ABSTRACT
In the isolated stomatogastric nervous system of the lobster Jasus lalandii, the strong modifications of the pyloric motor pattern induced by firing of the single anterior pyloric modulator neurone (APM) are due primarily to modulation by APM activity of the regenerative membrane properties which are responsible for the ‘burstiness’ of all the pyloric neurones and particularly of the non-pacemaker neurones (constrictor motoneurones).
This modulation has been studied under experimental conditions where the main extrinsic influences usually received by the pyloric constrictor neurones (intra-network synaptic interactions, activity of pacemaker neurones, and phasic central inputs from two premotor centres) are minimal. Under these conditions a brief discharge of neurone APM induces long plateaus of firing in all of the pyloric neurones.
The non-pacemaker neurones of the pyloric network are not simply passive follower neurones, but can produce regenerative depolarizations (plateau potentials) during which the neurones fire spikes. The ability of the pyloric constrictor neurones to produce plateau potentials (plateau properties) contributes greatly to the generation of the rhythmical pyloric motor pattern. When these neurones spontaneously express their plateau properties, firing of neurone APM amplifies these properties. When most of the central inputs usually received by the pyloric constrictor neurones are experimentally suppressed, these neurones can no longer produce plateau potentials. In such conditions, firing of the single modulatory neurone APM can reinduce plateau properties of the pyloric constrictor neurones.
In addition, firing in APM neurone slows down the active repolarization phase which terminates the plateau potentials of pyloric constrictor neurones. This effect is long-lasting and voltage-dependent.
Modulation by APM of the plateau properties of the pyloric neurones also changes the sensitivity of these neurones to synaptic inputs. This effect can explain the strong modifications that an APM discharge exerts on a current pyloric motor pattern. Moreover, it might render the motoneurones of the pyloric pattern generator more sensitive to inputs from a command oscillator, and contribute to switching on the pyloric motor pattern.
INTRODUCTION
The mechanisms by which central pattern generators are controlled, in particular how they are turned on and off and how their activity can be modulated by higher centres, have not yet been clearly elucidated. This problem can be approached using the stomatogastric nervous system of crustaceans, in which it has been shown that the motor output of a central pattern generator (the pyloric network) is lastingly modified by the discharge of a single interneurone (APM; Nagy & Dickinson, 1983). In some cases, the modifications of the pyloric pattern engendered by APM resemble those which correspond to the initiation of feeding in the intact animal. We can thus analyse the cellular mechanisms which underly the control of a pattern generator by asking what cellular properties of the pyloric neurones are altered by the discharge of the interneurone APM.
The ability of central pattern generators to produce rhythmical motor outputs depends not only on the synaptic connections within the pattern generator, but also on the cellular properties, such as endogenous bursting capabilities (Koester, Mayeri, Liebeswar & Kandel, 1974; Maynard & Selverston, 1975; Thompson & Stent, 1976) and post-inhibitory rebound (Siegler, Mpitsos & Davis, 1974) of the neurones comprising the generator. The rhythmic pyloric pattern is in part dependent on the existence of endogenous oscillators (pacemaker neurones) within the pyloric network (Maynard & Selverston, 1975; Selverston, Russell, Miller & King, 1976). In addition, all the pyloric neurones, including the non-pacemaking neurones, can produce long-lasting regenerative depolarizations, the ‘plateau potentials’, which are responsible for the burstiness of these neurones and thus play a major role in the generation of the pyloric rhythm (Russell & Hartline, 1978).
The expression of plateau properties in pyloric neurones depends upon inputs from higher centres, stimulation of the single input nerve to the stomatogastric ganglion being able to induce or amplify these properties in all the pyloric neurones (Russell & Hartline, 1978, 1981; Moulins & Cournil, 1982). The source of these inputs has, however, been obscure.
We show in this paper that one of these inputs is the single modulatory interneurone APM, which can induce plateau properties in the non-pacemaking pyloric neurones and amplify these properties in all the pyloric neurones, including the pacemakers. The induction and amplification of plateau properties in the pyloric neurones can fully explain the quantitative and qualitative modifications of the pyloric motor output provoked by an APM discharge (see Nagy & Dickinson, 1983).
MATERIALS AND METHODS
Experiments were performed on the isolated stomatogastric nervous system of male and female Cape lobsters, Jasus lalandii. With a few exceptions, the dissection, recording techniques and neurone identification were as outlined previously (Moulin & Nagy, 1981) and in the accompanying paper (Nagy & Dickinson, 1983). The isolated nervous system (Fig. 1A) was pinned on a Sylgard-lined Petri dish and maintained in oxygenated artificial sea water at 20 °C. Petroleum jelly barriers around the commissural ganglia (CG) permitted superfusion with a 0Ca2+ + 12mm-Co2+ saline (artificial sea water in which Ca2+ was replaced with Mg2+ and to which 12mm-Co2+ was added) to block synaptic activity in these ganglia. A similar barrier was built around the stomatogastric ganglion (STG) in those experiments in which this ganglion was perfused with 10−5M-picrotoxin (gift from Fluka) or with high K+ saline (three times normal K+, with the increased K+ replacing equivalent Na+). In most experiments involving injection of hyperpolarizing currents (see Figure legends), the neurones were penetrated with two microelectrodes, one for recording and one for injecting current. Current injected was monitored with a virtual ground circuit connected to the bath.
RESULTS
Background observations
In the isolated stomatogastric nervous system (Fig. 1A) of the lobster, the pyloric motor pattern, which is responsible for producing rhythmic movements of the pyloric filter, can be readily recorded. This rhythmic pattern is produced by the 14 neurones of the pyloric network (see Maynard, 1972; Maynard & Selverston, 1975; Selverston & Miller, 1980) and consists of bursts of action potentials produced alternately in the dilator and in the constrictor motoneurones. For the present purposes, we shall consider only a subset of the pyloric neurones: a group of three electrically coupled neurones, the two pyloric dilators (PD) and the anterior burster (AB), which are the pacemakers of the pyloric rhythm and can be considered as a single functional unit, and two types of constrictor motoneurones, the single lateral pyloric (LP) and the eight electrically coupled pyloric (PY) neurones. These neurones interact via a number of inhibitory synapses, shown in Fig. 1B. The dilators (PD) inhibit both types of constrictor (LP, PY); the constrictors LP and PY are mutually inhibitory; the constrictor LP also inhibits the dilators (PD).
The pyloric motor output (see Fig. 1B) is normally determined by the pacemaker activity of the dilator neurones (PD) and by the synaptic connections within the network (Maynard, 1972; Selverston et al. 1976). In addition, the pyloric rhythm is influenced by phasic inputs from the commisural ganglia (CG, premotor centres for the pyloric network) (Selverston & Miller, 1980; Robertson & Moulins, 1981). Because the goal of the present paper is to examine the action of APM’s discharge on the cellular properties of the pyloric neurones, particularly the non-pacemaking neurones (LP and PY), we attempted to study the pyloric network under conditions in which the influences mentioned above (activity of pacemaker neurones, phasic central inputs, intra-network synaptic interactions) were minimal. This can be realized in three ways: firstly by choosing preparations in which the background activity of PD neurones is minimal; secondly by experimentally blocking synaptic activity in the commisural ganglia, which suppresses phasic inputs to the pyloric network from these ganglia and greatly diminishes the activity of the pyloric pacemakers (PD); and thirdly by pharmacologically decreasing the influence of synapses within the network.
APM causes the pyloric neurones to fire in plateaus
When some of the influences that pyloric neurones usually receive are weakened (see above), a discharge of APM causes these neurones to fire in plateaus (Figs 1C, E, 2).
This happens when the pyloric pacemaker neurone PD is spontaneously minimally active. As was shown in the preceding paper (Nagy & Dickinson, 1983), spiking in APM when the pyloric rhythm is spontaneously weak strongly activates all the pyloric neurones, particularly the constrictors (LP, PY). In some cases this activation is accompanied by an extensive increase in the pacemaker activity of the dilators (PD) and hence an increase in the frequency of the rhythm; in other cases pacemaker activity is not activated. It is in cases of the latter category that the influence of PD is minimal. In such a case (Fig. 10), the major effect of an APM discharge is to provoke long plateaus of firing in the constrictor neurones (PY shown here). During these plateaus, spike frequency is elevated and corresponds to the long, intense bursts which can be recorded extracellularly (Fig. 1C, vlvn). At the peak of the effect, the inhibition by PD is considerably less effective than usual and cannot interrupt plateaus of the constrictor neurones (Fig. 1C, open triangles).
Such plateauing of pyloric neurones under APM influence is also obtained when synapses are blocked in commissural ganglia with 0Ca2++Co2+ saline (Fig.1D, E). This experimental situation not only minimizes phasic inputs from these ganglia on the pyloric network but offers an added technical advantage: under these conditions, and with the oesophageal nerve cut (on, Fig. 1A, D), extracellularly stimulating the inferior oesophageal nerves (Fig. 1A, ion, Fig. ID, 1 and 2) activates APM axons without activating other inputs to the stomatogastric ganglion (Fig. 1F, G; for further details see Fig. 1 legend). This is experimentally useful because the cell body of APM is often difficult and sometimes impossible to penetrate with a microelectrode (10–15 μm diameter, surrounded by a thick individual sheath).
In the experimental conditions of Fig. IE, duration of the plateau potentials induced in the pyloric neurones by APM firing is variable because it is largely determined by the inhibitory synapses within the pyloric network (see Fig. 1B). Therefore to observe more easily the effects of APM on individual pyloric neurones it is necessary, in addition to weakening descending inputs to the pyloric network, to weaken the synaptic interactions within the network itself. By bathing the stomatogastric ganglion in picrotoxin (Fig. 2A, rectangle), it is possible to block or partially block inhibition from the AB, LP and PY neurones (Marder & Paupardin- Tritsch, 1978; Bidaut, 1980). Under these conditions, plateaus considerably longer than usual are generated in both LP (35 s plateau, Fig. 2B) and PY (16 s plateau, Fig. 2C) when APM fires (6s discharge in these examples). The plateaus generated in PD under the influence of APM are still much shorter than those in the constrictors, but are nevertheless about ten times longer than usual (3 s in Fig. 2C). In picrotoxin, the duration of the plateaus is no longer limited by inhibitory synaptic influences within the network, as can be seen, for example, in Fig. 2C, in which the PY plateau is not terminated by the burst in the PD-AB group (the AB to PY synapse is picrotoxin sensitive). Therefore, when the pyloric neurones are nearly isolated from external influences, the final effect of an APM discharge is the induction of very long plateau potentials, which in turn lead to the firing of long and intense bursts of spikes.
APM induces plateau properties in the pyloric neurones
All the neurones of the pyloric network are capable of producing regenerative depolarizations or plateau potentials (Russell & Hartline, 1978), which appear to be of considerable importance in the generation of the pyloric rhythm. These neurones are characterized by a bistable membrane potential that can only transiently cross between two narrow ranges of membrane potential stability. As seen in Fig. 2C (for instance in PY), the plateau potential is comprised of an active transition from the lower level of stability to the depolarized stable level, then a period of time during which the neurone remains depolarized and finally an active repolarization which terminates the plateau. The depolarized stable level in the pyloric neurones is above the spike threshold, so that a burst of spikes, whose duration is determined by the length of time the cell remains at the upper level, is fired during each plateau potential. The transition between the two levels is an active process, but can be triggered by an external event (synaptic input or injected pulse of current). It is thus possible to test for the ability of a neurone to generate plateau potentials by the injection of brief pulses of current in its soma (Russell & Hartline, 1978; see Fig. 3A). A brief pulse (200ms) of depolarizing current triggers a longer (several seconds), regenerative depolarization (plateau) in the neurone (Fig. 3Ai). Injection of a brief pulse (200 ms) of hyperpolarizing current can prematurely trigger repolarization and terminate the plateau (Fig. 3A 2).
Using this method of testing for plateau properties we have shown that when inputs to the pyloric neurones are minimized (commissural ganglia synapses blocked and synapses in the stomatogastric ganglion weakened), neither LP nor PY exhibits plateau properties (Figs 3B and 4A, respectively). Simple depolarization or hyperpolarization of constrictor neurones does not restore their ability to generate plateaus. Thus these neurones have lost their plateau properties. The influence of APM alone is, however, able to restore them. When APM fires ionically at a low frequency, the same 200 ms depolarizing pulse injected into LP (Fig. 3C) or PY (Fig. 4B, C) produces a long-lasting plateau. This plateau potential can be prematurely terminated by a brief pulse (200 ms) of hyperpolarizing current (Fig. 4C), the membrane potential of the neurone then remains at its lower level of stability until the next depolarizing pulse. APM induced plateau properties in these neurones.
Fig. 5 further demonstrates that APM’s discharge does not simply provoke a depolarization of the pyloric neurones, but rather induces bistability, which is a major criterion for plateau properties. In this case, tonic firing of APM provokes long spontaneous plateaus in the neurone PY (Fig. 5A). Brief (100ms) pulses of hyperpolarizing current injected into the cell can then repolarize the neurone and terminate the plateaus; the cell remains silent at its lower level of stability for approximately 1·2 s (Fig. 5B). In contrast, the same neurone behaves differently when (APM being silent) it is simply depolarized with current injection until it reaches a spike frequency similar to that recorded during the plateau provoked by APM (Fig. 5C); pulses of hyperpolarizing current identical to those used in Fig. SB interrupt the neurone’s discharge only for the duration of each pulse, indicating that under these conditions (without APM firing), PY is no longer a bistable neurone.
APM amplifies plateau properties in the pyloric neurones
Under the conditions described above, the pyloric constrictor neurones lose their plateau properties. However, they are capable of generating plateau potentials to some extent, in most unmanipulated preparations. In such cases (Fig. 6), activity in APM amplifies rather than induces the plateau properties of these neurones.
In some cases, as in Fig. 6B, the plateau properties of the several PY neurones and developed to differing extents. Here, repetitive depolarizing pulses (100 ms) provoke only partial plateaus in one constrictor neurone (PY1), whose plateau properties are very weak. PY1 is electrically coupled (Fig. 6A) to another PY neurone (PY2) and the pulses of current injected into the soma of PY1 produce depolarizing electrotonic potentials in PY2 (Fig. 6B, open triangles). In this case, the plateau properties are stronger in PY2 than in PY1, and the electrotonic potentials transmitted from PY1 provoke fully developed plateaus in PY2. When APM is induced to fire tonically at a low frequency (Fig. 6C), PY1 produces true plateaus in response to the same pulses, and the plateaus in PY2 are stronger than before APM firing: they last longer, spike frequency within the plateau is elevated, and the amplitude of the plateau potential is increased.
Fig. 6E, F illustrates a situation in which APM amplifies naturally occurring plateaus. In this case, plateaus in a PY neurone were spontaneously produced more or less in synchrony with bursts in the oesophageal network, another rhythmic network of the stomatogastric nervous system (Moulins & Nagy, 1981) whose activity is recorded extracellularly from a son. During a low frequency discharge of APM (Fig. 6F), these plateaus are both longer and of greater amplitude than they are when APM is silent. Thus, APM can either induce or amplify plateau properties in the constrictor motoneurones (PY and LP).
APM modifies the repolarization phase of plateaus in the pyloric neurones
The repolarization phase is slowed down
We have shown that APM can amplify the plateau properties of the constrictor motoneurones and that one manifestation of this amplification is an increase in the duration of plateaus, such that the spontaneous repolarization phase, which usually terminates a plateau potential, occurs later. In addition, when the plateau properties are so amplified, this repolarization phase in constrictor neurones is less easily triggered by inhibitory synaptic inputs from the pacemaker neurones. In some cases, these inhibitory inputs are unable to interrupt a plateau potential when APM is active (see for example Fig. 1C, open triangles). Further, as demonstrated in the preceding paper (Nagy & Dickinson, 1983), APM decreases the efficacy of these same synapses when the pyloric rhythm is active. This suggests that APM may specifically modify the repolarization phase responsible for the termination of the plateaus. Modifications of this repolarization phase of plateaus in the constrictor neurones may be of particular importance functionally, for the duration of each burst of spikes in these neurones depends on the duration of the depolarized phase of the plateaus. We therefore considered the effects of APM’s discharge on the repolarization phase of the pyloric neurones, after triggering these repolarizations by injecting pulses of hyperpolarizing current into the neuronal somata.
Fig. 7 shows that, when APM fires, the repolarization from plateaus in the constrictors is distinctly slowed down. Repolarizations of LP in response to 1 s hyperpolarizing current pulses of increasing intensity injected during spontaneous plateaus are shown in Fig. 7B. In this instance LP’s plateau properties were well developed. The stomatogastric ganglion was bathed in 10−5M-picrotoxin to minimize synaptic interference from neurones within the pyloric network. During an APM discharge, the repolarizations caused by identical hyperpolarizing current pulses are drastically modified (Fig. 7C), the apparent time constant of the repolarizations being increased.
Careful examination, however, indicates that the initial, passive descent is unchanges (arrow), but that the active phase which completes the repolarization is considerably slowed.
That APM slows the active repolarization of the constrictor neurones is confirmed by the experiment illustrated in Fig. 7D, E, F. In this case, a PY neurone had completely lost its plateau properties (synapses blocked in commissural ganglia, Fig. 7D). PY was induced to fire tonically by injection of depolarizing current, and was then injected with pulses of hyperpolarizing current. The purely passive responses to 1 s hyperpolarizations of several intensities (Fig. 7E) can be compared with the responses to identical pulses during plateau potentials in PY induced by APM firing (Fig. 7F). When APM fires, the repolarizations of PY are increased in amplitude, and are slowed down in a manner analogous to that seen in LP in Fig. 7C. The initial rapid descent (Fig. 7F, arrows) corresponds roughly to the passive response in Fig. 7E. The retarded descent (starting at the arrows in Fig. 7F), which increases the amplitude of the response, probably corresponds to the active repolarization phase of a plateau, which is induced and slowed down by APM’s discharge.
Further evidence that APM acts specifically on the active phase of the repolarization from a plateau is provided by the injection of double steps of hyperpolarizing current (Fig. 8). It is known (Russell & Hartline, 1978; Tazaki & Cooke, 1979b) that the repolarization phase of a plateau potential is an active process with an all-or-none character; once set in motion, it continues until repolarization is complete. Thus, if a neurone is subjected to a hyperpolarization in two steps (of sufficient duration) during a plateau, the active response occurs during only one of the two hyperpolarizing steps, the one in which the membrane potential reaches the threshold for active repolarization. Fig. 8 shows that when such two-step hyperpolarizing currents are injected into LP or PY, the retardation of repolarization is likewise seen during only one of the two steps. Fig. 8A-F illustrates two-step hyperpolarizations of LP and PY, respectively, the two current steps in PY being of equal intensity, the second step in LP being stronger. Neither LP nor PY showed plateau properties in the absence of APM firing and responded passively to the double hyperpolarizations (Fig. 8A, D).
Fig. 8B and E shows the responses of the same neurones just after stimulation of the ion axons of APM. In both cases, only the response to the second hyperpolarizing step is slowed down. The experimental conditions in Fig. 8C and F are identical to those in B and E, respectively, except that, in addition to the two hyperpolarizing steps, a constant hyperpolarizing current was injected into each neurone, so that the first step in C and F brought the membrane to the same potential reached during the second step in B and E (compare arrows in B, C and in E, F). In these cases only the first hyperpolarizing response was slowed. Thus, the modifications provoked by APM’s discharge are visible during only one of the two steps, which supports the hypothesis that APM modifies active membrane properties and is able to induce these properties in pyloric neurones which do not spontaneously exhibit them (as is the case in Fig. 8A and D).
The induction of these active properties can also be seen as an increase in the amplitude of the response to a given hyperpolarizing step. In Fig. 8C and F (with APM’s discharge), the responses to the first steps are much larger than those in A and D (control), respectively, whereas the responses to the second hyperpolarizations are nearly equal in amplitude in A and C and in D and F. Similarly, the response to the second hyperpolarizing step in E (with APM’s discharge) is larger than the response to the second step in D (control), whereas the response to the first steps are similar. That the response to the second hyperpolarization in B (with APM’s discharge) is smaller, not larger, than that in A (control) can be explained by the fact that the slowed active phase in B did not. have sufficient time to reach maximum hyperpolarization before the hyperpolarization was terminated and the cell was depolarized.
A final line of evidence that APM slows the active repolarizing phase of plateaus is shown in Fig. 9, in which APM no longer modifies the repolarization when the active phase is experimentally blocked. Studies of the ionic mechanisms producing plateau potentials suggest that the active repolarization is dependent on K+ currents (Tazaki & Cooke, 1979c; Gola & Selverston, 1981). Such K+ currents can be decreased by increasing the external K+ concentration. In such conditions (3 X K+, Fig. 9B), two- step pulses of hyperpolarizing current injected into a PY neurone just after stimulation of APM’s axons in the ions, engender a hyperpolarization whose amplitude is considerably smaller than that produced when the stomatogastric ganglion is bathed in normal saline (31 mV instead of 48 mV, Fig. 9C). In addition, the slowing which normally accompanies the active repolarization induced by APM’s discharge is not present in high external K+ (compare the response to the first hyperpolarizing step in B with that in C).
It is not unlikely that APM also modifies the depolarizing phase of plateau potentials. However, except for the induction and amplification of plateau properties, which involve this phase, it was not studied in detail here.
The effects of APM are voltage-dependent
The two stage hyperpolarizations in Fig. 8 point to another fundamental characteristic of the action of APM: it is voltage-dependent. APM slows down the repolarization of pyloric neurones only within a certain range of voltages, which varies somewhat from neurone to neurone, but is quite stable within a given cell during a single experiment (−55 to −70 mV in LP, −40 to −60 mV in PY in the example of rig. 8). The membrane potential of the constrictor neurones (LP, PY) considered in Fig. 8 is brought into this range by the first hyperpolarizing step in C and F and by the second step in B and E.
The voltage dependence of APM’s effects is further illustrated in Fig. 10A. Currents of increasing intensity injected into the soma of PY, just after stimulation of APM’s axons, first produce passive hyperpolarizations of the PY neurone to different levels (indicated by arrows; −32 mV, upper trace; −39mV, middle trace; −43 mV, lower trace). These passive hyperpolarizations then trigger active repolarizations which are slowed down by the activity of APM. But it is clear that the more negative the membrane potential of the pyloric neurone is when the active response starts (i.e the stronger the initial passive hyperpolarization is), the less this active response is slowed by the influence of APM. Thus, the extent to which an active repolarization of a pyloric neurone is slowed as a result of an APM discharge is voltage-dependent. This characteristic of APM’s effects has important functional consequences for the expression of the pyloric motor pattern (see below).
APM’s effects are long lasting
One of the several striking features of APM’s effects on the pyloric network is its temporal characteristics: the global effects of an APM discharge continue well after the end of that discharge (see Nagy & Dickinson, 1983). Thus, the temporal evolution of APM’s effects on the active repolarization of pyloric neurones is of interest. Such changes with time can be seen in Fig. 10B, during a series of two-step hyperpolarizations of LP after stimulation of APM’s axons in the ions. The slowing of the repolarization, which occurs during the first of the two hyperpolarizing steps in this case, gradually decreases, but is still present 10 s after a 5 s APM stimulation. The persistence of the effects of APM is also shown in Fig. 10C, in which the response of LP to pulses of hyperpolarizing current injected at several times after stimulation of APM’s axons in ions are superimposed. Here again a gradual decrease in the extent to which the active response is slowed can be seen (compare traces at 6, 12 and 18 s after stimulation). It should also be noted that these active responses all start at the same level of membrane potential (arrow).
Functional consequences
The various modifications of membrane properties engendered by APM’s discharge are interesting not only in their own right, but also in terms of their effects on the pyloric rhythm. APM affects all the pyloric neurones (see Nagy & Dickinson, 1983), but its effects are particularly pronounced on the constrictor neurones, LP and PY. We have shown here that APM is capable of inducing and amplifying the plateau properties of the constrictor neurones. Moreover, APM slows the repolarization phase of the plateau potentials, especially in response to weak hyperpolarizing inputs. Because the repolarizations in response to stronger hyperpolarizing inputs are less retarded and those to weaker inputs are more retarded (see Fig. 10A), an ultimate effect of the slowing of the repolarizing phase of plateaus is a decreased sensitivity of the pyloric neurones to weak inhibitions and consequently a tendency to stabilize the neurone at its depolarized level. However, because the overall plateau properties of the pyloric neurone are amplified (or induced in certain cases), the response of the neurone to inputs which are strong enough to provoke the transition from one stable level to the other will be larger in amplitude than it would be in the absence of APM firing. APM is thus ultimately able to decrease the sensitivity of the constrictor neurones to weak inputs and to increase their response to strong inputs.
The cellular modifications provoked by APM enable us to understand the responses of the pyloric network to APM’s discharge not only when the pyloric rhythm is weak, but also when it is active. We showed in the preceding paper (Nagy & Dickinson, 1983) that, when the pyloric rhythm is active, a discharge of APM results in a strong activation of all the neurones in the network, a decrease in the efficacy of the inhibitory synapses of the pacemakers (PD) onto the constrictor neurones (synapses 1 and 2, Fig. 1 IB) and an increase in the efficacy of PY’s inhibitory synapses onto LP (synapse 3, Fig. 11B). There is thus an increase in the amplitude of oscillations in LP and PD and a decrease in the amplitude of oscillations in PY (see Fig. 11C). This would be explained if the bursts in the dilator neurones are insufficient in intensity and/or duration to cause full repolarization in the constrictors, whereas the bursts in the constrictor neurone PY are long and intense enough to repolarize LP, and the inhibitions from LP onto PD are able to provoke the repolarization of the dilator neurone PD. We thus suspect that the modifications of apparent synaptic efficacies which APM provokes are due both to differences in intensity of inhibitions and to modifications of the sensitivity of postsynaptic neurones to such inhibitions. To confirm this, we weakened the reciprocal inhibitory synapses of the constrictor neurones (synapses 3 and 4, Fig. 11B) and the inhibitory synapse of LP onto PD (synapse 5, Fig. 11B) with picrotoxin (Bidaut, 1980). After an APM discharge in these conditions (Fig. 11D), LP behaves like PY, with bursts of long duration and decreased amplitude, the inhibition due to PY no longer being sufficient to bring LI to its lower level of stable membrane potential. Similarly the bursts in the pacemaker PD tend to be longer, suggesting that APM exerts effects on PD that are similar to, although weaker than, those it exerts on the constrictor neurones. In picrotoxin, the inhibitions from LP are no longer sufficient to repolarize PD after an APM discharge.
Thus, it is largely through the intermediary of the induction, amplification and modification of the plateau properties of the pyloric neurones that the interneurone APM is able to activate the pyloric rhythm and to modulate the expression of the motor pattern produced by the pyloric network.
DISCUSSION
APM induces and amplifies the bursting ability of the pyloric neurones
Perhaps the most notable characteristic of the interneurone APM is the fact that it acts on the plateau properties of the pyloric neurones. These properties play an important role in the generation of the pyloric rhythm by allowing the pyloric neurones to develop the depolarizations necessary for the production of bursts of action potentials. They are therefore responsible for the ‘burstiness’ of the neurones which make up the pyloric network (Russell & Hartline, 1978). Plateau properties are characteristic of both the pacemaker (dilator) and the follower (constrictor) neurones (Russell & Hartline, 1978, 1981) of the pyloric network. The major feature distinguishing these two categories of neurone is the ability of the pacemakers to develop ‘pacemaker potentials’ (Gola & Selverston, 1981), the spontaneous slow depolarizations which bring the membrane potential to the threshold at which plateau potentials are triggered. Thus, the dilator neurones spontaneously and rhythmically produce plateaus and function as endogenous oscillators. In contrast, plateau potentials in the constrictor neurones must be triggered by synaptic inputs. An analogous situation in the cardiac ganglion of the crab, Portunus, has been studied in considerable detail (Tazaki & Cooke, 1979 a,b,c). In fact, plateau-like electrical activity appears to characterize a number of types of excitable cells, including both nerve cells (Ekerot & Oscarsson, 1981; Legendre, Cooke & Vincent, 1981) and non-nervous cells such as the cells of the endocrine pancreas (Cook, Crill & Porte, 1980) and cardiac muscle fibres in vertebrates (Vassale, 1979).
The expression of the plateau properties of the pyloric neurones depends on inputs of central origin. If spike conduction in the afferent nerve to the stomatogastric ganglion (the stomatogastric nerve) is blocked, the plateau properties of the constrictor neurones vanish (Russell & Hartline, 1978) and the plateau properties of the pacemaker neurones are generally weakened (Russell, 1979) or abolished (Moulins & Cournil, 1982). Electrical stimulation of the stomatogastric nerve can temporarily restore the plateau properties in these neurones; however, the nature of the inputs inducing them in the pyloric neurones, and particularly in the constrictor (nonoscillating) neurones, was heretofore unknown. We have shown that activity (even at moderate frequency) in the neurone APM can induce plateau properties in the nonpacemaker neurones of the network. Thus, a discharge of APM alone can confer on the non-pacemaker neurones of the pyloric network the ability to fire in bursts of action potentials, a mechanism which plays a fundamental role in the control of the motor pattern’s generation. This phenomenon, the induction of non-linear membrane properties that can alter the bursting capabilities of neurones, is not a well known phenomenon, and has previously been demonstrated only for neurone 11 of the visceral ganglion of the snail Otala. When this animal aestivates, neurone 11 loses its ability to fire in bursts; this ability can be reinduced by perfusion or iontophoresis with vasopressin or related peptides (Barker, Ifshin & Gainer, 1975; Ifshin, Gainer & Barker, 1975). The induction of plateau properties in the pyloric neurones by the activity of APM can likewise be compared to that of the electrical activity of the β cells of the endocrine pancreas, which can produce plateau potentials only in the presence of glucose, the duration of the plateaus being a function of the extracellular glucose concentration (Meissner, 1976).
A second aspect of APM’s actions is to amplify these properties when they are already present and active. This is reflected in the longer plateaus and the higher spike frequency within plateaus recorded after an APM discharge. The prolongation of the plateaus is largely due to the fact that APM slows down their repolarization phase. Although we tested the ability to produce plateau potentials under the influence of activity in APM only for the constrictor (non-pacemaker) neurones, indirect observations suggest that the interneurone APM is also able to augment the pacemaker neurones’ capacities for rhythmic activity. Thus an APM discharge provokes an increase both in amplitude of oscillation and in spike frequency within bursts in these neurones when the pyloric rhythm is normally active, (Nagy & Dickinson, 1983) and it provokes a ten-fold increase in plateau duration when the pacemaker neurones are isolated from other synaptic influences (present paper). It has recently been shown that an amplification of plateau properties specifically in the pacemaker neurones can also be induced by the electrical stimulation of a pair of nerve fibres afferent to the stomatogastric ganglion (Russell & Hartline, 1981). The amplification, under the influence of APM, of the rhythmic firing capabilities which are spontaneously manifested by all the pyloric neurones bears similarities to the control of endogenous rhythmic activity which has been examined in several preparations, including the cardiac ganglion of crustaceans (Watanabe, Obara & Akiyama, 1969; Cooke & Hartline, 1975; Lemos & Berlind, 1981) and neurone R15 of the parieto-visceral ganglion of Aplysia (Parnas, Armstrong & Strumwasser, 1974; Barker et al. 1975; Mayeri, Brownell, Branton & Simon, 1979). The control of the amplitude and duration of the vertebrate cardiac action potential (and particularly of its plateau phase) by various substances implicated in the nervous control of cardiac activity may also be comparable (Vassort et al. 1969; Reuter, 1974; Giles & Nobel, 1976; Ten Eick, Nawrath, McDonald & Trautwein, 1976).
Consequently, it appears that the types of effect induced by the interneurone APM are not restricted to the stomatogastric nervous system, but are instead fundamental to the control of several excitable systems. However, the system described here has a number of advantages as a model system for the study of this sort of control. Firstly, the modifications observed are due to the activity of a single neurone which can be found and identified from animal to animal. Secondly, it is possible to attribute a functional significance to the effects of APM: these effects allow the neurone to modulate both quantitatively and qualitatively the expression of a motor pattern generator (Nagy & Dickinson, 1983). In fact, these various modifications of the pyloric pattern can all be explained by the modifications of the plateau properties the neurones of the pyloric network.
Modifications of the plateau properties of the pyloric neurones underly modulation of the pyloric output by the interneurone APM
The modulation of the pyloric activity is seen in three principal effects: (1) an activation of previously silent neurones and an increase in the firing of active neurones; (2) an increase in the frequency of the pyloric rhythm to a maximum of around 1 Hz; (3) a modification of the efficacy of synaptic relationships within the network (Nagy & Dickinson, 1983).
The first effect is easily explained by the fact that the plateau properties form the basis for the bursting activity of the pyloric neurones. Thus the induction or amplification of these properties by activity in APM can either permit or augment the bursting activity of these neurones.
The second effect of an APM discharge is more complex. The pyloric network receives rhythmical excitation from a pair of command oscillators (the commissural pyloric oscillators, CPO) which have a rapid (approximately 1 Hz) and constant rhythm in in vitro preparations (Robertson & Moulins, 1981). These oscillators are able to entrain the rhythm of the pyloric pacemaker neurones (Nagy, 1981), although the coupling of the CPO to the pyloric oscillators is labile (Robertson & Moulins, 1981). It is this lability that is responsible for the existence of two distinct states of the pyloric rhythm, both in intact animals (Rezer & Moulins, 1980) and in isolated preparations (Nagy, 1981). A rapid and regular rhythm in which all the pyloric neurones participate corresponds to a tight coupling of the pyloric neurones and the CPO, whereas a slow and irregular rhythm appears to result from looser coupling. As was shown in the preceding paper (Nagy & Dickinson, 1983), an APM discharge can provoke an acceleration of the pyloric rhythm, the extent of which is dependent on the rhythm. In all cases, the maximum frequency attained is about 1 Hz (the frequency of the CPO); if the pyloric rhythm already has this frequency, activity in APM does not cause it to accelerate. In addition, an acceleration of the pyloric rhythm is always accompanied by an activation of all the formerly quiet neurones of the network. A reasonable explanation for these observations is that the modification of membrane properties by an APM discharge renders all of the pyloric neurones more sensitive to the rhythmic inputs from the CPO. In other words, the modulatory interneurone APM favours the coupling of the pyloric neurones to the command oscillators. This explanation is corroborated by the following observation: when synaptic activity in the commissural ganglia is experimentally blocked, the pyloric rhythm slows considerably or stops altogether (present paper), probably as a result of the suppression of the rhythmic inputs from the CPO, as well as other inputs. Under these conditions, an APM discharge still profoundly modifies the activity of pyloric neurones, but it never leads to the initiation or acceleration of a rhythmic activity. In the intact animal, the transition from a slow and irregular to a fast and regular pyloric rhythm occurs abruptly when the animal eats, and the rapid rhythm continues for up to several hours after the meal (Rezer & Moulins, 1980). These authors attribute this transition to an increased coupling of the CPO and the pyloric rhythm upon feeding. The characteristics of the action of the interneurone APM suggest that it could play a role in such a transition. The experimental model that we have presented thus suggests that the central nervous system can initiate a behavioural sequence simply by modifying certain membrane properties of the motoneurones in such a way that the motoneurones become more sensitive to inputs from a command oscillator.
The third aspect of the modulation of pyloric activity by the activity of the inter-neurone APM is the modification of the relative efficacies of synapses within the pyloric network. These modifications qualitatively change the pyloric rhythm by altering the phase relationships between the discharges of the different neurones. These modifications are due both to changes in the intensities of the inhibitions received by the neurones and to changes in the sensitivity of the postsynaptic neurones to the inhibitions resulting from the activity of APM. These effects can be explained by the fact that an APM discharge modifies the temporal characteristics of the repolarization phase of the plateau potential, which is normally triggered by inhibitory synaptic inputs in these neurones. The repolarization is retarded and this retardation is voltage dependent. Thus, as the membrane potential is taken passively by the input to more negative levels (corresponding to stronger inhibitory inputs), the retardation is less pronounced. Consequently, strong inputs will repolarize the membrane rapidly whereas weak inputs will do so only slowly. Thus, even a brief stimulus of sufficient intensity will repolarize the membrane. However, for a less intense input to repolarize the membrane, it must be of longer duration. One effect of the active repolarization is to amplify the inhibitory inputs a neurone receives; thus the effects of the largest inhibition (in either intensity or duration) received by a neurone will be those that are amplified when APM fires. Consequently, an APM discharge leads to an increase in the contrast of synaptic effects by augmenting the response to strong inputs and decreasing the response to weak inputs. It has previously been suggested that plateau properties can act as the basis for a synaptic amplifying mechanism (Russell & Hartline, 1978). The interneurone APM, by modifying these properties, provides the pyloric neurones with a means of filtering these synaptic inputs by specifically amplifying only certain of the inputs they receive.
ACKNOWLEDGEMENTS
We are grateful to Professor M. Moulins for helpful criticisms of the manuscript. This work was supported by the DGRST (Grants 80 P 6049 and 81 E0545) and by a CNRS/NSF exchange fellowship.