The Identification of Motor and Unpaired Median Neurones Innervating the Locust Oviduct

Like many other visceral muscles, the locust oviducts generate myogenic contractions. These contractions are under neurohormonal control and serve to propel the eggs to the ovipositor during egg laying (Girardie and Lafon-Casal, 1972; Miller, 1975; Cook et al. 1984). Locust oviducts, however, also receive a nervous supply from a pair of nerves arising from the seventh abdominal ganglion (Okelo, 1971; Lange and Orchard, 1984a). This nervous supply produces rhythmical contractions of the oviduct and apparently serves to retain eggs in the lateral oviducts (Lange et al. 1984a,b). The exact nature of this innervation, i.e. number, type and location of neurones, however, has not yet been fully resolved. According to Kiss et al. (1984), the oviducts of Locusta migratoria receive polyneuronal innervation from 3–5 paired and 9–11 unpaired neurones, located at the seventh abdominal ganglion. Lange and Orchard (1984a), however, describe only three paired and two unpaired neurones. Similar studies on the cockroach Periplaneta americana reveal the presence of 9 paired and 16 unpaired neurones in the terminal abdominal ganglion, projecting to the oviduct and the sphincter muscle (Stoya et al. 1989). In the moth Manduca sexta 4–6 paired and 8–0 unpaired neurones were found to innervate the oviducts (Thorn and Truman, 1989).

It also remains to be shown whether different regions of the oviduct are served by different sets of neurones. This is likely to be the case, as it is to be expected that the lateral oviduct, the common oviduct and their junctional area will exhibit different modes of contraction during activities such as egg propulsion, extrusion and retention. In this respect, modulatory neurones, such as the class of putatively octopaminergic dorsal unpaired median (DUM) neurones, may play a key role in differentially shaping the contractile properties of specific oviductal regions. Regarding the number of such DUM neurones in genital ganglia, sexual dimorphism is reported in several insect species (Giebultowicz and Truman, 1984; Yamagushi et al. 1985; Thorn and Truman, 1989). In the locust, however, the actual number of DUM neurones in the genital ganglion is the same in both sexes (P.-A. Stevenson and H.-J. Pflüger, unpublished results), but there are clear differences in their peripheral projections (Pflüger and Watson, 1988). Whereas in the female six DUM neurones send their axons through the sternal nerve to supply the oviducts, only one DUM neurone projects through the corresponding nerve of the male. The five other DUM cells of the male descend to the terminal ganglion to exit via the genital nerve, which is lacking in the female (Pflüger and Watson, 1988). Two of the oviductal DUM neurones have been shown to contain octopamine (DUMOV1 and 2, Orchard and Lange, 1985). Their stimulation raises the cyclic AMP level in the oviduct, as does also the application of octopamine (Lange and Orchard, 1986a). Octopamine appears to modulate the contractile properties of the oviduct, inhibiting both neurogenic and myogenic contractions in both locusts and cockroaches (Orchard and Lange, 1985, 1987). This contrasts with its effects on skeletal muscle, where it increases twitch amplitude and relaxation rate (Evans and O’Shea, 1978).

In this paper we employ both retrograde and anterograde cobalt stainings of branches of the oviductal nerve to determine the exact number, type and location of neurones serving the oviduct as a whole, as well as its various parts. We thus show that the branches of the motor neurones and the DUM neurones are unequally distributed on the different oviductal regions. Each individual motor neurone was identified and characterised both physiologically and anatomically by simultaneous intracellular and extracellular recording from the oviductal nerve and subsequent intracellular cobalt injection. Similarly, cobalt and differential cobalt/nickel backfilling, coupled with intracellular staining methods, revealed the number, distribution and intraganglionic anatomy of the median neurones with bilateral axons. The inhibitory effects of these neurones upon the oviductal contractions were confirmed by intracellular stimulation and tension measurements.

All experiments were performed on adult female locusts, Schistocerca gregaria (Forskål) and Locusta migratoria (L.). Animals were mounted ventral side up and a mid-ventral incision was made in the posterior abdominal segments, exposing the oviducts and the penultimate ganglion. The oviducts, their associated muscles, and the seventh and terminal (neuromeres 8–11) abdominal ganglia were removed and placed in a translucent Sylgard-filled Petri dish. For central backfills, the cut end of a branch of the oviductal nerve was placed in a Vaseline pool. However, as certain areas of the oviduct are innervated by numerous fine branches and their isolation in the pool proved to be very difficult, defined parts of the oviduct were cut and placed in the Vaseline pool, leaving the oviductal nerve intact. Vaseline pools were filled with distilled water, which was replaced after 5 min by a solution of 1.5–6% hexammine cobaltic chloride (Brogan and Pitman, 1981). The preparation was left at 4°C for 24–48 h, to allow passive diffusion of cobalt ions. Cobalt ions were then precipitated in 1% ammonium sulphide solution and the ganglia were fixed in 4% neutral formalin, dehydrated and cleared in methyl salicylate. In most preparations, cobalt stains were subsequently silver-intensified (Bacon and Altman, 1977). To reveal the course of the oviductal nerve and its numerous fine branches supplying different parts of the oviduct, peripheral cobalt fills were also made. These preparations were treated as above and also silver-intensified. Our results are based on at least 20 successful preparations.

For intracellular recordings and stainings of single cells, the seventh abdominal ganglion was exposed and stabilised on a wax-covered platform with fine pins. The ganglionic sheath was treated with 1% Protease (Sigma XIV) for 2 min to ease electrode penetration. Cell bodies of oviductal neurones were impaled with glass microelectrodes filled with 6% hexammine cobaltic chloride (resistance 70–100 MΩ). Signals were stored on magnetic tape and plotted on a chart recorder (Gould ES1000) for later analysis. After physiological recordings, cells were intracellularly filled with cobalt ions by passing depolarising pulses of 10 nA amplitude and 500 ms duration, at a repetition rate of 1 Hz for up to 30 min, via the recording electrode. The preparation was left for 20min in saline (Usherwood and Grundfest, 1965) to allow diffusion of the dye. Ganglia were then dissected out and treated as described above, fixed in 4% formalin and silver-intensified. All stained preparations were drawn with a camera lucida attached to a Zeiss Axiophot microscope and photographed.

Extracellular nerve recordings were obtained using monopolar hook electrodes. The contractions of the oviducts were recorded either in vivo or in vitro by attaching the anterior end of the lateral oviduct to a tension transducer; the other end of the oviduct was fixed with minuten pins.

The oviduct is innervated by branch N2B (also called the oviductal nerve, OVNv) of the sternal nerve (N2) of the seventh abdominal ganglion (AG7, see Fig. 1A,B). Upon reaching the oviduct, N2B bifurcates into branch Bl, which supplies the posterior part of the lateral oviduct (LO, see Fig. IB), and branch B2, which gives rise to branches B2a and B2b (Fig. 1A,B). Nerve B2a supplies the junctional area of the lateral oviduct and the common oviduct, muscle M257–258, as well as the sides of the common oviduct (CO) with numerous fine branches. Nerve branch B2b supplies muscles M257–258, M234 and M250 (Kiss et al. 1984). The finer branches of Bl and B2a and the anastomoses between them create a dense plexus on the oviduct. Selective backfill experiments of the three oviductal regions shown in the inset of Fig. 2 (regions A,B,C) showed that each receives a different pattern of innervation (Fig. 2A,B,C and Table 1).

The lateral oviduct (region A) innervated by branch Bl, is supplied by one motor neurone and 16–20 median neurones with bilateral axons, all within AG7 (Fig. 2Ai,ii). The cell body of the motor neurone has a diameter of 25 μm and lies ventrolaterally near the origin of the sternal root. This neurone was later identified as oviductal neurone 2 (OVN2, marked with arrow in Fig. 2Ai,ii).

The median neurones with bilateral axons are arranged in two groups, an anterior group with 8 cell bodies and a posterior group with 12 cell bodies. The majority of the anterior neurones (6 out of 8 cells, diameter 25–40 μm) are situated dorsally, near the origin of the anterior connectives of AG7 (Fig. 2Ai, anterior cells). The other two cells of the anterior group (diameter 30 μm) lie in a ventromedial position (Fig. 2Aii, anterior cells). Their primary neurites (diameter 2 μm) run posteriorly towards the centre of the ganglion, where they bifurcate, each of them giving rise to a pair of secondary neurites, which leave the ganglion through the sternal nerves.

The posterior neurones with bilateral axons are divided into three subgroups. The first group consists of the two largest (cell bodies 55–60 μm in diameter), located dorsally on either side of the origin of the median nerve (DUMOV1,2 according to Lange and Orchard, 1984a, black posterior cells in Fig. 2Ai). Their primary neurites run as a bundle anteriorly and both bifurcate into two secondary neurites (T-junction marked with 1 in Fig. 2Ai). The second group consists of three cells with relatively large cell body diameters (50 μm), which lie beneath the cells of the first group, but still dorsally (Fig. 2Ai, stippled posterior cells). The third posterior group consists of seven cells, which are located ventromedially and possess much smaller cell bodies (Fig. 2Aii, posterior cells). The primary neurites of the second and third groups project anteriorly into two separate bundles and bifurcate more posteriorly than the primary neurites of the cells of the first group (T-junction of second group marked with 2 in Fig. 2Ai; the T-junction of third group is marked 3 in Fig. 2Aii).

The junctional area between the lateral and the common oviduct (region B) is supplied through branch B2a by three motor neurones, later identified as OVNI, OVN2 and OVN3, and two median neurones with bilateral axons (Fig. 2B). The cell bodies of the three motor neurones (diameter 25–30 μm) lie ventrolaterally in AG7. The cell bodies of OVNI and OVN2 lie posteriorly near the origin of the sternal root, while the cell body of OVN3 lies more anteriorly (Lange and Orchard, 1984a). Their central arborisations extend to cover both the ipsilateral and contralateral areas of the ganglion. In 30% of the preparations OVN3 was not stained, possibly because it innervates a relatively small part of the oviduct. The two median neurones are located dorsally, near the origin of the median nerve. They are clearly the same neurones as those previously described as forming the first posterior group innervating the lateral oviduct (DUMOV1 and DUMOV2).

The common oviduct (region C) is supplied through fine branches of nerve B2a by two motor neurones, OVNI and OVN2, and two median neurones with bilateral axons (Fig. 2C). The motor neurones have cell bodies located ventrola-terally and possessing mainly ipsilateral arborisations. The median neurones, to judge from their very dorsal position and the size of their cell bodies, are probably DUMOV1 and DUMOV2, which also project to the lateral oviduct and the junctional area (regions A and B).

Intracellularly impaled neurones were identified by a 1:1 correspondence between each intracellularly recorded soma spike and an action potential recorded extracellularly from the oviductal nerve, at a constant latency. Prolonged intracellular cobalt injection of up to 1 h made it possible to follow the stained axon of the impaled cell to the oviduct. Each oviductal motor neurone (OVN) was stained at least four times.

Extracellular records from the oviductal nerve reveal three large action potentials of different amplitude, which are produced by the three motor neurones (OVNI, OVN2, OVN3 in Fig. 3D) and some smaller potentials, which are probably produced by the oviductal median neurones with bilateral axons. The oviductal motor neurones were identified both physiologically (Fig. 3) and morphologically (Fig. 4). All oviductal motor neurones showed attenuated soma spikes, suggesting that the spike-initiating zone is distant from the soma.

Simultaneous intracellular records from the cell body of OVNI and extracellular records from the oviductal nerve (OVNv in Fig. 3) showed that OVNI produces the smallest extracellular action potential (Fig. 3A,D). After intracellular cobalt injection, the branching pattern of OVNI within AG7 was revealed (see OVNI in Fig. 4). Its cell body (diameter 25 μm) lies ventrolaterally, near the origin of the sternal nerve. The dendrites of OVNI extend mainly in the dorsal area of the ipsilateral half-ganglion, but some fibres cross the midline. A single branch runs to the ventral area.

This neurone produces the action potentials with intermediate amplitude recorded from the oviductal nerve (Fig. 3B,D). Its cell body has a diameter of 25 μm and its dendrites arborise both ipsi-and contralater-ally (OVN2 in Fig. 4). Although its dendritic field is less dense than those of OVNI and 0VN3, this neurone has, in addition to dorsal arborisations, projections in the medial and ventral areas of the neuropile. Prolonged intracellular dye injection showed that this cell supplies all innervated areas of the oviducts through branches Bl and B2a (see Fig. 1B).

This neurone produces the largest action potentials recorded extracellularly from the oviductal nerve (Fig. 3C,D). The motor neurone was injected with cobalt to reveal its intraganglionic morphology (OVN3 in Fig. 4). The cell body of OVN3 (diameter 30 μm) lies ventrally and slightly anterior to the origin of the sternal nerve. Its primary neurite runs towards the midline and, about 50 μm before reaching it (see arrowhead in Fig. 4), gives rise to a thick profile (10 urn in diameter), termed the giant contralateral process (GCP, Lange and Orchard, 1984a). Secondary neurites originate from the GCP and their dense arborisations cover the contralateral half of the ganglion exclusively. Its axon originates from the neurite in the ipsilateral half-ganglion (see arrowhead in Fig. 4) and exits through the sternal nerve. Prolonged intracellular dye injection showed that OVN3 supplies the oviducts only through branch B2a (see Fig. 1B).

All experiments were performed in vitro, by isolating the oviducts together with the seventh and terminal abdominal ganglia. Intracellular records from the cell bodies of large posterior median neurones showed that they generate large overshooting action potentials (60–80 mV) with a marked undershoot (6–15 mV, see upper traces in Fig. 5). The action potentials of anterior median neurones are of similar shape but their amplitudes are generally smaller (40–50 mV) than those of their posterior counterparts.

Intracellular records from the cell body of a posterior median neurone (DUM, in Fig. 5) and extracellular records from both the right (rOVNv) and the left (lOVNv) oviductal nerve, revealed a 1:1 correlation of the soma action potential and pairs of extracellular potentials (marked with dots in Fig. 5). Very often, one of the two extracellular spikes was recorded before the intracellular potential (see lOVNv in Fig. 5A). Since it has been shown that up to four spike-initiating zones (DUMETi, Heitler and Goodman, 1978) exist in a single DUM neurone, the extracellular electrode was probably closer to one of the axonal spike-initiation sites than the intracellular electrode in the cell body. Following physiological characterisation, each impaled neurone was intracellularly stained to reveal its intraganglionic morphology (Fig. 6). The anterior (Fig. 6A) and large posterior (Fig. 6B,C) median neurones show considerable differences in their branching patterns in the seventh abdominal ganglion.

There are two anterior median neurones (Fig. 6Ai,Aii) with different branching patterns in the ganglion. They both belong to the dorsal group of anterior cells described previously (see Fig. 2Ai) and their cell bodies have a diameter of 40 μm. Their primary neurites travel posteriorly and, before reaching the centre of the ganglion, divide into two lateral neurites. The neuropilar branches of the neurone shown in Fig. 6Ai originate exclusively from the lateral parts of its pair of secondary neurites and from there they extend to cover the whole of the dorsal area of the neuropile. Most of the neuropilar branches of the neurone shown in Fig. 6Aii also originate from the more lateral parts of its lateral neurites and they ramify mainly within the lateral parts of the neuropile. In contrast to the neurone shown in Fig. 6Ai, a few branches also originate from its primary neurite. The primary and lateral neurites of both cells have a diameter of 2 μm, which is considerably smaller than that of the neurites of the posterior median neurones.

The cell bodies of the five posterior neurones of groups 1 and 2 are larger (diameter 50–60 μm) than those of their anterior counterparts and their neurites are also thicker (4–5 μm) (Fig. 6B). In contrast to the anterior median neurones, they possess three distinct dendritic fields; a central field originating from the primary neurite, close to the T-junction (this field is absent or very limited in the anterior neurones), and a pair of lateral fields made up of second-order and higher-order neurites that originate from the distal parts of their lateral neurites.

These neurites extend both anteriorly and posteriorly within AG7 (Fig. 6Bi). The neurone shown in Fig. 6Bi is one of the two large neurones of the first group (previously identified as DUMOV1,2). The neurone shown in Fig. 6Bii belongs to the second group of dorsal posterior cells (see Fig. 2Ai) and its cell body is located dorsomedially between the posterior connectives. One posterior neurone (possibly from the second group) was found to possess a unilateral projection (see arrowhead in Fig. 6Biii), which descends through the posterior connective and, after ramifying sparsely within the terminal abdominal ganglion, exits through the sternal nerve of the eighth abdominal neuromere.

Two neurones of the third posterior group (see Fig. 2Aii), with large ventrally located cell bodies, were intracellularly stained (Fig. 6Ci,ii). The neurone shown in Fig. 6Ci has a branching pattern similar to that of its dorsal counterparts and sends fine projections through the posterior connectives to the terminal abdominal ganglion (Fig. 6Ci). The neurone shown in Fig. 6Cii possesses a large ventral cell body but its arborisations within AG7 are very limited.

When a female is interrupted during copulation or oviposition, spontaneous rhythmic activity can be recorded from the oviductal nerve. Three representative patterns are shown in Fig. 7, in which three motor units corresponding to the three oviductal motor neurones (OVNI,2,3) can be clearly distinguished. During each cycle these units are active in the sequence OVN3-OVN2-OVN1, with some overlap between the different units (Fig. 7A). Differences from the pattern shown in Fig. 7A, which is the most frequently observed pattern, are due to irregular bursting of OVN3 (bursts of OVN3 are marked in Fig. 7B,C). Nevertheless, it is possible to identify the single units on the basis of their relative spike amplitude and discharge patterns in each cycle. As outlined in the next section, simultaneous intracellular records from the cell bodies of different motor neurones allowed precise identification of the extracellularly recorded action potentials (Fig. 8).

OVNI (Fig. 8A) exhibits a very regular bursting pattern with an intraburst frequency of 40–50Hz, burst duration of 2–2.5 s and burst interval of 3–3.5 s (Fig. 8A). OVN2 fires continuously but in bursts with a spike frequency of about 25 Hz dropping between bursts to about 10 Hz (Fig. 8B). OVN3 is also active in bursts with an intraburst frequency ranging from about 10 to 50 Hz (see Fig. 7B,C). In most cases its burst duration and burst interval are 2–3.5 s (Fig. 7C). In some preparations OVN3 fired with an intraburst frequency of 10–15Hz, but the burst interval was markedly reduced to 0.5 s (Fig. 8C).

Simultaneous intracellular records from the cell bodies of OVN3 and OVN2 during depolarising current injection into either OVN3 (Fig. 9) or OVN2 revealed no indication of coupling between these neurones. This was also true for similar experiments pairing OVN3 with OVNI or OVN2 with OVNI.

The mechanical events occurring in the posterior region of the lateral oviduct and at the junctional area of the lateral and common oviduct were recorded in vivo using a tension transducer (Fig. 10, upper traces). When these measurements were obtained simultaneously with extracellular records of the oviductal rhythm (Fig. 10, lower traces), a correlation was evident between the bursting activity of OVNI (smaller action potential) and the oviductal contractions (the bursts of OVNI are marked in Fig. 10B). This was even more evident when OVNI ceased firing for some milliseconds and caused a reduction in the contraction amplitude (see arrowhead in Fig. 10B).

To examine the influence of the bilaterally projecting oviductal median neurones on oviductal contractions the locust oviducts were removed from the animal with only the seventh abdominal ganglion attached. This preparation exhibits rhythmical contractions (Fig. 11A,B,C, lower traces); each contraction lasts for 7-8 s and is repeated at a rate of 0.1 Hz. Injection of depolarising current of increasing strength into a posterior median neurone with bilateral axons (first and second group in Fig. 2Ai) (Fig. 11A–C) revealed a graded effect upon the strength of oviductal contractions. A comparison of the contraction amplitude before (a in Fig. 11A) and immediately after (b in Fig. 11A) the injection of a relatively small current into the cell body of such a posterior median neurone (Fig. 11A, upper trace) showed only a slight reduction in the contraction amplitude (see inset of Fig. 11A). Intracellular excitation of a posterior median neurone with a larger current (Fig. 11B) resulted in a 25% reduction in the contraction amplitude (see inset of Fig. 11 B) and a slight decrease in the oviductal contraction frequency. The initial amplitude of the contraction was re-established10 s after the end of current injection into the neurone. This inhibitory effect was even more evident when a very large current was injected into a posterior median neurone (Fig. 11C), making the cell fire at a frequency of 15–20 Hz. In this case, the contraction amplitude was reduced by more than 50% (see inset of Fig. 11C) and the oviductal rhythm slowed down considerably. Fig. 11D shows the decrease in the contraction amplitude in relation to the posterior median neurone frequency during intracellular current injection.

Both myogenic and neurogenic oviductal contractions participate in the egglaying behaviour of the female locust. These contractions are under both neural and neurohormonal control (Okelo, 1971; Girardie and Lafon-Casal, 1972; Miller, 1975; Lange et al. 1984a,b). During egg laying the eggs are transported to the ovipositor by the myogenic contractions of the oviducts. At inappropriate times, such as during oviposition digging (pre-ovipositional behaviour) or following interruption during egg laying, egg propulsion must be prevented (Lange et al. 1984b). A similar mechanism controlled by the central nervous system (CNS) was found in the stick insect Carausius morosus (Thomas and Mesnier, 1973; Thomas, 1979). In this insect ovipositional activity is nocturnal. During the day an ovulated egg remains within the genital chamber, a membranous structure located at the base of the ovipositor. Sense organs, probably located in the genital chamber wall, detect the presence of the egg and activate the motor neurones that innervate the common oviduct. Activity of these motor neurones contracts the common oviduct and thus prevents the progression of another egg (Thomas, 1979). In contrast, the myogenic contractions of the stick insect oviducts, which serve to propel the eggs, are under neurohormonal control. A similar model of dual control of egg laying has been proposed in the locust (Lange et al. 1984b). In this insect, whereas eggs are propelled by myogenic oviductal contractions, the retention of the eggs in the anterior region of the lateral oviducts is controlled by the rhythmic activation of sets of motor neurones, located in the seventh abdominal ganglion, which innervate the oviducts (oviductal rhythm, Lange et al. 1984b). Their activity causes the junctional area of the oviducts to contract, thus preventing the passage of the eggs to the common oviduct (Lange et al. 1984b). In the present study, extracellular records from the oviductal nerve show that three motor neurones produce the neurogenic oviductal rhythm, confirming previous reports showing a polyneuronal innervation of the oviduct by three motor units (Orchard and Lange, 1986). While no information is available concerning the generation of the oviductal rhythm, recent studies have shown that the central pattern generator (CPG) for oviposition digging in the locust is located within the terminal abdominal ganglion (Thompson, 1986a,b). It was also suggested that the ovipositional CPG is inhibited by neuronal circuits from the brain and the thoracic ganglia, since isolation of the abdomen from the thorax initiates the oviposition digging behaviour and stimulation of the connectives between the metathoracic ganglion and the first free abdominal ganglion inhibits the oviposition digging rhythm (Thompson, 1986a). Since the oviductal and the oviposition digging rhythms are temporarily phase-locked and must, therefore, be coordinated, the two rhythms may be driven by a common centre. In favour of this hypothesis is the observation that in females the ovipositor valves undergo continuous digging movements when the oviductal rhythm is initiated. When the connectives between the seventh ganglion and the terminal abdominal ganglion are transected, the oviductal rhythm is dramatically changed (Kalogianni, 1991).

A pre-requisite for the study of the generation of the oviductal rhythm, as well as its coordination with the oviposition digging rhythm, is the intracellular identification of the oviductal motor neurones. The present study has shown that these neurones (OVNI, OVN2, 0VN3) can be easily distinguished by morphological and physiological criteria. Our experiments show (a) that the junctional area of the oviduct, the contractions of which aid egg retention, is the only region innervated by all three motor neurones and (b) that the lateral oviduct, the myogenic contractions of which cause egg progression, is only innervated by one motor neurone, which is more or less continuously active (OVN2). Of the other two motor neurones, OVNI exhibits a phasic pattern in most preparations. Our results show that the oviductal neurogenic contractions are phase-locked with the bursting activity of OVNI. The tonic nature of the firing of OVN2, the only neurone supplying all innervated areas of the oviduct, suggests that it maintains the muscle tone. OVN3 was very variable in its firing pattern, ranging from complete silence to strong activity. In our experiments we did not observe a correlation between the activity of OVN3 and the oviductal contractions, possibly because in these experiments OVN3 was not firing at a high frequency.

Intracellular dye injection showed that OVN3 possesses a central branching pattern confined exclusively to the contralateral half of the seventh abdominal ganglion. In contrast, the ramifications of OVNI and OVN2 extend mainly within the ipsilateral half of the ganglion. None of the three neurones was found to send any projections to the neighbouring ganglia (AG6, TG).

The results of the present study confirm some previous reports (Kiss et al. 1984) but contradict others (Lange and Orchard, 1984b) by revealing the presence of 16–20 median neurones with bilateral axons located at the seventh abdominal ganglion and projecting to the locust oviduct. A possible explanation for this discrepancy is the unequal distribution of these neurones upon the different oviductal regions. The lateral oviduct receives innervation from all the median neurones (16–20) and one motor neurone, whereas the junctional area receives innervation from only two of the median neurones (previously described as DUMOV1,2, Lange and Orchard, 1984a) and three motor neurones. Similar disparities in the spatial distribution of motor and median neurones were reported for the oviduct of the moth Manduca sexta (Thorn and Truman, 1989). There, the lateral oviduct is supplied by 6 median neurones, the junctional area by 4–6 motor neurones and the common oviduct by 2–4 median neurones.

Intracellular stainings revealed striking morphological differences between the anterior and posterior bilaterally projecting oviductal neurones. Whereas the dorsal posterior median neurones (five neurones of the first and second posterior group) have the typical morphology previously described for abdominal DUM neurones (Pflüger and Watson, 1988), the anterior median neurones with bilateral axons possess only lateral dendritic fields and resemble the so-called bilaterally projecting neurones (BPNs) that innervate the insect heart (Ferber and Pflüger, 1990). The oviductal median neurones with more ventrally located cell bodies of large diameter that were stained in this study possess a branching pattern which does not differ from that of their dorsal counterparts. Previous studies have also shown the existence of such ventral median neurones, possibly innervating the locust oviduct (ventral unpaired median, VUM, neurones, Lange and Orchard, 1986b). Biochemical assays of two identified oviductal DUM neurones (DUMOV1,2, Orchard and Lange, 1985) and of 4-6 VUM neurones (Lange and Orchard, 1986b) have measured their octopamine content (1.27 mmol l⁻¹ and 2.16mmol l⁻¹, respectively). Similar octopamine concentrations were detected in DUMETi, an identified DUM-neurone innervating the locust extensor tibiae muscle (Evans and O’Shea, 1978) and the DUM neurones innervating the lantern tissue in the male firefly (Christensen and Carlson, 1981, 1982; Christensen et al. 1983; Carlson and Jalenak, 1986). Finally, a recent immunohistochemical study of the locust CNS has shown that eight large posterior median neurones of the seventh abdominal ganglion of the female locust express octopamine-like immunoreactivity (Stevenson et al. 1992). Octopamine has been found to enhance the amplitude and the relaxation rate of skeletal muscle contractions. The neurogenic and myogenic contractions of visceral muscles, however, such as those of the cockroach and locust oviduct, are inhibited by octopamine (Orchard and Lange, 1985, 1987). Since intracellular stimulation of the posterior dorsal median neurones reduced the contractions of the locust oviduct in the present study, we propose that these cells are neuromodulators of the oviductal contractions through the secretion of octopamine. Octopamine, however, is not the only substance to which the insect oviducts are sensitive. Proctolin enhances the frequency and amplitude of the spontaneous oviductal contractions in a dose-dependent manner (Stoya and Penzlin, 1988; Orchard and Lange, 1987). Histological studies, using a proctolin antibody, labelled axons in the oviductal nerve and median cell bodies of the seventh abdominal ganglion (Eckert et al. 1989). It has been suggested that, unlike octopamine, which acts as an inhibitory neuromodulator, proctolin acts as an excitatory transmitter in both the locust and cockroach oviduct (Orchard and Lange, 1987). The above data clearly show that a much-needed further step in this study is to characterize the oviductal median neurones with respect to their presumed transmitters. Our data suggest that, from the total of 16–20 median neurones with bilateral axons projecting to the locust oviduct, the dorsal posterior (five neurones) and the large ventral posterior (two neurones) median neurons are true DUM (VUM) neurones, since they are unpaired and support large overshooting soma action potentials with a characteristic hyperpolarising phase, like the DUM neurones described elsewhere (Hoyle and Dagan, 1978; Braunig, 1988; Brookes and Weevers, 1988). The median neurones with smaller cell bodies of the anterior and ventral posterior group are probably BPNs, similar to those recently located in the pregenital (fourth-sixth) abdominal ganglia of the locust, which innervate the heart (Ferber and Pflüger, 1990). These BPNs may modulate the heartbeat through the release of peptides into the haemolymph (Ferber and Pflüger, 1992).

The results of this study show that the motor and the median neurones with bilateral axons project to different regions of the oviduct. It is suggested that the motor neurones cause the neurogenic contractions of the junctional area of the oviducts, whereas the median neurones mainly inhibit the myogenic contractions of the lateral oviducts. The intracellular identification of these neurones is a prerequisite for future studies intended to reveal the higher neuronal centres responsible for their activation. It could also provide a useful model for the study of the effects and interactions of different transmitters, such as octopamine and proctolin, released from identified neurone populations.

Financial support by the DFG is greatly appreciated (Pf 128/6-1 and Pf 128/6-2). E.K.’s prolonged visit to the laboratory of H.-J.P. at the Free University of Berlin was made possible by a grant from the DAAD. We thank Dr P. A. Stevenson for help with the English text and for critically reading the manuscript. This study forms a part of E.K.’s doctoral thesis.