Properties of synapses made by IVN command-interneurones in the stomatogastric ganglion of the spiny lobster Panulirus interruptus

Two nerve cells that project from the brain to the stomatogastric ganglion of Panuliris interruptus are now known to be command neurones (Sigvardt & Mulloney, 1981). When these two neurones fire prolonged bursts of impulses, in response either to sensory input or to experimental stimulation, the motor patterns that normally occur in the stomatogastric ganglion change radically (Dando & Selverston, 1972; Sigvardt & Mulloney, 1981). The neurones of the stomatogastric ganglion and their synaptic organization are known in detail (Maynard & Selverston, 1975; Mulloney & Selverston, 1974a, b;Mulloney, 1977; Selverston & Mulloney, 1974). A description of the synapses in the ganglion made by these command interneurones would make possible an explanation of the changes in the motor pattern that their firing causes.

This paper describes the p.s.p.s caused by the two IVN interneurones in all the stomatogastric neurones with which they synapse. The dynamics of some of these synapses are complex, and contribute significantly to the changes that these interneurones command.

Our methods and the anatomy of the stomatogastric nervous system were described in detail in the companion paper (Sigvardt & Mulloney, 1981). In the experiments reported here, the preparation was isolated completely from the animal and the stomach and consisted of the stomatogastric and oesophageal ganglia connected by the stomatogastric nerve (SGN). The preparation was pinned under saline solution in a petri dish lined with a transparent, inert silicone rubber (Sylgard). Extracellular recordings and stimulations were made with pin electrodes attached to selected peripheral nerves (Mulloney & Selverston, 1974 a).

The two command interneurones we studied have axons in the Inferior Ventricular nerve (IVN) (Dando & Selverston, 1972), and only the axons of these two neurones run from the IVN to the stomatogastric ganglion through the SGN (Kushner, 1979; Sigvardt & Mulloney, ibid.). Therefore we could stimulate them at the IVN and record their impulses en passant in the SGN before the impulses caused p.s.p.s in the neurones of the stomatogastric ganglion (Sigvardt & Mulloney, ibid.). Data were recorded directly by photographing an oscilloscope trace, or by using a Gould 220 chart recorder. The results of this paper were obtained from 20 Expts.

To record p.s.p.s at different membrane potentials in the postsynaptic neurone, we penetrated the soma of each neurone with two microelectrodes, filled either with 4 M-K acetate or 2·5 M-KCI. We used one electrode to record potential through a Getting M4 or M5 preamplifier and the other to pass the constant currents needed to move the steady-state membrane potential of the neurone to the desired value. These potentials were calibrated with a WP-Instruments Inc. Precision Millivolt Source.

The saline solutions used in these experiments contained in mMoles:

The control saline had 5 × normal Ca²⁺, and was used to increase the size of the p.s.p. and to reduce spontaneous firing in the postsynaptic neurones. D-tubocurare and picrotoxin were dissolved in control saline. In both control and high Mg²+, we reduced the NaCl to compensate for the additional Cl^- added as CaCl₂ or MgCl₂; the Cl^- concentration was the same within 2% in the four solutions. To compensate osmotically for the loss of the Na+ in these solutions, we added sucrose to raise the osmolarity to within 1 % of that of the normal saline, assuming that all the salts dissociated completely.

PD, pyloric dilator neurone; VD, ventricular dilator neurone; Int 1, interneurone 1; LPGN, lateral posterior gastric neurone; GM, gastric mill neurone; IVN, inferior ventricular nerve; SON, superior oesophageal nerve; SGN, stomatogastric nerve; StG, stomatogastric ganglion.

The two IVN interneurones synapse with 11 neurones in the stomatogastric ganglion : the three PD-AB neurones and the VD neurone of the pyloric system and Int i, the two LPGNs and the four GMs of the gastric system. These 11 synapses appear to be direct by two criteria. First, the individual p.s.p.s follow each action potential in the IVN interneurones at a constant latency (Fig. 7 of Sigvardt & Mulloney, 1981) and the latency of the p.s.p.s in all postsynaptic neurones is the same in any given experiment. Second, the p.s.p.s recorded in each neurone follow impulses in the IVN interneurones even at frequencies greater than 50 Hz, and they continue to do so when the extracellular Ca²⁺ concentration is raised fivefold. If an interneurone were interposed between the IVN units and these 11 postsynaptic neurones, raising the Ca²⁺ concentration would raise this interneurone’s threshold and would increase the probability that p.s.p.s would fail to follow IVN impulses (Kehoe, 19726). We observed no failures, and conclude these synapses are direct.

The thresholds of these two interneurones to stimulation of the IVN are very similar, and often they can not be stimulated individually. In experiments where the thresholds of the two axons were sufficiently distinct to permit us to study the consequences of stimulating only the lower-threshold unit, we looked for differences in the connexions made by the two IVN units. We did not discover any differences; increasing the stimulus intensity above threshold for the second interneurone simply recruited a second p.s.p. of identical latency and similar other properties in every postsynaptic neurone we examined. In the rest of this paper, we will treat the two as having the same properties.

The IVN interneurones have very different effects on the output of each of the postsynaptic neurones in the stomatogastric ganglion because the properties of each synapse differ. When the IVN is stimulated at low frequencies in normal saline, the p.s.p.s in all neurones except VD are small (Fig. 1). During experiments to characterise the properties of each synapse, we increased the size of the p.s.p.s by perfusing the ganglion with control saline (5 x normal Ca²⁺; see Methods).

Neurons of the PD-AB group have a biphasic p.s.p. following each impulse in the IVN interneurones (Fig. 2). This p.s.p. consists of an initial rapid, excitatory depolarization followed by a slower, long-lasting inhibitory hyperpolarization. The PD-AB neurones are endogenous bursters (Gola & Selverston, 1981) whose membrane potentials oscillate periodically through a range as large as 20 mV, so the effect of these IVN p.s.p.s depends in part on the potential of the postsynaptic membrane when they arrive and in part on the frequency at which they arrive. At low presynaptic impulse frequencies, the p.s.p.s trigger impulses in the PD-AB neurones during the depolarized parts of the endogenous oscillations, but not during the hyperpolarized parts. These low frequency p.s.p.s have little effect on the endogenous pattern of bursting in PD-AB. When the IVN interneurones fire at higher frequencies, however, the long-lasting inhibitory components of the p.s.p.s sum and firing stops in PD-AB (Figs. 1, 2). After a bout of high-frequency impulses in the presynaptic fibres, PD-AB fire vigorous bursts that return to normal only after several seconds (Fig. 1).

During a 25 Hz train of impulses in the IVN interneurones, the excitatory and inhibitory components of the p.s.p.s interact. For example, the first three e.p.s.p.s in one 25 Hz train recorded in a PD neurone starting at – 56 mV had amplitudes of 5, 3 and i mV, even though the membrane potential was moving farther from their reversal potential (Fig. 4). In contrast, the e.p.s.p.s from a 2·5 Hz train recorded at the same initial membrane potential (Fig. 3, control –56 mV) are almost identical. This reduction in the size of the excitatory component is not due to classical synaptic depression because the slower inhibitory component does not decrease proportionately. We think the reduction of the excitatory component occurs because the excitatory current is effectively shunted by the summed inhibitory conductance (Takeuchi, 1977). The net effect of the two components during a high frequency train of p.s.p.s is to excite the PD-AB neurones transiently, and then to inhibit them for the duration in the train (Fig. 1).

VD normally fires bursts of impulses that alternate with the bursts in the PD-AB neurones because PD-AB inhibit VD effectively (Maynard & Selverston, 1975). Impulses in the IVN interneurones cause large e.p.s.p.s in VD (Figs, i, 2); each of these p.s.p.s usually causes an impulse in VD throughout the normal range of IVN impulse frequencies (Sigvardt & Mulloney, 1981). When the IVN interneurones fire at low frequencies, additional impulses are simply intercalated in VD’s characteristic train of bursts (Fig. 1). When the IVN interneurones fire at high frequencies, bursting is suppressed in the PD-AB neurones and VD then fires a train of impulses each of which is triggered by a p.s.p. from the IVN interneurones (Fig. 1). In high Ca²⁺ saline, VD may fail to follow the IVN interneurones impulse-for impulse (Fig. 2), but in normal saline VD continues to fire during a long train of IVN p.s.p.s.

The postsynaptic response of LPGN appears to be inhibitory when the amplitude of the p.s.p. is increased by increasing extracellular calcium (Fig. 2). The i.p.s.p. is long-lasting, and stimulation of IVN at 25 Hz produces a rapid summation to a potential that is probably the reversal potential for the p.s.p. There may be an initial depolarizing component in normal saline (Fig. 1 ), but this component is overwhelmed by the inhibitory component in high-calcium saline and during high-frequency stimulation in normal saline. Since the excitatory component is so small, the biphasic p.s.p. can be described functionally as an i.p.s.p. The IVN to LPGN synapse facilitates; at 2·5 Hz the i.p.s.p. is about 0·5 mV (Fig. 1) but at 25 Hz the third stimulus produces a 2·0 mV i.p.s.p. and the response increases to a maximum of about 4 mV. Its facilitation characteristics are such that it usually takes several stimuli to prevent firing of LPGN when LPGN is firing ionically.

The GM neurone receives inhibitory input from the IVN interneurone (Figs. 1, 2). Individual i.p.s.p.s are small and therefore have no effect on the cell’s firing pattern when they occur at low frequencies. The timecourse of the i.p.s.p. is slow, and summation to a subthreshold level is rapid at high frequency (Figs. 1, 2). This cell shows little or no postinhibitory rebound following IVN input.

Interneurone 1 (Int 1) is a critical component in the network that generates the gastric rhythm (Mulloney & Selverston, 1974b). The IVN interneurones synapse directly with Int 1 (Fig. 7 of Sigvardt & Mulloney, 1981). When the IVN interneurones fire at low frequencies, the postsynaptic response is excitatory; additional impulses are intercalated in the ongoing train of Int 1 impulses whenever the IVN interneurones trigger a p.s.p. At higher frequencies, a long-lasting inhibitory component becomes apparent, probably because of temporal summation during the high-frequency train. This inhibitory component effectively prevents firing in Int 1 until the IVN interneurones stop and Int 1 recovers. This inhibition of Int 1 can also be inferred from the changing frequency and amplitude of the i.p.s.p.s caused by Int 1 in GM (Fig. 2).

Biphasic potentials like these in PD must arise from the integration of more than one synaptic current. In the PD-AB neurones, the excitatory and inhibitory currents overlap in time, and we do not have pharmacological methods to isolate them. Therefore, to study the contributions of different ions to the inhibitory current, we measured the size of the p.s.p. under conditions that held the excitatory current constant. Then, although each p.s.p. had a contribution from the excitatory current, differences between experimental and control p.s.p.s recorded at the same holding potendial Could measure changes in the contributions of the inhibitory currents themselves. The time-to-peak of the hyperpolarizing part of the p.s.p. was not constant under different experimental conditions when the IVN interneurones were stimulated at 2·5 Hz (Fig. 3) so we measured the size of the summed p.s.p. immediately following a 2 s, 25 Hz train of stimuli to IVN (Fig. 4). This summed compound p.s.p. had an excitatory component, but this component was comparatively constant, and so could be accounted for in the analysis of i.p.s.p.s.

If K+ is the charge carrier for the inhibitory current, a two-fold change in the external concentration of K+ should change the size of the i.p.s.p.s measured at a controlled starting voltage by 17 mV. We measured the size of the summed p.s.p. following a 25 Hz train of IVN stimuli (arrows in Fig. 4), and plotted these measurements as a function of the holding potential at which each p.s.p. was measured (Fig. 5). The reversal potential for the control p.s.p.s was – 68·3 mV, and the reversal potential changed to – 52·6 mV when external K⁺ increased twofold. The difference between the predicted 17 mV and measured 15 mV is within the limits of accuracy of our recording apparatus. In these neurones, a pure K⁺ current in normal saline should have a reversal potential of – 80 mV (Marder & Paupardin-Tritsch, 1978 a), so the difference between this and the measured reversal potential reflects the contribution of other currents.

If Cl⁻ also acts as a charge-carrier for the inhibitory current, then altering the external or internal Cl⁻ concentration should also change the size of the i.p.s.p. These experiments are complicated by the presence in these neurones of a significant non-synaptic resting Cl⁻ conductance (Marder & Paupardin-Tritsch, 1978a) that will allow Cl⁻ to redistribute across the membrane rapidly. Therefore we injected Cl⁻ through a microelectrode into an individual neurone by prolonged hyperpolarization (Kehoe, 1972a), and measured the size of the summed p.s.p. at the end of a 2 s, 25 Hz, train of IVN stimuli (Fig. 4), immediately after the injection was complete. It was not possible to measure the amount of Cl⁻ injected into the neurone, and so we can not predict the corresponding change in the reversal potential of a purely Cl⁻ carried i.p.s.p., but Figs. 4 and 5 show that increasing the internal Cl⁻ concentration changed the size and reversal potential of this summed, compound p.s.p. We conclude that Cl⁻ also contributes to the inhibitory current at this synapse. The agreement between the measured and predicted change in reversal potential when K+ concentration changed two-fold does not mean that only K+ acts as a charge carrier because Cl⁻ could redistribute across the membrane during the 15 min intervals between the change in K+ concentration and stimulation of the IVN interneurones.

The excitatory component of this p.s.p. increased in size as the neurone was hyperpolarized. It did not reverse in the range of potentials below threshold, and it did not appear to be the product of an electrical synapse because it was blocked reversibly by a low Ca²⁺, high Mg²+ external medium (Fig. 6). We suggest that this component is an inward Na⁺ current, but we have not explored this question further.

The synapse of the IVN interneurones with VD appears to be purely excitatory. If VD is hyperpolarized, the IVN p.s.p. fails to evoke an impulse in VD, and the underlying e.p.s.p. is revealed (Fig. 7). The p.s.p. showed some decrement with repeated stimulation but reached a steady-state amplitude after new stimuli. This e.p.s.p. is probably due to an increase in Na⁺ conductance, because the e.p.s.p. increased in amplitude as the membrane was hyperpolarized, never reversed and was unaffected by increases in external potassium ion concentration (Fig. 7) or internal Cl⁻ injection. However, we did not test the Na⁺ hypothesis directly.

The identity of the transmitter released by the IVN command interneurones is not known, but the pharmacology of receptors for various possible transmitters has been described for different neurones of the StG (Marder & Paupardin-Tritsch, 1978a, b). Therefore, we used two different pharmacological agents in an attempt to separate the different ionic currents on the p.s.p.s in PD and VD.

Picrotoxin, 10⁻⁴ M, in control saline had no effect on transmission at these synapses (Fig. 8). This concentration is known to block transmission at certain other synapses in the ganglion (Bidaut, 1980). 5 × 10⁻⁵ M D-Tubocurare (D-TC) did not affect transmission, but 5 ×10⁻⁴ M blocked the inhibitory component completely and reduced the excitatory component (Figs. 8, 9). The reduction of the excitatory component accurred in both PD and VD (Fig. 9).

The effect of the biphasic p.s.p. in PD depends on the burst structure of the IVN interneurones. When the IVN units fire at a low-frequency, the p.s.p. is excitatory and can cause the cell to fire if its membrane potential is near threshold. At the beginning of a high-frequency IVN burst, the excitatory components of the IVN p.s.p.s can cause the cell to fire. After the first few p.s.p.s the inhibitory components sum to short-circuit the excitatory components and prevent the cell from firing (i.e. the postsynaptic response effectively changes sign).

Another synapse with a frequency-dependence similar to that of the IVN-PD synapse is the synapse from L10 to L7 in Aplysia (Wachtel & Kandel, 1967, 1971). L10 produces an e.p.s.p. in L7 at low frequency that converts to an i.p.s.p. at higher frequencies. However their results suggest a different mechanism for the conversion than the one we propose for the IVN-PD synapse; at the L10-L7 synapse the excitatory receptor desensitizes while the inhibitory receptor becomes sensitized to the transmitter. The biphasic p.s.p. described by Berry & Cottrell (1975) is always functionally inhibitory because the excitatory component is initially small compared to the inhibitory component and the inhibitory component facilitates.

Several biphasic or dual-action synapses have been described in molluscan nervous systems (Berry & Cottrell, 1975; Gardner & Kandel, 1972; Kehoe, 1969, 1972a, b;Levitan & Tauc, 1975; Wachtel & Kandel, 1967, 1971). They are interesting because, following the release of a single transmitter, the postsynaptic cell has a multicomponent response that results from the activation of more than one type of receptor in the postsynaptic membrane. In most of these cases, the biphasic potential consists of a depolarizing phase and a hyperpolarizing phase, although the synaptic potential described by Kehoe (1969, 1972a) has two inhibitory components - a rapid i.p.s.p. superimposed on a slow, long-lasting hyperpolarizing wave. The postsynaptic receptors that underlie these dual-action synapses vary in ways that include the associated conductance change, threshold for activation, kinetics, and facilitation or desensitization properties. As a result, the biphasic potential may be very sensitive to the pattern of firing of the presynaptic interneurone; its amplitude and effect depend on the temporal characteristics of the burst. These biphasic potentials thus permit more integrative complexity than conventional synaptic potentials that may vary in amplitude but not in sign.

The two IVN interneurones are command neurones that originate centrally and provide a pathway for specific alteration of the motor output of the ganglion. A burst in the IVN interneurones is initiated by various types of sensory input including input from mechanoreceptors in the pyloric region of the stomach (Sigvardt & Mulloney, 1981). In isolated preparations, where no sensory fields remain attached, these fibres will often fire prolonged bursts spontaneously (Selverston-et al. 1976). In both situations, the bursts have similar structure; they start with impulses occurring at about BHZ, rapidly accelerate to over 50 Hz and then slow to a stable plateau of about 10 Hz (Sigvardt & Mulloney, ibid., Fig. 5). Bursts may last as long as 20 s. Because of this| variation in frequency during a burst, the dynamic properties of the different synapses are significant determinants of the responses of the postsynaptic neurones during and after normal IVN bursts. For example, low-frequency impulses in the IVN interneurones will each trigger impulses in VD but allow PD to burst without interruption, and will not affect the ongoing activity of the neurones of the gastric system. At the other end of the scale, a sudden high-frequency burst in the IVN interneurones will drive a high-frequency burst in VD but inhibit completely PD, GM, LPGN and Int 1.

The IVN interneurones are a convenient experimental pathway that can be used to perturb the ganglion’s spontaneous activity in a controlled way. Ayers & Selverston (1977, 1979) stimulated the IVNs at frequencies ranging from 0·59 to 1·52 Hz to produce e.p.s.p.s in PD at periods just shorter than the inherent period of PD bursts, in an attempt to demonstrate the cellular mechanism for the entrainment of coupled oscillators described by von Holst (1939). In the stimulus regime used by Ayers & Selverston, the excitatory component of the biphasic IVN p.s.p. in PD predominates (cf Fig. 1) and PD oscillations can be entrained by these cyclic excitatory inputs. Strong excitatory input to any oscillatory system will entrain that system in a manner that is largely a function of the difference between the period of the input and the period of the free-running oscillator (e.g. Stein, 1976). However, the spontaneous and sensory-initiated bursts have structures very different from the experimental trains used in these entrainment experiments.

It is probable that the inhibitory component of the biphasic p.s.p. in PD is the result of an increase in conductance to both K+ and Cl⁻ because manipulations of the concentrations of each had a large effect on the reversal potential of this component. Both increasing extracellular K+ and increasing intracellular Cl⁻ shifted the reversal potential in the depolarizing direction, as is predicted from the Nernst relation if each ion can act as a charge-carrier here.

Although the results support qualitatively the hypothesis of an increased conductance to both ions, it is not possible to consider the results quantitatively because of the high resting Cl⁻ conductance of stomatogastric neurones. Any manipulation of the intracellular or extracellular concentration of one of the ions simultaneously affects the resting membrane potential and the equilibrium potentials of K⁺ and Cl⁻. The dual-component synapse described by Kehoe (1969, 1972 a) is a result of an increase in both Cl⁻ and K+ conductance, but the Cl⁻ mediated and K+ mediated phases have different latencies and different durations, and so cause a p.s.p. that has two inhibitory phases rather than a single inhibitory phase. The other dual excitatory-inhibitory p.s.p.s whose ionic mechanisms have been described (Wachtel & Kandel, 1971 ; Gardner & Kandel, 1972) have an excitatory component that is Na⁺ dependent and an inhibitory component that is Cl⁻ dependent.

Marder & Paupardin-Tritsch (1978 a) have described the pharmacological properties of the responses of some of the stomatogastric neurones to various transmitit substances. PD shows K+ dependent and Cl⁻ dependent inhibitory responses to both glutamate and GABA. However, all of these responses except the GABA Cl⁻ response are blocked by picrotoxin, which did not block the response of any StG neurone to IVN interneurones. Therefore, it is unlikely that the IVN interneurones release GABA or glutamate. Acetylcholine causes a depolarizing response in PD and VD that is blocked by D-tubocurare. Although we were unable to block completely the excitatory component of the biphasic p.s.p. in PD, Marder reports that the IVN e.p.s.p. in PD is blocked by curare and in our study the IVN e.p.s.p. in VD was partially blocked by curare.

Marder & Paupardin-Tritsch (19786) also discovered a biphasic response to acetyl-β-methylcholine that has an early excitatory component and a longer-lasting, K+ dependent inhibitory component. Acetylcholine is the putative transmitter for several of the stomatogastric neurones including PD and VD, and all of these neurones cause i.p.s.p.s in other stomatogastric neurones (Marder, 1974, 1976). Thus, although acetylcholine remains a candidate for the transmitter released by the IVN interneurone, the identity of the transmitter is still unknown.

We thank Hilary Anderson, Duncan Byers, Donald H. Edwards, Jr, Gad Geiger, Richard Nassel, Dorothy H. Paul and Kate Skinner for criticizing drafts of this paper. This research was supported by USPHS Grant NS 12295, by a USPHS Postdoctoral Fellowship to K.A.S. and by the Alfred P. Sloan Foundation.