Modulation of Ca²⁺ and K⁺ Conductances in an Identified Insect Neurone by the Activation of an α-Bungarotoxin-Resistant Cholinergic Receptor

Evidence for three distinct classes of acetylcholine (ACh) receptor in the insect central nervous system (CNS) has been provided by binding studies. There are three specific binding components best characterised by their ability to bind (i) α-bungarotoxin (α-bgt), (ii) quinuclidinyl benzilate (QNB) and (iii) decamethonium or QNB (for reviews, see Pitman, 1985; Breer and Sattelle, 1987). These components may correspond to (i) ‘nicotinic’, (ii) ‘muscarinic’ and (iii) ‘mixed pharmacology’ ACh receptors respectively.

These studies have been supported by electrophysiological experiments which have demonstrated that there are several pharmacologically distinct functional cholinergic receptors in the insect CNS. α-bgt-sensitive (‘nicotinic’) receptors have been located on a number of postsynaptic neuronal membranes of the cockroach Periplaneta americana, including those of the slow depressor motoneurone (D_s) (Carr and Fourtner, 1980), giant interneurone 2 (Sattelle et al. 1983) and the fast coxal depressor motoneurone (D_f) (David and Sattelle, 1984). In D_f, these ‘nicotinic’ receptors are relatively insensitive to QNB and decamethonium. α-bgt-sensitive receptors have also been reported in dissociated neurones of the locust Locusta migratoria (Benson, 1992).

In addition to these essentially ‘nicotinic’ receptors, a number of different insect CNS preparations also appear to possess receptors with properties similar to those of vertebrate muscarinic receptors (Lapied et al. 1990, 1992; Le Corronc and Hue, 1991; Trimmer and Weeks, 1989, 1993; Trimmer, 1994, 1995). The gene for a muscarinic receptor (Dm1) has been isolated from Drosophila melanogaster (Onai et al. 1989; Shapiro et al. 1989). These receptors have characteristics similar to those of mammalian M₁ and M₃ muscarinic receptor subtypes when expressed in mammalian cells (Shapiro et al. 1989; Blake et al. 1993) and appear to be particularly dense in Drosophila melanogaster antennal lobes. QNB-sensitive receptors which are activated by muscarinic agonists have been found in unidentified dissociated neuronal somata of the locust (Benson, 1992) and D_f somata of the cockroach (David and Pitman, 1993a). These two receptors show distinct pharmacological differences, the most noticeable being that, in D_f, both the M₁-selective agonist McN-A-343 (McN) and nicotine are potent, while in locust somata the actions of these two ligands are very weak. The McN-sensitive receptor in D_f is insensitive to α-bgt and yet sensitive to a number of other nicotinic and muscarinic agents, including decamethonium, nicotine and muscarine; this pharmacological profile closely resembles that of the decamethonium binding component isolated from insect CNS (a putative ‘mixed’ receptor) (Mansour et al. 1977; Cattell et al. 1980). The above evidence suggests that the receptors mediating the electrophysiological response to McN may be identical to the putative ‘mixed’ pharmacology cholinergic binding component. In view of the unique properties of the receptor mediating responses to McN in D_f, we shall refer to this receptor as the ‘McN-sensitive’ or ‘mixed pharmacological profile muscarinic receptor’ (m_MAChR). This definition distinguishes the responses mediated by McN-sensitive receptors from the depolarizing responses evoked by the muscarinic agonist oxotremorine that have been reported in D_f somata at resting membrane potential (Bai and Sattelle, 1994). Responses to oxotremorine are not mimicked by McN and are apparently sensitive to α-bgt since, at the resting potential, no response to ACh is observed in the presence of this antagonist (David and Sattelle, 1984).

We have previously shown that, in D_f, bath-applied McN suppresses Ca²⁺-dependent but not voltage-dependent outward currents (David and Pitman, 1995a). Here we have examined the time course and ionic basis of currents induced by bath-applied McN in order to characterize more fully the mechanisms underlying the effects of activation of m_MAChRs in this neurone. A brief account of some aspects of this work has been published previously (David and Pitman, 1993b).

Nerve cords of adult male cockroaches Periplaneta americana L. were prepared for recording as described previously (David and Pitman, 1993a). The saline had the following composition (in mmol l⁻¹): NaCl, 214.0; KCl, 3.1; CaCl₂, 9.0; sucrose, 50.0; TES, 10.0 (pH adjusted to 7.2 using 1.0 mol l−1 NaOH). A portion of the nerve cord consisting of the three thoracic and first three abdominal ganglia was placed in a chamber containing saline (volume, 2 ml) which was circulated using 100 % O₂. Pharmacological agents were added to the bath in 20 μl volumes and removed by continuous perfusion at a rate of 2 ml min⁻¹. The method of oxygenation resulted in a rapid circulation of the saline within the chamber which produced a virtually complete distribution of added solutions within 2 s. The cell body of the metathoracic ‘fast’ coxal depressor motoneurone (D_f) was voltage-clamped using two microelectrodes, which were normally filled with 2.0 mol l⁻¹ potassium acetate and had an electrical resistance of 8–12 MΩ. The voltage-clamp amplifier had a compliance of ±100 V. Following penetration of the cell body by both microelectrodes, the preparation was left for 30–60 min to stabilise before voltage-clamping the membrane at −80 mV. To reveal inward currents passing through voltage-dependent Ca²⁺ channels, K⁺ conductances were blocked by filling the electrodes with 1 mol l⁻¹ CsCl and replacing CaCl₂ in the saline with equimolar BaCl₂. In those experiments where outward K⁺ currents were blocked to reveal inward currents, leakage current correction was performed using pulses from −80 to ⁻¹20 mV. Command steps were generated and data recorded using a computer and CED1401 (Cambridge Electronic Design, Cambridge, UK) interface and associated software. Drug-induced currents were determined by subtracting each current trace obtained in the absence of drug from the current obtained in the presence of drug at the same command potential. All current/voltage (I/V) relationships were constructed using currents measured 10 ms after the beginning of a voltage step. In all cases, the I/V relationships shown are representative of a minimum of three experiments and, for clarity, show data from a single experiment. For experiments involving the injection of the Ca²⁺ chelator 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetra-acetic acid (BAPTA), microelectrodes were filled with a 0.1 mol l⁻¹ solution of the tetrapotassium salt and current pulses (200 ms, 2 Hz) sufficient to hyperpolarize the cell by 30 mV were passed through the electrode. BAPTA injection was continued until [Ca²⁺]_i had been lowered sufficiently to evoke Ca²⁺-dependent action potentials in response to a depolarizing current pulse (see Pitman, 1979). In experiments which required selective induction of I_K,Ca, Ca²⁺ was injected into the cell by ionophoresis using positive current pulses applied via a 100 MΩ resistor to a third intracellular microelectrode filled with 1 mol l⁻¹ CaCl₂.

The following drugs were used: acetylcholine chloride, α-bungarotoxin (α-bgt), BAPTA (Sigma Chemical Co. Ltd, UK) and McN-A-343 [(4-hydroxy-2-butynyl)trimethylammonium chloride] (Research Biochemicals Inc. USA).

With the cell voltage-clamped near its normal resting potential (−80 mV), ACh induces an inward current which desensitizes with repeated applications; this current results from the activation of receptors with many of the properties of vertebrate peripheral nicotinic receptors: it is blocked by externally applied α-bgt (1 μmol l⁻¹ for 60 min) and is carried by Na⁺, K⁺ and Ca²⁺ (David and Sattelle, 1990). Although at −80 mV ACh produces little or no response in the presence of α-bgt (David and Sattelle, 1984), when the membrane potential is stepped to 0 mV by applying a 15–50 ms voltage command pulse once every minute, a response can be observed. This response is normally biphasic, consisting of an increase and subsequent reduction in net outward current (Fig. 1A). The muscarinic agonist McN-A-343 (McN) has been used routinely in this study for the following reasons: (i) its effects are similar to those produced by ACh in the presence of α-bgt when the membrane potential is jumped to 0 mV (Fig. 1B) (see David and Pitman, 1993a, 1995a), eliminating the need to pretreat preparations with antagonist to eliminate actions upon nicotinic receptors (the McN response is unaltered by α-bgt); (ii) the effects of McN (a vertebrate M₁ receptor agonist) appear to be specific, since they can be blocked by the M₁ receptor antagonist pirenzepine as well as by a number of receptor subtype non-selective muscarinic antagonists (David and Pitman, 1993a); (iii) the effects of McN are reversible and have no long-term deleterious effect upon the preparation; and (iv) this agonist does not activate a population of cholinergic receptors in this neurone that has been reported to mediate a depolarizing response from the resting potential (Bai and Sattelle, 1994).

Although at 100 μmol l⁻¹, both ACh and McN in the presence of α-bgt normally produce biphasic effects at 0 mV, the relative contribution of each phase varied in different preparations; in 12 out of 32 neurones to which McN was applied, no early phase of increased outward current was observed at 0 mV (see below). In the remaining 20 neurones, the increase in outward current induced by 100 μmol l⁻¹ McN was maximal 1.9±0.2 min (mean ± S.E.M.) after applying the agonist and had a mean peak amplitude of 85.2±17.9 nA. The component consisting of an inwardly directed current (or reduction in outward current) observed at membrane potentials more positive than −40 mV results from suppression of I_K,Ca (David and Pitman, 1993a, 1995a). At 0 mV, this inward current was maximal 10.8±1.3 min after applying 100 μmol l⁻¹ McN and had a mean peak amplitude of 196±17.7 nA (N=32).

The voltage-dependence of the McN response was determined by applying a series of voltage command pulses of 15 ms duration at 0.5 Hz over the potential range ⁻¹20 to +80 mV during a single agonist application. The time course of these effects was followed by repeating this series of voltage commands once every minute. Currents evoked using this protocol were used to construct I/V relationships.

In normal saline, the I/V relationship of the D_f neurone soma is essentially linear over the potential range ⁻¹20 to −40 mV. At potentials more positive than −40 mV, the membrane shows strong rectification due to an increase in K⁺ conductance consisting of at least two different components: a voltage-dependent current (I_K,V) and a current that is activated by the entry of Ca²⁺ through voltage-dependent channels (I_K,Ca) (Thomas, 1984). The time-dependent effect (1–7 min exposure) of 100 μmol l⁻¹ McN upon membrane currents generated by depolarizing steps to +80 mV is shown in Fig. 2A; at 1 min, McN caused an increase in outward current, while at 3 and 7 min, the net outward current was reduced to below the initial control level. In this experiment, the preparation was pretreated with α-bgt, showing that the effect of McN is not blocked by this antagonist. The effect of McN (3 and 7 min exposure) on the I/V relationship of the neurone is shown in Fig. 2B (currents sampled 10 ms from the start of the command step). The most marked effect is a reduction in the rectification (i.e. a reduction in outward current) seen at potentials more positive than −20 mV. This shift corresponds to the more robust second phase of the current shown in Fig. 1B, the pharmacology of which formed the subject of a previous paper (David and Pitman, 1993a). The voltage-and time-dependence of the current induced by McN (determined by subtracting the currents obtained in normal saline from those obtained during similar command steps in the presence of McN) is shown at various times after applying the agonist in Fig. 2C,D. The data presented in Fig. 2C,D were obtained during the same experiment. The information has been presented in two graphs to allow the currents obtained 1, 2 and 3 min after the onset of drug application (Fig. 2C) to be shown on a more expanded scale than those obtained after 4, 5 and 6 min (Fig. 2D). The reduction in outward current induced by McN (Figs 1, 2A,B) is represented as an induced inward current following computation of the difference current. This second phase of the McN response is further referred to as a McN-induced inwardly directed current.

In Fig. 2C, a McN-induced inwardly directed current developed at potentials more positive than 0 to +20 mV between 1 and 2 min after application of the agonist. This period varied between preparations, and it is this variability which is apparently responsible for the absence in some cells of the early outward component of the current induced by McN at 0 mV (Fig. 3A). The mean amplitude of the inwardly directed current observed at +20 mV following a 6 min exposure to 100 μmol l⁻¹ McN was 137±24 nA (mean ± S.E.M., N=17). The inwardly directed component of the McN-induced current did not wane with prolonged (up to 30 min) exposure to the agonist (N=5), indicating that it exhibits no significant desensitization. The effect of such prolonged applications of McN upon the I/V relationship was reversed after washing in drug-free saline for 60 min.

The first phase of the McN-induced current (seen as an increase in outward current at 0 mV in Fig. 1) corresponds to the McN-induced currents observed over the potential range ⁻¹20 to −20 mV in the example shown in Fig. 2C,D. This consisted of a current that was outwardly directed at −20 mV and declined at more negative potentials, reversing to become inwardly directed beyond −60 to −80 mV (mean −72±3 mV, N=17). This value is close to the reversal potential for K⁺ (E_K) as measured under conditions similar to those used in these experiments. (The value of E_K under differing conditions is considered in the Discussion.) The mean amplitude of the current at −20 mV induced by the application of 100 μmol l⁻¹ McN for 2 min was 50.1±5.6 nA (N=17).

The McN-induced current between 0 mV and +80 mV is noticeably U-shaped between 1 and 3 min after application of the agonist (Fig. 2C) and is then monotonic between 4 and 6 min after agonist application (Fig. 2D). This change probably results from superimposition of the two current components induced by McN, each of which has a different time-dependency, bringing about a progressive shift in the relative contributions made by the each component during the course of the response; when the first phase (outwardly directed between −20 to +80 mV) is relatively large, it produces an outward inflection and, consequently, the McN-induced current has a U shape. With increasing time after agonist application, the second phase (inwardly directed between −20 to +80 mV) of the response becomes dominant and, as a result, the outward component of the U shape in the I/V relationship of the drug-induced current progressively declines and disappears.

To establish whether the initial outwardly directed component of the response to McN observed at command potentials close to 0 mV is triggered by an influx of Ca²⁺ from the external solution, Ca²⁺ in the saline was replaced by Ba²⁺ (for 30 min prior to recording), since this divalent cation is capable of passing through Ca²⁺ channels, but cannot activate I_K,Ca (Connor, 1979). Under these conditions, Ba²⁺ saline results in selective inhibition of I_K,Ca in D_f somata (David and Pitman, 1995b). Experiments reveal that, although the early outwardly directed component of the McN-induced current persisted in Ba²⁺ saline (amplitude at −20 mV after 2 min of McN application 60.8±13.4 nA, N=5), the delayed inwardly directed component observed in Ca²⁺-containing saline (Fig. 3A) no longer developed in Ba²⁺ saline (Fig. 3B) (amplitude at +20 mV after 6 min of McN application 111.0±13.8 nA outward, N=5). These observations indicate that the early outwardly directed component of the McN-induced current is not dependent upon Ca²⁺ influx across the surface membrane, although it may be triggered by release of Ca²⁺ from intracellular stores (see below). The absence of a delayed inwardly directed current in Ba²⁺ saline is consistent with earlier work which has shown that McN suppresses the Ca²⁺-dependent outward current component of the I/V relationship in D_f somata, with no significant effect upon the voltage-dependent outward current (David and Pitman, 1995a). Since I_K,Ca is not activated by Ba²⁺, this current could not be further attenuated by McN. The principal mechanisms that could underlie such a reduction in I_K,Ca are (i) a direct action on Ca²⁺-dependent K⁺ channels, (ii) a reduction in Ca²⁺ entry through voltage-dependent channels, or (iii) a change in the ability of a rise in intracellular [Ca²⁺] to activate I_K,Ca. Support for the second proposal has been provided by examining the effects of McN on Ca²⁺ currents and upon I_K,Ca induced by intracellular Ca²⁺ injection.

K⁺ currents were suppressed by impaling neurones with microelectrodes filled with 1 mol l⁻¹ CsCl and, in some cases, by replacing all of the Ca²⁺ in the saline with Ba²⁺. Under these conditions, command pulses more positive than approximately −50 mV revealed an inward current which partially inactivated during the depolarizing step (Fig. 4A). A large proportion of this inward current is carried through voltage-dependent Ca²⁺ channels since such currents evoked in either Ca²⁺ (N=4) or Ba²⁺ (N=3) saline were blocked by external Cd²⁺ (Fig. 4C,D). The I/V relationship of Ca²⁺ currents (I_Ca) observed under these conditions differs from the I/V relationship of I_K,Ca. This appears to be largely caused by a discrepancy between the apparent reversal potential for I_Ca (+10 to +20 mV) and true E_Ca (+150 to +180 mV, Thomas, 1984) which probably results from outward current carried either through Ca²⁺ channels by ions other than Ca²⁺ or through other types of channel which were not blocked under the experimental conditions. This issue is discussed in more detail by David and Pitman (1995b). At −20 mV, the Ba²⁺ current was attenuated by 100 μmol l⁻¹ McN applied for 15 min (mean percentage attenuation 80±31, N=4) (Fig. 4A). The effects of 100 μmol l⁻¹ McN on the I/V relationship in Ba²⁺ saline using Cs⁺ electrodes are shown in Fig. 4B. A McN-induced current was observed at potentials more positive than approximately −50 mV, peaked near 0 mV and declined to near zero at about +20 mV.

These experiments indicate that activation of m_MAChRs by McN indirectly decreases I_K,Ca by reducing current flowing through voltage-dependent Ca²⁺ channels. Since I_K,Ca is large relative to I_Ca, the net effect of McN is to produce a reduction in outward current. The early outward phase of the McN-induced current (Fig. 1), however, is the result of an increase in K⁺ conductance.

To examine the possibility that, in addition to suppressing I_Ca, McN acts directly upon I_K,Ca, Ca²⁺ was injected into the neurone soma by ionophoresis through a third microelectrode containing 1 mol l⁻¹ CaCl₂ while the cell membrane potential was controlled by voltage-clamp. In such experiments, a 20 ms pulse of 1–3 μA (Ca²⁺ microelectrode positive) resulted in a brief current, which showed outward rectification (Fig. 5A). This current reversed between −80 and −90 mV (mean −82±4 mV, N=3) and was outwardly directed at more positive potentials. This value is close to E_K determined under similar conditions (David and Sattelle, 1990) (see Discussion). The characteristics of this current are those previously attributed to I_K,Ca in this neurone (see Thomas, 1984). The current was unaffected by application of 100 μmol l⁻¹ McN for up to 11 min (N=4) (Fig. 5B). These findings show that McN appears neither to alter the Ca²⁺-sensitivity of I_K,Ca nor directly to modulate the component of I_K,Ca that is activated by intracellular Ca²⁺ injection. It is, however, possible that McN can alter a component of I_K,Ca that is activated by influx of Ca²⁺ through the surface membrane but is insensitive to Ca²⁺ injection.

Since injection of inositol-1,4,5-trisphosphate (InsP₃), which induces the release of Ca²⁺ from intracellular stores, into D_f somata results in the generation of biphasic effects similar to those evoked by McN (David and Pitman, 1993c), it seemed likely that both components of the McN-induced current were triggered by a rise in [Ca²⁺]_i. This possibility was investigated by establishing whether McN-induced currents persist when [Ca²⁺]_i is buffered by intracellular injection of the Ca²⁺ chelator BAPTA. In fact, BAPTA injection suppressed both current components induced by McN; the peak outward current (measured at −20 mV) was suppressed by 82±5 % and the peak inward current (measured at +20 mV) was suppressed by 83±10 % (N=4) (Fig. 6). These observations suggest that a rise in [Ca²⁺]_i mediates the electrophysiological effects of McN.

Experiments performed on the cell body of D_f indicate that the application of either the muscarinic ligand McN-A-343 or of ACh in the presence of α-bgt normally induces a biphasic current at 0 mV but very little current at the normal resting potential of −80 mV. In some (approximately 30 %) cases, however, only an inwardly directed current is observed at 0 mV after exposure to these ligands. The time course of the change in current observed at 0 mV is such that the outward component, when visible, always precedes the inwardly directed component. Analysis of the effects of m_MAChR activation over a range of membrane potentials indicates that where no early increase in outward current is observed at 0 mV (c.f. David and Pitman, 1993a) this is because it is masked by rapid development of the inwardly directed current; in such cases, however, an early component is visible at more negative potentials (e.g. Fig. 2C).

The initial increase in outward current appeared to result from an increase in K⁺ conductance since it was eliminated by internal Cs⁺ and reversed at membrane potentials between −60 and −80 mV. The reversal potential for K⁺ (E_K) in resting D_f somata has been estimated at −97 mV using a value for [K⁺]_i determined using K⁺-sensitive microelectrodes (David and Sattelle, 1990). Electrophysiological observations, however, indicate that E_K can undergo a positive shift in neurones that are repeatedly depolarized (such as during many voltage-clamp protocols) (Thomas, 1984). Analysis of tail currents generated by these voltage-clamp protocols indicates that E_K may attain values of between −40 and −80 mV, depending upon the magnitude, duration and interval between depolarizing steps (J. A. David and R. M. Pitman, personal observations). Thomas (1984), from measurements of neuronal capacitance, attributed this inward shift in tail currents following depolarizing command steps to an extensive infolding of the surface membrane. This proposal is supported by ultrastructural studies, which show extensive membrane invagination and close glial ensheathment of cockroach neurone somata (Smith and Treherne, 1965; Lane et al. 1982). These observations explain the difference between E_K observed in pulse protocol experiments (−60 to −80 mV) and following Ca²⁺ injection (−80 to −90 mV), in which the membrane potential was clamped relatively close to the normal cell resting potential throughout an experiment.

The early outwardly directed component of the McN-induced current appears to involve an increase in I_K,Ca rather than I_K,V, since it is blocked by intracellular injection of BAPTA and because it coincides with a rise in [Ca²⁺]_i produced by McN (David and Pitman, 1996). The rise in [Ca²⁺]_i responsible for this increase in I_K,Ca appears to be caused by release of Ca²⁺ from intracellular stores, since it is not suppressed by substitution of external Ca²⁺ with Ba²⁺. The outward component of the McN-induced current increases with time in Ba²⁺ saline (Fig. 3B), suggesting that, in normal saline, the outward component appears to be transient only because it becomes masked by the development of the delayed inwardly directed current. This is consistent with the analysis of the effects of McN observed over a range of membrane potentials discussed above.

The involvement of I_K,Ca in the inwardly directed component of the McN response can be observed most readily at relatively positive membrane potentials (+50 to +150 mV), where the I/V relationship in D_f somata shows a characteristic N shape which results from activation of I_K,Ca by Ca²⁺ entering through voltage-dependent channels (Thomas, 1984). It has been shown previously that McN suppresses this N-shaped region of the I/V relationship, but has relatively little effect on outward currents more positive than +150 mV, suggesting that the application of McN results in suppression of I_K,Ca and not I_K,V (David and Pitman, 1995b). McN, however, does not act directly upon this current, but exerts its effect indirectly by suppressing the flow of current through voltage-dependent Ca²⁺ channels. This conclusion is based on two lines of evidence. First, application of McN had no effect upon I_K,Ca evoked by brief intracellular Ca²⁺ injections. Second, when K⁺ currents have been suppressed by intracellular injection of Cs⁺ and by substitution of external Ba²⁺ for Ca²⁺, positive command steps to potentials between −40 and 0 mV produced inward currents which were reduced or eliminated by application of McN. It is unlikely that this inward current has a Na⁺ component, since it is blocked by external Cd²⁺ and by organic Ca²⁺ channel blockers (data not shown). Further evidence for this comes from the fact that neither of the active electrical events normally recorded from D_f somata (plateau potentials and action potentials) has any detectable Na⁺ component (Pitman, 1979; Hancox and Pitman, 1991, 1992); a contribution from this ion to action potentials is only ever observed chronically after colchicine treatment or axotomy or following relatively prolonged hypoxia (Pitman, 1975, 1988). The delayed inwardly directed current induced by McN, therefore, appears to be produced by suppression of a Ca²⁺ current which, in turn, produces an even greater suppression in outward I_K,Ca. Any decrease in I_K,Ca which exceeds the reduction in I_Ca would be seen as a net inward shift in currents recorded from the neurone. It is likely that the McN-induced reduction in voltage-dependent Ca²⁺ current is a consequence of the early rise in [Ca²⁺]_i triggered by McN, since elevation in [Ca²⁺]_i produced by injecting InsP₃ into D_f produces a biphasic current similar to that evoked by McN (David and Pitman, 1993c). In fact, it is likely that the effects of McN upon D_f somata are mediated by a rise in InsP₃ turnover, since McN can cause an increase in InsP₃ levels in cockroach thoracic ganglia (David and Pitman, 1994). The muscarinic antagonist pirenzepine blocks both this biochemical effect of McN on whole nerve cords and the electrophysiological effects of McN on D_f somata. Inactivation of Ca²⁺ channels produced by a rise in [Ca²⁺]_i, as proposed here, has been reported in a number of other preparations including Paramecium, molluscan neurones, stick insect skeletal muscle and mammalian heart cells (e.g. Eckert et al. 1981; Plant et al. 1983; Ashcroft and Stanfield, 1982; Lee et al. 1985).

The pharmacological profile of m_MAChRs on the soma of D_f reported by David and Pitman (1993a) was determined during the delayed inwardly directed component of the current induced by muscarinic agonists. It is possible that this and the early outward component of the response to McN or ACh (in the presence of α-bgt) are each mediated by separate classes of receptor. It is more likely, however, that both components of the response are mediated by a single receptor coupled to a rise in [Ca²⁺]_i, because release of intracellular Ca²⁺ by intracellular injection of InsP₃ mimics the biphasic effect of McN (David and Pitman, 1993c). It appears that the soma of this neurone does possess at least one class of ‘muscarinic’ receptor besides that investigated in the current study; Bai and Sattelle (1994) reported that, although McN had no effect upon current-clamped D_f somata (resting potential −75 to −85 mV), oxotremorine produced depolarization. The effects of McN may have been overlooked because the resting potential is sufficiently close to E_K that any increase in K⁺ conductance was virtually invisible and because this potential is too negative for activation of the delayed inwardly directed current. Oxotremorine, therefore, must act upon a receptor with a different pharmacology and operating via a different ionic mechanism from that mediating McN responses. Some biphasic cholinergic responses recorded from insect central neurones do result from activation of two distinct receptor types; ACh and nicotine evoke two-component depolarizing responses in cockroach dorsal unpaired median (DUM) neurone somata (Lapied et al. 1990). Receptors mediating the slow component are sensitive to α-bgt, d-tubocurarine, pirenzepine and gallamine, while receptors mediating the fast component are resistant to all these antagonists. In the same type of neurone, McN produces biphasic responses consisting of a fast hyperpolarization followed by a slow depolarization, although ACh and muscarine induce only a depolarization (Lapied and Hue, 1991; Lapied et al. 1992). Although McN-induced hyperpolarizations in DUM and D_f neurones both appear to result from an increase in membrane K⁺ conductance, they appear to involve different receptors, since ACh does not mimic McN responses observed in DUM cells. The McN-induced slow depolarization in DUM neurones also seems to be distinct from the inwardly directed phase of the response reported here, since it has an extrapolated reversal potential of −28 mV, suggesting that it results from a non-specific cation conductance rather than from suppression of a K⁺ current. A muscarinic-receptor-induced reduction in K⁺ conductance has been reported in ventral giant interneurones of the cockroach Periplaneta americana (Le Corronc and Hue, 1993). This muscarinic response, however, differs from the McN response of D_f both in its pharmacology (McN is a weak agonist) and in its ionic basis (the response is not modified by external Cd²⁺). In the locust Locusta migratoria, a synaptic input to suboesophageal vasopressin-like immunoreactive (VPLI) interneurones evokes a biphasic excitatory postsynaptic potential (EPSP) with two pharmacologically distinct components: a fast, transient component mediated by nicotinic receptors and a slow, sustained component mediated by muscarinic receptors which are blocked by scopolamine (Baines and Bacon, 1994). Voltage-clamp observations revealed a muscarine-induced inward current which reverses between −60 and −50 mV and increases non-linearly with depolarization. This current in VPLI interneurones may, like the delayed inwardly directed component of the McN response in D_f, result from a decrease in K⁺ conductance; the mechanisms of these two responses, however, apparently differ, since a voltage-dependent K⁺ current is implicated in the former (Baines and Bacon, 1994), while the latter involves a reduction in I_K,Ca resulting from suppression of a Ca²⁺ current. Oxotremorine induces a predominantly Na⁺ inward current in proleg retractor motoneurones (PPR) of larval Manduca sexta by activating receptors which appear to be muscarinic, since they are blocked by scopolamine and pirenzepine but not by mecamylamine (Trimmer and Weeks, 1989, 1993; Trimmer, 1994). The oxotremorine response is inhibited by Ca²⁺ channel antagonists such as Cd²⁺, suggesting that Ca²⁺ entry is required for the muscarinic-receptor-induced increase in Na⁺ conductance.

A speculative, physiological role for a biphasic effect of mAChR activation may be to adjust the sensitivity of the neurone according to the extent of synaptic input; initial activation would evoke an increase in I_K, which would tend to hold the neurone close to the resting potential and limit the effects of further synaptic input. During prolonged muscarinic activation, the reduction in voltage-dependent I_K,Ca would tend to increase the responsiveness of the neurone to synaptic excitation in a manner similar to that produced by M-current inhibition in sympathetic ganglia (Adams et al. 1986).

We thank the BBSRC for grant GR/G63520 which supported this research.