in the intact animal. Some coordination with ratios of gastric: oesophageal cycle of 1:2 and 1 has also been observed (Selverston et al. 1976). But it has been shown that an oscillator which is entrained by a periodic input firing at nearly its period, will also be entrained (with a ratio different from 1:1) by the same input firing with a period close to double or half that of the oscillator (Wendler, 1974; Stein, 1977). Thus in these conditions, PSRs might again be able to entrain the oesophageal rhythm at the gastric frequency. It has been reported that in deafferented preparations some gastric neurones receive trains of epsp’s from the oesophageal neurones of the commissural ganglia (Selverston et al. 1976). Thus, central relations exist between the two rhythm generators. We propose that the PSRs pathway constitutes a peripheral loop through which the gastric and the oesophageal rhythm generators can be coupled, and that this loop can reinforce the central coupling which exists between them.

It appears that PSRs can also discharge with a cardiac sac rhythm (see Fig. 2),. and complex central coupling between the oesophageal and cardiac sac rhythms has been described (Moulins & Vedel, 1977). Here again a peripheral mediated coupling through PSRs may reinforce the central one.

Complex behaviour can be understood as the addition of discrete behaviours, each one being organized by a given neuronal network (Carlson, 1977). The different networks involved in such a complex behaviour must be linked to insure a coordinated motor activity. Rhythmic activity of the oesophagus is an example of a discrete behaviour integrated into a complex one, the motor activity of the foregut, the final goal of which is to ensure progression and mechanical processing of the food. The other components of this behaviour are the movements of the cardiac sac, tritutation by the gastric mill teeth, and food filtering by the pyloric stomach. Each of these is governed by an identified neuronal network (Selverston et al. 1976; Moulins & Vedel, 1977).

Movements elicited by some of these networks (gastric and cardiac ones) rhythmically stimulate the PSRs (Dando et al. 1974; present work). PSRs could be also stimulated by mouthparts movements during feeding, as some of their sensory come probably monitor mandibular movements (Wales et al. 1976). The PSRs in turn influence the rhythmic discharge of the gastric, pyloric (Dando et al. 1974; Hermann & Dando, 1977; Nagy, 1977) and oesophageal networks (present work). Therefore the PSRs have the necessary characteristics to contribute to the coordination of complementary behavioural sequences (feeding, swallowing, food processing in stomach).

We are grateful to Dr P. Dickinson for critical review of the manuscript.

This work was supported by the CNRS (ATP 3625) and the DGRST (Grant 77.7.1877).