α-Bungarotoxin Blocks Excitatory Synaptic Transmission Between Cercal Sensory Neurones and Giant Interneurone 2 of the Cockroach, Periplaneta Americana

The radiolabelled, receptor-specific ligands ^12SI-α-bungarotoxin and [³H]quinuclidinyl benzilate, which probe respectively the nicotinic and muscarinic acetylcholine receptors of vertebrates (cf. Yamamura, Enna & Kuhar, 1978), have also enabled the characterization of two, pharmacologically distinct, putative cholinergic receptors in insects (cf. Sattelle, 1980). Nevertheless, to date there has been detailed study of the competence of these receptor ligands to modify transmission at presumed cholinergic synapses in the insect central nervous system.

In the terminal abdominal ganglion of the cockroach, Periplaneta americana (L.), putative cholinergic synapses are thought to mediate a direct, monosynaptic connection between sensory neurones of the cercus and giant interneurones which ascend the ventral nerve cord (Callec, 1974). It has been demonstrated that cholinergic ligands are particularly active at these insect central synapses (Yamasaki & Narahashi, 1960; Shankland, Rose & Donniger, 1971; Callec, 1974; Sattelle, 1980).

The cercal afferent, giant interneurone synapses are suitable for detailed pharmacological investigation of cholinergic agonists and antagonists. Of particular value is the facility with which single afferents can be selectively activated by mechanical stimulation of individual cereal mechanoreceptors (Callec, Guillet, Pichon & Boistel, 1971). It is also possible to record synaptic potentials for long periods from a single giant interneurone by the oil-gap, single-fibre technique (Pichon & Callec, 1970), using the dissected axon of one of the giant interneurones. After pharmacological experiments the giant axon can be cobalt backfilled for identification (Harrow, Hue, Pelhate & Sattelle, 1980).

In an earlier, brief report (Harrow, Hue, Pelhate & Sattelle, 1979) pharmacological and anatomical approaches were combined to demonstrate that α-bungarotoxin at a concentration of 1·0 X 10⁻⁸M blocks synaptic transmission between cereal afferents and giant interneurone 3 (GI3 - nomenclature of Harris & Smyth, 1971). For this study we chose, instead, giant interneurone 2 (GI2) whose axon is more readily dissected than that of any other giant interneurone in the cockroach and is therefore well suited for pharmacological studies using the oil-gap, single-fibre technique. Also, the cell body of GI 2 is located conveniently for microelectrode impalements, thereby enabling comparative studies on the pharmacology of synaptic and extrasynaptic acetylcholine receptors of the same identified neurone (Harrow & Sattelle, 1983).

The work reported in this paper investigates in detail the effect of the cholinergic receptor probes α-bungarotoxin and quinuclidinyl benzilate on synaptic transmission between cereal sensory neurones and GI 2 in the terminal (sixth) abdominal ganglion of the cockroach Periplaneta americana. In addition to these electrophysiological studies, autoradiography has been employed to examine the distribution of ^l25I-α- bungarotoxin binding sites in the same ganglion.

Adult male cockroaches (Periplaneta americana L.) were used throughout this study.

Purified α-bungarotoxin was obtained from the Miami Serpentarium Laboratories, Florida, U.S.A. The ¹²⁵I-α-bungarotoxin used in this study was prepared by the Chloramine T method (Greenwood, Hunter & Glover, 1963) as described previously Bchmidt-Nielsen, Gepner, Teng & Hall, 1977). Quinuclidinyl benzilate was provided by Dr J. F. Donnellan of the Shell Bioscience Laboratories, Sittingbourne, U.K.

Samples of α-bungarotoxin (2mgml⁻¹) in 0·0625m Tris (hydroxyethyl) aminomethane-HCl (Tris-HCl) buffer (pH 6·8) containing 2% sodium dodecylsulphate (SDS), 12·5% polyacrylamide, 10% glycerol, 5% 2-mercaptoethanol and 0·001% bromophenol blue were heated for 1·5 min in a boiling water bath and 10αl of this sample solution was applied to each gel. The procedure of Laemmli (1970) was adopted for the preparation and running of gels. The composition of the electrode buffer was as follows: 0·1% SDS, 0·192M glycine, 0·025 M Tris, pH 8·3. Tube gels of 3 mm diameter and 100 mm in length were run at 22 °C for 4·5 h using a current of 1 mA per tube. Gels were stained with Coomassie blue by the method of Fairbanks, Steck & Wallach (1971).

Frozen sections (10μm thick) were prepared from a freshly dissected sixth abdominal ganglion, mounted on subbed slides and air-dried, following procedures detailed elsewhere (Schmidt-Nielsen et al. 1977). After fixation of sections for 30 min at 0 °C in 4% paraformaldehyde, 0·14M sucrose, 0·075 M sodium phosphate buffer (pH 7·4) the slides were washed for 5 min in 0·14M sucrose in 0·075 M phosphate buffer (pH 7·4). Following a brief rinse (30 s) in distilled water, the slides were air-dried. Sections were incubated for 10—20 min at 21 °C in a humidified chamber with ¹²⁵1-α-bungarotoxin (5·0×1O⁻⁸m toxin; specific activity 70Ci mmol⁻¹) in wash saline of the following composition (in HIM): NaCl, 120; K₂HPO₄, 1-8; KH₂PO₄,0·2; CaCl₂, 1·8; Tris-HCl, 40·1; Tris-base, 4·8 (pH7·4) containing, in addition, 0·1% bovine serum albumin. Unbound toxin was removed by washing in three changes of this saline and sections were postftxed for 20 min at 0 °C in 2% glutaraldehyde in 1·0M phosphate buffer (pH 7·4). In some experiments sections were pre-treated with J-tubocurarine chloride (1·0X10⁻³M) or nicotine hydrogen(+)-tartrate (1·0 X 10⁻³ M), for 15 min prior to incubation with ^12sI-α-bun-garotoxin.

Finally, slides were washed with three changes (1 min each) of distilled water and coated with Kodak NTB 2 emulsion diluted 1:1 with distilled water. After 4—12 days autoradiograms were developed and the sections were dehydrated (by passing the slides through an ethanol series), given three changes in xylene (2min each) and mounted in Permount.

Although axonal position noted during dissection proved a fairly reliable indicator that the cell under test was GI 2, this could only be confirmed on completion of a pharmacological experiment by backfilling the isolated axon with cobaltous chloride, using a procedure described earlier (Harrow et al. 1980).

The axon of G12 was dissected from connectives linking the fourth and fifth abdominal ganglia. Using the experimental chamber and recording techniques described by Pelhate, Hue & Chanelet (1978), oil-gap, single-fibre recordings of action potentials were obtained from the superfused isolated axon under current-clamp conditions at 12°C. Physiological saline of composition (mm): NaCl, 210; KCl, 3·1; CaCl₂, 5·4; MgCl₂, 5·0; NaHCCh, 2·0; Na₂HPO₄,0·1; pH 7·2 was employed.

A preparation comprising the cerci, the cereal nerves and abdominal ganglia was excised and transferred to a drop of saline on the stage of a binocular microscope. Using fine stainless-steel needles, the sixth abdominal ganglion was desheathed to facilitate access to the neuropile of bath-applied pharmacological agents. One of the connectives linking the fifth and sixth abdominal ganglia was desheathed and the axon of GI 2 was isolated. All other axons in the connective were cut and the preparation was transferred to a Perspex experimental chamber suitable for oil-gap, single-fibre recordings. Details of the chamber have been described elsewhere (Pichon & Callec, 1970; Callec, 1974). Finally, the intact connective was severed leaving only one isolated axon (bathed in paraffin oil) linking the fifth and sixth abdominal ganglia. The desheathed sixth abdominal ganglion was continuously superfused (1 ml min⁻¹) with a physiological saline of composition as described above for recording action potentials, but without MgCl₂. These experiments were performed at room temperature (21-23 °C).

Silver-silver chloride electrodes were connected to a WP Instruments model M701 DC amplifier. Excitatory postsynaptic potentials (EPSPs) of GI 2 were recorded using a Tektronix 545 oscilloscope, and changes in resting potential were recorded on a Tekman TE220 pen recorder. EPSPs were evoked by mechanical stimulation of a single mechanoreceptor (long bristle hair - Nicklaus, 1965) on the cercus contralateral to the connective containing the axon of GI 2, using a fine needle attached to the cone of a loudspeaker which was driven by a Farnell pulse-generator. Compound EPSPs were evoked by direct electrical stimulation of many cereal afferents using silver hook electrodes applied to the contralateral cereal nerve XI (nomenclature of Roeder, Tozian & Weiant, 1960). Action potentials were elicited by applying current pulses to the postsynaptic giant axon. Unless otherwise stated, stimulus intervals of 5 s were used throughout pharmacological experiments. A detailed description of the electrical circuitry has already been published (Hue, Pelhate, Callec & Chanelet, 1978).

When 20pg (2·5 × 10⁻⁹mol) of α-bungarotoxin was subjected to SDS-polyacryl-amide gel electrophoresis a single band of molecular weight 8000 was found. No trace contaminants were detected (Fig. 1A).

Sections of the sixth abdominal ganglion (N=4) incubated with ¹²⁵I-α-bun-garotoxin revealed two separate regions of binding. In the neuropile region, two distinct patches of specific binding were noted on either side of the midline (Fig. IB). Pretreatment with either d-tubocurarine (Fig. 1C) or nicotine at a concentration of 1·0 × 10⁻³M completely removed these dense patches of binding in the neuropile, indicating that all the binding in this region is specific. This region of neuropile roughly corresponds with the location of the cereal afferent, GI 2 synapses which are the subject of electrophysiological experiments in this paper.

In addition, dense binding was also observed in the peripheral regions of the ganglion. The distribution of binding in the periphery was not restricted to cell bodies but extended throughout regions occupied by glial cells (Fig. IB). Only part of this peripheral binding was removed by pretreatment with 1·0 × 10⁻³M d-tubocurarine (Fig. 1C). Thus, both specific and non-specific binding components were detected in these peripheral regions of the ganglion.

Action potentials recorded under current-clamp conditions from the axon of G12 (N = 5) were unaffected by perfusion of α-bungarotoxin(1·0 X 10−⁵M) for 60 min (Fig. 2). The single axon backfill procedure (Harrow et al. 1980) allowed identification of G12 upon completion of electrophysiological experiments. Each giant interneurone has a unique morphology in the terminal abdominal ganglion (Daley, Vardi, Appignani & Camhi, 1981). Soma position, neurite shape and positions of major dendritic branches confirmed that the cell under test was G12. The comparison of Fig. 3A and B shows that the act of desheathing the ganglion does not disrupt substantially the normal morphology of G12.

Excitatory postsynaptic potentials (EPSPs) were recorded by the oil-gap method from GI2. When α-bungarotoxin(1·0×10⁻⁷M) was bath perfused over the desheathed sixth abdominal ganglion, the EPSP recorded in response to the deflection of a long bristle hair on the lateral, ventral surface of the cercus, progressively reduced in amplitude. Synaptic transmission was finally blocked about 90 min after the application of this concentration of α-bungarotoxin(Fig. 4). No significant change in membrane potential of the giant interneurone was detected during synaptic blockade. The compound EPSP recorded in the same experiment in response to direct electrical stimulation of cereal nerve XI was also completely blocked by α-bungarotoxin(1·0 × 10⁻⁷M). After 110 min of exposure to toxin, increasing the stimulus intensity in order to recruit further afferents failed to restore synaptic transmission, indicating that all excitatory afferent input to the giant interneurone via cereal nerve XI was blocked. Superfusing the ganglion with normal saline for 150 min following synaptic block by α-bungarotoxindid not result in any recovery of synaptic transmission. In the same experiment, action potentials elicited by direct stimulation of the portion of the giant axon in the sixth abdominal ganglion were unaffected by exposure to (1·0 × 10⁻⁷M) α-bungarotoxin (Fig. 4). In three of the eight experiments in which α-bungarotoxinwas tested at this concentration, blockade of synaptic transmission was preceded by a slight (10-20%) increase in amplitude of the compound EPSP. Control experiments showed no attenuation of EPSPs recorded by the oil-gap method from cereal afferent, GI 2 synapses over a period of 8-12 h.

The time required to block synaptic transmission completely between cereal afferents and GI 2 was dependent upon the concentration of α-bungarotoxinperfused over the desheathed sixth abdominal ganglion. The results of several experiments of this type are summarized in Fig. 5. Exposure to concentrations of α-bungarotoxinas low as 1·0×10⁻⁹ M was effective in blocking transmission provided that the toxin was applied for ∼200min.

For any given toxin concentration, the time required for blockade is dependent on the frequency of stimulation. For example, the time required for 5·0 x 10⁻⁸ M a-bungarotoxin to reduce the amplitude of the EPSP by 50% is 75 ± 35 min at a stimulation frequency of 0·2 Hz (Table 1). However, stimulation at a higher frequency of 1 0 Hz increases the time for 50% blockade by a factor of 2·4 (Table 1). We propose that increased frequency of stimulation leads to higher synaptic concentrations of the neurotransmitter candidate, acetylcholine, which competes with the α-bungarotoxinfor the postsynaptic receptor.

To test this hypothesis, we first investigated the effect of an acetylcholinesterase inhibitor (eserine) on time to 50% blockade by α-bungarotoxin. Simultaneous exposure of the desheathed ganglion to 7·5 X 10⁻⁸M eserine and 5·0 X 10⁻⁸M a-bungarotoxin increased the time to 50% blockade of the compound EPSP evoked by electrical stimulation (0·2 Hz) of the cereal nerve XI (Table 1). In a second set of experiments, a potassium-channel blocking agent, 4-aminopyridine (Pelhate & Pichon, 1974; Meves & Pichon, 1977) was tested for its effect on time to blockade by α-bungarotoxin. Previous studies established that 4-aminopyridine enhances neurotransmitter release at cereal afferent, giant interneurone synapses as a result of the increased duration of the presynaptic action potential (Hue et al. 1976). Exposure of the desheathed ganglion to 5·0 × 10⁻⁵M 4-aminopyridine and 5·0 × 10⁻⁸M a-bungarotoxin prolonged the time to 50% blockade at a stimulation frequency of 0·2 Hz (Table 1).

A stock solution of 0·1 M quinuclidinyl benzilate was prepared in ethanol and diluted with physiological saline to yield various concentrations of the ligand, always with 0·01% or less ethanol present. Control experiments showed that exposure of desheathed sixth abdominal ganglia to this concentration of ethanol for 300 min did not modify either the resting potential or the compound EPSP recorded from G12 following electrical stimulation of nerve XI (at a frequency of 0·2 Hz). At a concentration of 1·0 × 10⁻⁶M, quinuclidinyl benzilate (N = 5) did not affect synaptic transmission between cereal afferents and GI2 (Fig. 6). Quinuclidinyl benzilate at a concentration of 1·0 × 10⁻⁵ M resulted in partial block of transmission. The sample of quinuclidinyl benzilate used in these studies was tested for its ability to inhibit the binding of [³H]quinuclidinyl benzilate to rat brain extracts (J. M. Young & D. B. Sattelle, Unpublished observations). A K_i of 2·0 × 10⁻¹¹M was determined from the binding data, confirming that the sample of quinuclidinyl benzilate used in the present studies was highly effective at rat brain muscarinic cholinergic receptors.

A specific, saturable ¹²⁵1-α-bungarotoxinbinding component, with the pharmacological properties of an acetylcholine receptor has been described in membrane extracts of the following insect preparations: fly heads of Drosophila melanogaster (Schmidt-Nielsen et al. 1977; Hall, 1980; Dudai & Amsterdam, 1977; Dudai, 1977, 1978, 1980; Rudloff, Jimenez & Bartels, 1980) and Musca domestica (Dudai, 1977; Harris et al. 1979; Cattell, Harris & Donnellan, 1980; Jones, Sudershan & O’Brien, 1981), brain tissue of the moth Manduca sexta (Sanes, Prescott & Hildebrand, 1977; Hildebrand, 1980), cerebral ganglia of the locust Locusta migratoria (Breer, 1981) and in abdominal nerve cords of the cockroach Periplaneta americana (Gepner, Hall & Sattelle, 1978). These preparations have a high concentration of α-bungarotoxinbinding sites. For example, there are 300pmol binding activity per gram of antennal lobe tissue in brains of Manduca (Sanes et al. 1977), 112 pmol binding activity per gram of abdominal nerve cord in Periplaneta (D. B. Sattelle, J. I. Gepner & L. M. Hall, unpublished observations) and 88 pmol binding activity per gram of Drosophila heads (Schmidt-Nielsen ci al. 1977). For the purified receptor-toxin complex of Drosophila, a dissociation constant (Kb) of 1·5 × 10⁻¹⁰M has been calculated using association and dissociation rate constants (Gepner, 1979; Hall, 1980). In order to show that the α-bungarotoxinbinding component of insect central nervous tissues is a constituent of a functional acetylcholine receptor, it is necessary to demonstrate that ¹²⁵I-α-bungarotoxinbinding is present in neuropilar regions known to contain cholinergic synapses and that α-bungarotoxinblocks synaptic transmission at a concentration close to the calculated K_D.

In the current investigation, autoradiographic localization of ¹²⁵I-α-bungarotoxinbinding in the sixth abdominal ganglion reveals binding in the neuropile on either side of the midline. This binding is predominantly situated towards the posterior half of the ganglion, corresponding to regions in which the dendritic fields of giant interneurones are located. The bulk of the sensory neurones that enter the sixth abdominal ganglion via cereal nerve XI terminate in this same region of the neuropile (Daley et al. 1981). Although proof that ¹²⁵I-α-bungarotoxinis binding to synaptic membranes awaits an ultrastructural localization study, the present experiments demonstrate specific binding of ¹²⁵1-α-bungarotoxinin the neuropile of the cockroach sixth abdominal ganglion.

The dense binding of ¹²⁵1-α-bungarotoxinnoted in the peripheral regions of the ganglion, part of which is non-specific, contrasts with the previously reported absence of binding in the peripheral regions of ganglia in Drosophila melanogaster (Hall & Teng, 1975; Schmidt-Nielsen et al. 1977) and Manduca sexta (Hildebrand, Hall & Osmond, 1979) but is consistent with findings for other insects such as the cricket Acheta domesticus (L. M. Hall, B. C. Osmond & J. G. Hildebrand, unpublished observations). Part of the specific binding in peripheral regions of the cockroach terminal abdominal ganglion is probably due to α-bungarotoxin-sensitive cholinergic receptors on the somal membranes of neurones (Harrow & Sattelle, 1983). The possibility that α-bungarotoxinis binding to glial cells cannot be excluded. Further experiments are needed to characterize the dense toxin-binding in the peripheral regions of the ganglion.

It has been shown in this study that the acetylcholine receptor probe a-bungarotoxin at a concentration of 1·0 × 10⁻⁹M blocks synaptic transmission between cereal afferents and GI2 in the sixth abdominal ganglion of the cockroach. At micromolar concentrations, however, this toxin has no effect on the action potentials. These electrophysiological studies indicate that the primary site of action of a-bungarotoxin on GI 2 is synaptic. Earlier investigations have demonstrated that α-bungarotoxin(1·0 × 10−⁸M) blocks cereal afferent input to GI 3 in the same ganglion (Harrow et al. 1979). Using the metathoracic ganglion of the same insect, Carr & Fourtner (1980) have shown that 1·0 × 10⁻⁷M α-bungarotoxinblocks synapses between trochanteral hairplate afferents and a leg motoneurone (D₈). There is considerable pharmacological evidence that acetylcholine is the likely transmitter molecule at all these synapses (Shankland, et al. 1971; Callec, 1974; Sattelle, McClay, Dowson & Callec, 1976; Sattelle, 1978; Carr & Fourtner, 1980).

Several observations in the present investigation indicate that an antagonist action at an acetylcholine receptor is the basis of the synaptic block induced by α-bungarotoxin. a-Bungarotoxin blocks cereal afferent input to GI 2 at concentrations as low as 1·0 × 10⁻⁹M. This value is close to the KD of 1·5 × 10⁻¹⁰M calculated for the purified receptor-α-bungarotoxin complex of the fruit fly Drosophila melanogaster (Gepner, 1979; Hall, 1980), indicating that the blocking action is likely to be a specific action of the toxin.

In addition, conditions which increase synaptic concentrations of acetylcholine or prolong its life time are found to interfere with the time course of blocking action of α-bungarotoxin. The simplest interpretation of these results is that acetylcholine competes with α-bungarotoxin for the postsynaptic receptor.

Recently, in studies on the extrasynaptic cell body membrane of GI 2, the depolarizing response to ionophoretically-applied acetylcholine was blocked by α-bungarotoxin(Harrow & Sattelle, 1983). Since the cell body membranes of neurones in the cockroach sixth abdominal ganglion appear to be devoid of synapses (Smith & Treherne, 1965), a presynaptic action of the toxin appears to be ruled out in this case. The effects of α-bungarotoxin at cereal afferent, GI 2 synapses can most simply be explained, therefore, by a competitive antagonism of a postsynaptic cholinergic receptor.

Our studies reveal that a substantial length of time is required for α-bungarotoxin blockade in this cockroach preparation. Kinetic analysis of ¹²⁵I-α-bungarotoxin binding to membrane homogenates from cockroach nerve cords has shown that the toxin has a reasonably fast rate constant of association (7·8 × M-¹S⁻¹), (D. B. Sattelle, J. I. Gepner&L. M. Hall, unpublished observations). This is the same order of magnitude as rate constants for the association of toxin with the acetylcholine receptor at neuromuscular junctions (Colquhoun & Rang, 1976; Barnard, Coates, Dolly & Mallick, 1977). The observed rates would predict half-times for the association rate of the order of 1 min at a toxin concentration of 1·0 × 10⁻⁷ M, and yet in electrophysiological studies of the intact but relatively accessible end-plate of vertebrate muscle the observed times are 10-30 min. These slow times are generally attributed to local diffusion barriers (Barnard, 1977; Dolly et al. 1977). Even greater diffusion barriers would be expected for synapses located deep within the neuropile of a desheathed but relatively intact cockroach ganglion. Thus, we feel diffusion barriers are likely to be a major factor contributing to the long times for toxin blockade (cf. discussion by Barnard following Harrow, David & Sattelle, 1982).

A comparison between electrophysiological and ligand binding studies for various vertebrate and invertebrate neuronal tissues in which nicotinic cholinergic synapses are present indicate that the ability to bind radiolabelled toxin is not always an indication of the effectiveness of this receptor probe as a synaptic blocking agent (Table 2). This confirms the importance of parallel studies using both techniques in any characterization of receptor properties.

Specific binding of [³H] quinuclidinyl benzilate has been observed in particulate extracts of fly heads of both Drosophila melanogaster (Dudai & Ben-Barak, 1977; Haim, Nahum & Dudai, 1979) and Musca domestica (Jones & Sumikawa, 1981), the terminal abdominal ganglion of the cricket Acheta domesticus (Meyer & Edwards, 1980), and cerebral ganglia of the locust Locusta migratoria (Breer, 1981). Studies on insect head homogenates reveal that the concentration of [³H] quinuclidinyl benzilate binding sites is lower than that of ¹²⁵I-α-bungarotoxin binding sites by a factor of five in the case of Drosophila (Dudai, 1980) and by a factor of 20 in the case of both Musca (Jones & Sumikawa, 1981; Jones et al. 1981) and Locusta (Breer, 1981). A KD for the [³H]quinuclidinyl benzilate-receptor complex of Drosophila (1·5 × 10⁻¹⁰M) has been calculated from association and dissociation rate constants (Haim et al. 1979). It is to be expected that concentrations of quinuclidinyl benzilate close to the KD would produce blockade of synaptic transmission if postsynaptic receptors sensitive to this ligand predominate at the chemical synapse under investigation.

In this study we have found that a concentration of 1·0 × 10⁻⁶M quinuclidinyl benzilate is ineffective at cereal afferent, GI 2 synapses of the cockroach. Although partial synaptic blockade is detected at 1·0 × 10^−sM and higher concentrations of quinuclidinyl benzilate, this effect is only observed at concentrations ∼ 100 000-fold greater than the estimated KD for [³H]quinuclidinyl benzilate binding to insect preparations. Meyer & Edwards (1980) detected a specific [³H]quinuclidinyl benzilate binding component in extracts of the terminal abdominal ganglion of the cricket Acheta domesticus. These authors suggested a possible role for a muscarinic receptor at cereal afferent, giant interneurone synapses in the cricket. The present electrophysiological studies in the cockroach reveal a much greater potency of the nicotinic receptor probe, α-bungarotoxin, compared to the muscarinic probe, quinuclidinyl benzilate, at cereal afferent, GI 2 synapses. This apparent discrepancy may simply represent a difference between species. Another possibility is that [³H]quinuclidinyl benzilate binds to other neuronal elements in the cricket terminal abdominal ganglion. A further possibility is that although the experiments reported here indicate that α-bungarotoxin blocks all the postsynaptic cholinergic receptors, it is still possible that a muscarinic receptor exists on the presynaptic membrane. It is of interest to note, therefore, the observation by Breer (1982), that muscarinic agents modulate potassium-induced acetylcholine release from synaptosomes prepared from the locust. Further experiments are needed to test these ideas.

The experiments described in this paper further characterize the α-bungarotoxin binding component in the cockroach nerve cord (Gepner et al. 1978). At least a portion of this binding component represents cholinergic receptors which have a postsynaptic function at synapses between cereal sensory neurones and GI 2.

The authors wish to thank Mrs B. C. Osmond for assistance in the preparation of frozen sections and Dr J. F. Donnellan (Shell Bioscience Laboratories, Sittingbourne, U.K.) for the generous gift of quinuclidinyl benzilate. DBS. acknowledges the support of a Ciba Foundation Anglo-French Exchange Bursary. IDH. was in receipt of an SRC-CASE studentship in collaboration with Shell Research Ltd, U.K. LMH. acknowledges the support of the Council for Tobacco Research, U.S.A., Inc. and the U.S. National Science Foundation.