Pharmacological and Electrophysiological Characterization of a Postsynaptic Muscarinic Receptor in the Central Nervous System of the Cockroach

The neurotransmitter acetylcholine (ACh) is widely distributed throughout the central nervous system (CNS) of insects. Extensive biochemical studies have demonstrated that ACh is present locally at high concentrations and have detected high activities of cholinergic enzymatic markers (Sattelle and Breer, 1990).

Two major types of ACh receptors have been characterized in insects: nicotinic acetylcholine receptors (nAChrs) and muscarinic acetylcholine receptors (mAChrs). Numerous studies have revealed the pharmacological properties of nAChrs on synapse-free cell bodies of neurones as well as on the synapse–neuropile complex (Sattelle, 1985; Breer and Sattelle, 1987; Benson, 1992, 1993) and have demonstrated their physiological role in fast excitatory synaptic transmission (Callec, 1985).

Although the vast majority of cholinergic receptors in insects exhibit nicotinic properties, conventional binding studies using muscarinic antagonists, such as quinuclidinyl benzilate (Dudai and Ben-Barak, 1977; Breer, 1981; Aguilar and Lunt, 1984; Lummis and Sattelle, 1985; Huang and Knowles, 1990b; Abdallah et al. 1991; Orr et al. 1991) and scopolamine (Lummis and Sattelle, 1986), have revealed a smaller population of mAChrs. The pharmacological characterization of mAChrs in some insect brains has been carried out using mammalian subtype-selective antagonists (e.g. Knipper and Breer, 1988; Orr et al. 1991; Abdallah et al. 1991). Muscarinic binding sites are found to be localized on cell bodies (Knipper and Breer, 1988) and in precise regions of the brain and thoracic neuropiles (Lummis and Sattelle, 1986; Knipper and Breer, 1988; Orr et al. 1991), suggesting a role for mAChrs in synaptic transmission.

In vertebrates, mAChrs typically modulate neuronal activity by affecting various intracellular transduction mechanisms, which trigger changes in the opening and closing of ionic channels (Kuba et al. 1989; Akaike, 1992). Some of these electrophysiological effects have been described recently in different insect nerve cell bodies. On the soma of the cockroach motoneurone D_f (David and Pitman, 1990) or in dissociated neuronal somata of the locust (Benson and Neumann, 1987; Benson, 1992), a muscarinic inward current can be evoked and inhibited by muscarinic agonists and antagonists, respectively. The muscarinic response can be complex, as described in isolated dorsal unpaired median (DUM) neurones, where a biphasic response composed of a fast initial hyperpolarization followed by a slow depolarization is evoked by the muscarinic agonist McN-A-343 (Lapied et al. 1992). The two components of this response are supported by two distinct mAChrs. To date, no physiological role has been attributed to mAChrs on insect neuronal cell bodies.

For mAChrs located in the neuropile of ganglia, two different physiological actions have been described. Using locust synaptosomes, Breer and Knipper (1984) have demonstrated that the ACh release is decreased by oxotremorine and enhanced by atropine, suggesting a negative feedback regulation involving muscarinic autoreceptors. This modulatory action of mAChrs has also been demonstrated electrophysiologically in the cercal afferent to giant interneurone synapses of the cockroach (Hue et al. 1989; Le Corronc et al. 1991) and in the tobacco hornworm (Trimmer and Weeks, 1989a). At the postsynaptic site, ionophoretic applications of muscarinic agonists within the dendritic tree of the identified motoneurone PPR in the tobacco hornworm induce a small, long-lasting depolarization and modify the effectiveness of the incoming signals by lowering the spike threshold (Trimmer and Weeks, 1989a).

In the sixth abdominal ganglion of the cockroach, mAChrs have been found on the somata of DUM neurones and on the afferents presynaptic to the cercal to giant interneurone synapses, but no demonstration of a precise postsynaptic localization of mAChrs has been reported. The purpose of the present study was to characterize mAChrs in this synaptic system at the postsynaptic site (i.e. on a ventral giant interneurone). We show that these receptors change the postsynaptic spike threshold and can play a functional role under the experimental conditions applied in this study.

All experiments were performed at room temperature (20–23°C) on adult male cockroaches (Periplaneta americana L.) reared at 26°C in the laboratory. The cockroaches were dissected dorsally and the abdominal nerve cord, a cercus and the corresponding cercal nerve were isolated. The sixth abdominal (A6) ganglion, containing the synapses of interest, was carefully desheathed to facilitate penetration of the drug-ejecting micropipette and the application of drugs to the nervous tissue. Using fine stainless-steel needles, a single giant interneurone (GI) from the ventral group, usually GI2 (nomenclature of Harris and Smyth, 1971), was isolated between the A6 and A5 ganglia and as close as possible to the A6 ganglion. The preparation was mounted in an oil-gap recording chamber described in detail elsewhere (Callec, 1974) and the A6 ganglion was continuously superfused (1mlmin⁻¹) with saline. The recording electrode was connected to the input of a high-impedance amplifier, whose output was passed to a digital oscilloscope, a pen chart recorder and a video cassette recorder for later off-line analysis. Under these conditions, unitary excitatory postsynaptic potentials (uEPSPs) resulting from the activity of presynaptic cercal mechanoreceptors were recorded. Composite EPSPs (cEPSPs) were triggered by a short electrical stimulation of the ipsilateral cercal nerve, eliciting an action potential in the GI when the spike threshold was reached. Direct measurements of postsynaptic membrane conductance changes were achieved using a balanced Wheatstone bridge circuit allowing application of square current pulses through the intraganglionic part of the GI. Variations of the postsynaptic membrane resting potential were monitored on a pen chart recorder. In the single-fibre oil-gap method, the resting potential of the GI is determined indirectly. According to the procedure described by Callec (1974), the resting potential can be measured after full depolarization of the intraganglionic part of the GI with isotonic KCl. In our experiments, the resting potential recorded using this method was – 71±3mV (N=6) and was similar to that obtained with intracellular microelectrodes (about – 75mV).

When necessary, voltage-clamp experiments were performed using a single-electrode voltage-clamp method adapted for use with the single-fibre oil-gap method (Mony et al. 1988).

The physiological saline had the following composition (in mmol l⁻¹): NaCl, 208; KCl, 3.1; CaCl₂, 5.4; NaHCO₃, 2; sucrose, 26; pH7.4. In experiments carried out in low-Na⁺ salines, Tris was substituted at equal molarity (pH was adjusted to 7.4 with Tris). For 150mmol l⁻¹ K⁺ saline, an equal molarity of Na⁺ was removed from the normal saline. Muscarinic agonists and the high-K⁺ saline were applied at constant pressure (1kgcm⁻²) within the dendritic tree of the GI using broken micropipettes (tip diameter 20 μm) connected to a pneumatic pressure-ejection system (Medical System Corp., USA), except where stated otherwise. Under our experimental conditions, pressure-ejection of saline never produced physiological effects on synaptic transmission. However, a small pressure artefact was occasionally observed. The antagonists and ion channel inhibitors were dissolved in the saline and applied externally.

Drugs were obtained as follows: arecoline hydrobromide (ARE), atropine sulphate (ATR), d-tubocurarine chloride (dTC), (–)-scopolamine hydrochloride (SCO), tetraethylammonium chloride (TEA⁺), tetrodotoxin (TTX) from Sigma; 4-diphenylacetoxy-N-methyl-piperidine methiodide (4-DAMP), (4-hydroxy-2-butynyl)-1-trimethylammonium m-chlorocarbanilate chloride (McN-A-343), methoctramine hydrochloride (MET), (±)-muscarine hydrochloride (MUS), (±)-quinuclidinyl benzilate (QNB) from RBI. Pirenzepine dihydrochloride (PZ) was a generous gift from Boehringer (Germany). QNB was first dissolved in a few drops of 0.1mol l⁻¹ HCl and then added to the physiological saline, which was checked for pH. The stock solution (10⁻³ mol l⁻¹) was diluted in saline to a final concentration of 10⁻⁶ mol l⁻¹.

When quantified, the results were expressed as mean ± S.E.M. Although many of the figures show results from single experiments, we confirmed each observation in at least three different preparations.

When bath-applied for 1–2min, 10⁻⁴ mol l⁻¹ ARE induced a small, slow depolarization which was followed by a slow repolarization (Figs 1A, 4A). The response typically observed during the bath-application of 10⁻⁴ moll⁻¹ ARE for 1min was a slow depolarization reaching 1mV in amplitude with a duration averaging about 15min (see Fig. 4A). However, in some preparations, similar bath-applications of ARE evoked a larger depolarization, which reached 3.5mV in amplitude and had a total duration of 60min. To examine muscarinic responses in the A6 ganglion further, ARE and other muscarinic agonists were pressure-ejected within the neuropile in order to be able to limit the duration of drug action and to study the kinetics of drug responses more precisely. Pressure-ejections of 10⁻⁴ mol l⁻¹ ARE (800ms in duration) within the dendritic tree of the GI mimicked the effects obtained during bath-application. However, the amplitude of the depolarization was smaller and the rise time was shorter (Fig. 1B). The peak amplitude of the slow depolarization was reached about 40s after the ejection of ARE (Fig. 1C). The ejection of 10⁻⁴ mol l⁻¹ MUS (800ms in duration) into the neuropile of GI elicited either the same small depolarization observed with ARE (not shown) or a slow depolarization which seemed to consist of two components: a small transient depolarization followed by the small prolonged depolarization (Fig. 1D). In contrast, pressure ejections of 10⁻³ mol l⁻¹ McN-A-343 (200ms in duration) produced a rapid depolarization with a fast decay phase (Fig. 1E) but failed to reveal the long-lasting depolarization.

The slow depolarization evoked by ARE was dose-dependent (Fig. 2). When the peak amplitude of the depolarization was plotted against the pressure-ejection duration (20ms to 1s) of 10⁻⁴ mol l⁻¹ ARE, a sigmoid curve was obtained. An ejection of ARE lasting 50 ms was sufficient to elicit a slow response, and the maximal depolarization (1.2mV) was obtained at around 800ms. The smooth line corresponding to the best fit (correlation coefficient r=0.992) through the mean data points (4–10 different preparations; Fig. 2) according to the Hill equation had a Hill slope factor of 1.65.

When the concentration of ARE in the ejecting pipette was increased to 10⁻³ mol l⁻¹, the induced depolarization consisted of two phases (Fig. 3A,B): a fast transient depolarization following the start of pressure-ejection, followed by a slower persistent depolarization. Ejections of ARE lasting 70ms elicited mainly the slow depolarization; the fast depolarization was elicited by pressure-ejections of 100ms or longer (Fig. 3A). The amplitudes of the fast and slow responses (Fig. 3A) and the duration of the slow response (not shown) increased with the duration of the pressure-ejection. During pressure-ejection of 10⁻³ mol l⁻¹ ARE (200ms duration), the maximal reduction of membrane resistance (about 20%) coincided with the peak of the transient depolarization (Fig. 3Ci), whereas a smaller reduction of membrane resistance (less than 5%) was obtained during the slow depolarization (Fig. 3Cii). The biphasic depolarization suggests that two distinct receptors might be activated by ARE. Antagonists selective for muscarinic and nicotinic receptors were therefore applied to characterize the cholinergic receptors mediating the slow response induced by 10⁻⁴ mol l⁻¹ ARE (1min bath-application) and receptors mediating the biphasic response induced by ejection of 10⁻³ moll⁻¹ ARE (200ms in duration).

The depolarizing effect of 10⁻⁴ mol l⁻¹ ARE was partly blocked by the muscarinic antagonist ATR (10⁻⁶ mol l⁻¹) and fully blocked by 10⁻⁵ mol l⁻¹ ATR (Fig. 4A). Bath-application of 5 × 10⁻⁵ mol l⁻¹ dTC blocked the fast depolarization of the biphasic response and simultaneous bath-application of dTC (5 × 10⁻⁵ mol l⁻¹) and SCO (10⁻⁶ mol l⁻¹) inhibited the slow response (Fig. 4B). Whatever the sequence of application of these antagonists, the effects were similar (Fig. 4C). The slow depolarization of the biphasic response was also reduced by 10⁻⁶ moll⁻¹ ATR and completely inhibited in the presence of 10⁻⁵ mol l⁻¹ ATR (not shown). For these antagonists, a wash of 60min was necessary to reverse the blockade. QNB (10⁻⁶ mol l⁻¹) also inhibited the slow depolarization of the biphasic response but this effect was only partially reversed by a washing period of 60min (Fig. 4D). These observations strongly suggest that ARE at various concentrations was able to elicit two distinct pharmacological responses at the postsynaptic membrane of the GI: a fast nicotinic depolarization and a slow muscarinic depolarization.

To characterize the pharmacological properties of the postsynaptic muscarinic response better, the responses obtained by pressure-ejections (200ms in duration) of 10⁻³ mol l⁻¹ ARE were examined in the presence of several specific muscarinic vertebrate subtype-selective antagonists: PZ, a vertebrate M₁ receptor subtype antagonist; MET, a vertebrate M₂ receptor subtype ligand; and 4-DAMP, a selective vertebrate M₃ receptor antagonist (Hammer et al. 1980; Doods et al. 1987; Melchiorre et al. 1987). As shown in Fig. 5, only PZ (10⁻⁵ moll⁻¹) reduced the slow muscarinic response, while MET and 4-DAMP, applied at the same concentration, did not. The PZ blockade was reversed after a washing period of 30min.

In vertebrates, McN-A-343 is known to be an M₁-selective agonist (Eglen et al. 1987; Micheletti and Schiavone, 1990). McN-A-343 has been used in different invertebrate CNS preparations to evoke a muscarinic response, with complex results (see Discussion). In a previous study (Le Corronc et al. 1991), we have shown that McN-A-343 acts at the presynaptic level and at the postsynaptic site. Pressure-ejections of a high concentration (10⁻³ mol l⁻¹) of McN-A-343 evoked a depolarization of the GI. Fig. 6A illustrates the effects of dTC and SCO on the depolarization induced by a pressure-ejection (180ms in duration) of 10⁻³ moll⁻¹ McN-A-343. 10⁻⁶ mol l⁻¹ SCO only slightly reduced the duration of the McN-A-343-induced depolarization, whereas 5X10⁻⁵ mol l⁻¹ dTC strongly inhibited the remaining depolarization. Identical experiments have been performed in the presence of 10⁻⁶ moll⁻¹ PZ because (i) in DUM neurone somata isolated from adults, McN-A-343 evokes a biphasic muscarinic response composed of a fast hyperpolarization fully antagonized by MET and a slow depolarization antagonized by 10⁻⁸ mol l⁻¹ PZ (Lapied et al. 1992), and (ii) in motoneurone D_f, the inward shift in the current–voltage relationship induced by McN-A-343 is blocked by 10⁻⁶ mol l⁻¹ PZ (David and Pitman, 1992). As shown in Fig. 6B, 10⁻⁶ mol l⁻¹ PZ had no effect on the amplitude or the duration of the depolarization induced by McN-A-343. This suggests that the postsynaptic mAChr of the GI is probably different from mAChrs in motoneurone D_f and in DUM neurones.

Fig. 7A shows that the biphasic response evoked by pressure-ejection of 10⁻³ moll⁻¹ ARE (200ms in duration) was unchanged in the presence of TTX (10⁻⁶ mol l⁻¹). This demonstrates that the muscarinic response was independent of the Na⁺ action potential mediated by input from another neurone. However, non-spiking interneurones may be involved in the muscarinic depolarization. In the presence of 10⁻⁶ mol⁻¹ TTX and 10⁻³ mol l⁻¹ Cd²⁺ (see below), the muscarinic response was unchanged, demonstrating that ARE acted directly on mAChrs of the GI under test. In tobacco hornworm (Trimmer and Weeks, 1991) and in lobster (Freschi and Livengood, 1989), an inward muscarinic current is described as TTX-insensitive but it was reduced and abolished when the external Na⁺ concentration was decreased or totally replaced. In our preparation, when the external Na⁺ concentration was reduced from 208mmol l⁻¹ to 100mmol l⁻¹ no significant changes in the slow depolarization were observed (Fig. 7B), but there was a decrease in the amplitude and duration of the fast depolarization. The effect of zero external Na⁺ could not be tested because, when all external Na⁺ was replaced with Tris, large oscillations of the membrane potential were observed. It should be noted that a hyperpolarization of the GI (by about 8mV) was recorded in the presence of the low-Na⁺ saline and that, before each ejection of ARE, the membrane potential was restored to its original level by injection of depolarizing current.

The possibility that the closure of a K⁺ conductance, such as that demonstrated in various muscarinic systems, is involved in the slow depolarization was tested by application of TEA⁺. When 10⁻³ mol l⁻¹ TEA⁺ was added to the extracellular bath solution, a consistent inhibition of the fast depolarization of the biphasic response was observed, without any effect on the amplitude of the slow depolarization. However, 5 × 10⁻³ mol l⁻¹ TEA⁺ was an effective blocker of the muscarinic response (Fig. 7C). TEA⁺, by itself, induced a small depolarization when applied to the A6 ganglion, but before pressure-ejections of ARE, the membrane potential was restored to the original resting potential by injection of a hyperpolarizing current. The effect of TEA⁺ on the fast depolarization was probably caused by a direct blockade of postsynaptic nicotinic receptors, as has been demonstrated previously in insect neurones (Benson, 1988). The effect of 5 × 10⁻³ mol l⁻¹ TEA⁺ supports the view that the slow depolarization is caused by a reduction of a potassium conductance. However, we cannot discard the possibility that TEA⁺ exerts an inhibitory action on mAChrs, because Drukarch et al. (1989) have shown that TEA⁺, at millimolar concentrations, displaces the specific binding of the muscarinic antagonist dexetimide.

To test the hypothesis that the muscarinic excitation was caused by the suppression of a K⁺ conductance, an inward K⁺ current was recorded in the presence of ARE according to the protocol described by Akaike et al. (1990). During the perfusion of a saline containing 10⁻⁶ mol l⁻¹ TTX and 10⁻³ mol l⁻¹ Cd²⁺ (see below) and under voltage-clamp at the resting potential, 100ms pressure-ejections of 150mmol l⁻¹ K⁺ saline into the neuropile induced an inward K⁺ current (Fig. 8A). ARE (10⁻⁴ mol l⁻¹), applied to the A6 ganglion for 1min, induced a slow inward current and reduced the inward K⁺ current (Fig. 8A), suggesting that the slow depolarization resulted from the muscarinic blockade of a K⁺ conductance. The reduction of the K⁺ current was not associated with an increase in the membrane resistance of the GI. In current-clamp experiments, a simultaneous slight reduction of the membrane resistance (Fig. 8B) and a marked decrease of the K⁺-induced depolarization (not shown) were observed in the presence of ARE (10⁻⁴ mol l⁻¹). These results could be explained by a decrease in a K⁺ conductance and by an increase in a supplementary cationic conductance.

David and Pitman (1991) have shown that the muscarinic excitation in motoneurone D_f is caused by the closure of voltage-dependent calcium channels and indirectly by the closure of calcium-activated potassium channels, so a direct action of postsynaptic mAChrs on calcium entry cannot be ruled out in GI. As illustrated in Fig. 9A–C, 2 × 10⁻³ mol l⁻¹ Co²⁺, 10⁻³ mol l⁻¹ Cd²⁺ and 10⁻³ mol l⁻¹ La³⁺ did not change the biphasic depolarization induced by 10⁻³ mol l⁻¹ ARE (200ms pressure ejection). In contrast, uEPSPs were inhibited by these calcium channel blockers, indicating the effective inhibition of presynaptic calcium channels (Fig. 9A–C). The present results indicate that, at the resting membrane potential and under our experimental conditions, postsynaptic voltage-dependent calcium channels were not activated or inhibited during the muscarinic excitation.

Fig. 9D shows that the perfusion of preparations with a saline containing 5.4 × 10⁻³ moll⁻¹ Ba²⁺ (i.e. at the same concentration as Ca²⁺) inhibited the slow depolarization of the biphasic response without affecting the fast depolarization and the unitary activity. Barium ions are potent blockers of potassium channels (Armstrong et al. 1982) and also of cation conductances activated by mAChrs (Inoue and Kuriyama, 1991). Calcium channel blockers and Ba²⁺ induced a small depolarization by themselves when applied to the A6 ganglion, but before pressure-ejections of ARE, the membrane potential was restored to the original resting potential by injection of a hyperpolarizing current.

In the tobacco hornworm Manduca sexta, the spike activity in the identified motoneurone PPR is increased in the presence of MUS because the spike threshold is lowered (Trimmer and Weeks, 1989a). In the GI of the cockroach, repetitive spiking cannot usually be recorded and it was therefore preferable to observe possible changes in the spike threshold induced by ARE on action potentials triggered by stimulation of afferent inputs. As described in Materials and methods, electrical stimulation of the cercal nerve induced a cEPSP in GI, which, when its amplitude reached the spike threshold, triggered a propagated action potential. Because ARE reduces the release of ACh by acting on presynaptic mAChrs and induces a small postsynaptic depolarization, the presynaptic electrical stimulation was slightly increased to restore the cEPSP to the threshold level and the membrane potential was returned to its original value before each evaluation of the threshold. A bath-application of 10⁻⁴ mol l⁻¹ ARE lasting 1min decreased the threshold for spikes evoked by afferent stimulation (Fig. 10A). Conversely, in the presence of 10⁻⁵ mol l⁻¹ ATR, the spike threshold was increased (Fig. 10B). Action potentials can be evoked in GI by mechanical stimulation of the cercal mechanoreceptors by a puff of air if the afferent uEPSPs summate to spike threshold. A puff of air, applied on the cercus through a broken glass micropipette, elicited two propagated action potentials (Fig. 10C). However, in the presence of 10⁻⁵ mol l⁻¹ ATR, the same wind stimulation elicited no action potentials, only a cEPSP. This implies that the spike threshold was at a higher level. These results demonstrate that postsynaptic mAChrs modified the spike initiation threshold of the GI under our experimental conditions.

Our results demonstrate the presence of postsynaptic physiologically active mAChrs at cercal afferent to giant interneurone synapses in the cockroach Periplaneta americana. Two muscarinic agonists (ARE and MUS) induced a small, long-lasting depolarization that was specifically inhibited by ATR, SCO and QNB but was insensitive to dTC. These observations suggest that the slow depolarization was mediated by an mAChr. Higher concentrations of ARE also evoked a fast transient depolarization that was specifically blocked by dTC but was not modified by muscarinic antagonists. This observation was consistent with the activation of postsynaptic nAChrs widely present within the dentritic tree of the GI (Callec, 1974, 1985). Similar effects have been observed in isolated neuronal somata from the thoracic ganglia of the locust, in which applications of ARE evoke a muscarinic and a nicotinic current (Benson, 1992). The systematic application of a range of cholinergic receptor ligands to motoneurone D_f of the cockroach (David and Sattelle, 1984) and to motoneurone PRR in tobacco hornworm (Trimmer and Weeks, 1989a) has revealed that the ACh receptors in these preparations are predominantly nicotinic. However, with muscarinic agonists, or with cholinergic agonists in the presence of nicotinic antagonists, a small muscarinic component is revealed (Benson, 1992).

In vertebrates, mAChrs have been subdivided into at least three subtypes (M₁, M₂ and M₃) defined on the basis of binding and functional studies using selective antagonists (for a review, see Mei et al. 1989). In the same way, several authors have used these selective antagonists to discriminate among mAChrs in the insect CNS. On the basis of the different affinity of PZ for various mAChrs, Knipper and Breer (1988) have provided biochemical evidence for the heterogeneity of mAChrs in adult locust ganglia. Two distinct populations of mAChrs have been localized: an M₁ subtype preferentially found on cell bodies and an M₂ subtype more predominant in nerve endings. More recently, two studies have characterized mAChrs in brain membranes from different insects by the displacement of bound QNB by selective antagonists for the M₁, M₂ and M₃ subtypes. From these studies, it appears that one subtype of mAChr is present, but the order of potency of muscarinic antagonists does not correspond to the vertebrate classification, although some similarities exist with M₁ or M₃ mammalian subtypes (Orr et al. 1991; Abdallah et al. 1991). In whole bulb mite homogenates, one population of mAChrs corresponding to the M₁ subtype has been described on the basis that PZ is effective in displacing bound QNB and MET is not (Huang and Knowles, 1990a).

In several neuronal preparations from insects, electrophysiological approaches have revealed a heterogeneity in the pharmacological profile of mAChrs and in the electrical responses evoked by muscarinic agonists. In isolated neuronal somata from the thoracic ganglia of the locust, MUS, oxotremorine, ARE and pilocarpine induce a slow inward current which is inhibited by non-specific muscarinic antagonists (QNB, SCO and ATR) and by the specific subtype antagonists PZ and 4-DAMP at concentrations below 10⁻⁶ mol l⁻¹ (Benson, 1992). Specific antagonists for the M₂ subtype (MET and AF-DX 116) are ineffective and the muscarinic agonist McN-A-343 is only very weakly active. The receptor bears some pharmacological resemblance to M₁ and M₃ receptors but does not correspond in detail to any of the defined mAChrs in mammals. In motoneurone D_f of the cockroach, an inward shift in the current–voltage relationship is induced after activation of mAChrs by ACh (in the presence of a-bungarotoxin), oxotremorine, MUS and McN-A-343. The responses of McN-A-343 and MUS are blocked by ATR, QNB, SCO and dexetimide at concentrations of 10⁻⁵ mol l⁻¹ or less. PZ (10⁻⁶ mol l⁻¹) and MET (10⁻⁶ mol l⁻¹) block the muscarinic response (David and Pitman, 1990, 1992). In the Manduca sexta motoneurone PPR, stimulation of afferents, in the presence of mecamylamine, evokes a slow long-lasting depolarization that is mimicked by oxotremorine or MUS, but McN-A-343 at 3X10⁻⁶ mol l⁻¹ is without effect. Low doses of SCO (10⁻⁷ mol l⁻¹) block the postsynaptic response, whereas PZ is effective at concentrations above 10⁻⁵ mol l⁻¹ but 4-DAMP is without effect (Trimmer and Weeks, 1989a,b). In isolated adult DUM neurones of the A6 ganglion of the cockroach, MUS evokes a slow depolarization, whereas McN-A-343 produces a biphasic response composed of a fast depolarization followed by a slow depolarization, indicating a peculiar effect of this agonist. The hyperpolarization is specifically blocked by 10⁻⁶ mol l⁻¹ MET and the depolarization by 10⁻⁸ mol l⁻¹ PZ (Lapied and Hue, 1991; Lapied et al. 1992). At the cercal afferent to giant interneurone synapses of the cockroach, presynaptic mAChrs are sensitive to ARE, carbachol, McN-A-343, oxotremorine and bethanechol and are partially blocked by 10⁻⁷ mol l⁻¹ AF-DX 116 and 10⁻⁷ mol l⁻¹ MET, but not by PZ (10⁻⁶ mol l⁻¹) and 4-DAMP (10⁻⁷ mol l⁻¹) (Le Corronc et al. 1991). In the present study, the postsynaptic muscarinic response is reduced by 10⁻⁵ mol l⁻¹ PZ, but 10⁻⁵ mol l⁻¹ MET and 10⁻⁵ mol l⁻¹ 4-DAMP are without effect. Furthermore, McN-A-343 at 10⁻³ mol l⁻¹ has a small effect on postsynaptic mAChrs and mainly shows an affinity for nAChrs. Our results agree with those reported by Trimmer and Weeks (1989b) for tobacco hornworm. This profile differs from the M₁, M₂ and M₃ mammalian classification and from other profiles reported in insects from binding or electrophysiological studies and also differs from the types of mAChrs described in lobster (Freschi, 1991).

Until now, few studies have investigated the ionic dependence and ionic currents involved in muscarinic responses in insects. In the cockroach motoneurone D_f, at membrane potentials more positive than –40mV, the inward current evoked by muscarinic agonists is reversed by TEA⁺. The outward current is blocked by Cd²⁺, indicating a reduction in the voltage-dependent calcium current and a consequent reduction in a Ca²⁺-activated K⁺ conductance (David and Pitman, 1990, 1991). In the Manduca sexta motoneurone PPR, the MUS-induced depolarization is not blocked by the absence of external Ca²⁺ (Trimmer and Weeks, 1989a) and the oxotremorine-induced inward current is TTX-insensitive but is reversibly abolished by the absence of external Na⁺ (Trimmer and Weeks, 1991). This current is larger near the resting potential of about – 50mV and declines with depolarization or hyperpolarization. In DUM neurones (resting membrane potential approximately – 50mV), the biphasic response is voltage-dependent, because the fast hyperpolarization decreases with hyperpolarization and gives an extrapolated reversal potential of – 91mV, which is closed to the equilibrium potential for K⁺. In contrast, the slow depolarization decreases with depolarization and gives a projected reversal potential of – 28mV, suggesting the participation of a non-specific conductance or a mixture of specific conductances (Lapied et al. 1992).

From the present results, we suggest that two conductances are probably involved in the slow muscarinic excitation in the GI: a decrease in a K⁺ conductance and an increase in a cationic conductance. In the vertebrate nervous system, slow muscarinic excitation is usually generated by the blockade of non-voltage-dependent or voltage-dependent K⁺ conductances. In bullfrog sympathetic neurones (Brown and Adams, 1980; Akasu et al. 1984; Jones, 1985), and in rat (Madison et al. 1987) and in guinea pig hippocampal neurones (Benardo and Prince, 1982), a well-characterized voltage-dependent K⁺ current, different from the delayed rectifier current, is selectively suppressed by muscarinic agonists; it has been called the M-current. mAChrs also depress the slow after-hyperpolarization that follows a series of spikes evoked by a depolarizing current pulse (Benardo and Prince, 1982; Cole and Nicoll, 1984; Madison et al. 1987; Washburn and Moises, 1992). This hyperpolarization is largely due to an increase in Ca²⁺-activated K⁺ conductance (gK_Ca) because it is blocked by calcium antagonists. mAChrs are thought to block gK_Ca at a step subsequent to calcium entry. The third type of K⁺ current reduced by mAChrs is a non-voltage-dependent K⁺ current that is activated at the resting membrane potential. Muscarinic inhibition of the resting K⁺ conductance has been reported in myenteric neurones (Morita et al. 1982), in hippocampal cells (Madison et al. 1987) and in adrenal chromaffin cells (Akaike et al. 1990). At negative membrane potentials, voltage-dependent K⁺ currents should be deactivated. Therefore, the effect of mAChrs is caused by the suppression of the K⁺ channels that maintain the resting membrane potential. In our experiments, ARE reduces the inward K⁺ current and the depolarization induced by high-K⁺ saline. The reduction of a K⁺ current is postulated, rather than the increase of a voltage-dependent calcium current or the reduction of gK_Ca, because the slow depolarization is insensitive to calcium channel blockers. We cannot differentiate the type of K⁺ channel (voltage-dependent or non-voltage-dependent) that contributes to the slow depolarization. However, a decrease in the membrane resistance is associated with the muscarinic response, which suggests inhibition of a K⁺ conductance and a concomitant increase of an inward conductance. To explain the weak effect of ARE on the membrane resistance, we can also hypothesize that the affected channels are so widely spaced that resistance changes cannot be measured. It is also possible that the muscarinic effect is voltage-sensitive, which might mask resistance changes measured by current pulse injections. In bullfrog symphathetic neurones, slow excitatory muscarinic currents are generated not only by a decrease in K⁺ conductance but also by a concomitant increase in Na⁺ and other cations conductances (Akasu et al. 1984; Jones, 1985).

In lobster, 5 × 10⁻³ mol l⁻¹ Ba²⁺ does not alter the muscarinic inward current (Freschi and Livengood, 1989). Furthermore, in the same preparation, TEA⁺ (3 × 10⁻² mol l⁻¹) does not block the muscarinic excitation, which is mainly carried by Na⁺. In GI, TEA⁺ and Ba²⁺, two K⁺ channel blockers, inhibit the slow depolarization. The pharmacological sensitivities of these channels were probably different from those described for lobster. In bullfrog sympathetic ganglion cells, Ba²⁺ imitates the action of MUS after a specific inhibition of the M-current (Constanti et al. 1981), but TEA⁺ and Ba²⁺ have been found to suppress a muscarinic-activated cation channel on chromaffin cells (Inoue and Kuriyama, 1991).

ARE modifies the effectiveness of afferent inputs by lowering the spike threshold, whereas ATR increases the same spike threshold. Similar results have been observed in Manduca sexta motoneurone PPR, where ionophoretic applications of MUS lower the spike threshold (Trimmer and Weeks, 1989a). In cercal afferent GI preparations showing a high unitary activity (i.e. a high release of ACh), the spike threshold triggered in GI is generally lower than in preparations showing poor unitary activity (B. Hue and H. Le Corronc, unpublished observations). Therefore, it appears that in physiological conditions ACh acts on postsynaptic mAChrs, suggesting that postsynaptic mAChrs might have a physiological role in the CNS of the cockroach. In the vertebrate nervous system, the depolarizing effects of muscarinic agonists are generally associated with an increase in excitability. Activation of a K⁺ current (M-current) holds the cell below threshold so that muscarinic inhibition of K⁺ currents causes a profound change in the firing characteristics of the cells even if they are clamped to the original resting potential (Constanti et al. 1981; Cole and Nicoll, 1984; Jones, 1985). In contrast, the muscarinic inward current induced by the increase of a conductance simply depolarizes the cell and has much less excitatory action (Jones, 1985).

In conclusion, these results demonstrate the presence of mAChrs on GI itself and suggest that they play a physiological role by lowering the spike threshold and consequently modifiying the integrative properties of the GI.

We would like to thank Dr Jack A. Benson for his comments on our manuscript and for correcting the English.