Anatomy of Motoneurones Innervating Mesothoracic Indirect Flight Muscles in the Silkmoth, Bombyx Mori

It is well known that the male silkmoth, Bombyx mori, performs a mating dance in response to the female sex pheromone. The major behavioural components of the dance are walking, wing vibration, and occasional abdominal curvature. The mating dance plays an essential role in locating the calling female (Obara, 1979). Close observations have revealed the behavioural properties of wing vibration in the mating dance, and extracellular recordings from identified mesothoracic indirect flight muscles have elucidated the electrical activity of the muscles during the dance (Y. Obara & Y. Kondoh, in preparation). The motor pattern of wing vibration during the mating dance is similar to that during flight. Deviation from the normal flight motor pattern is caused by irregular inhibitory inputs from legs that occasionalls contact the ground during the dance. Surgical experiments suggest nevertheless that some neuronal command elements for the mating dance exist in the supra-oesophagial ganglia (Y. Obara & Y. Kondoh, in preparation). These command elements are activated by stimulation with a pherormone and remain active for as long as 4–5 min afterwards. The neural mechanisms controlling the mating dance, however, remain to be demonstrated at the cellular level.

The present study was carried out first to elucidate which of the indirect flight muscles is involved in wing vibration and then to reveal the anatomy of their moto-neurones. The results will be discussed in terms of the motoneuronal architecture of the insect flight system.

Males of the TWz-strain of Bombyx mori were used. This strain lacks the dark pigments on the surface of ganglia in the ventral nerve cord, present in the wild type, which make it difficult to observe the fine dendritic processes of the moto-neurones. The moths were reared in mass culture at 26 °C under a long day ( 16 L : 8 D) photoperiod regime. They were kept at 26 °C for two days after eclosion and subsequently at 4 °C until use.

The anatomy of the mesothoracic musculature and nervous system was investigated by dissecting moths which had been preserved for 24 h in a fixative consisting of 25 ml of 40% formalin, 1·25 ml of acetic acid and 10 g of chloral hydrate in 100 ml distilled water (Chauthani & Callahan, 1966). The nervous tissues were stained with aniline blue (1 % in distilled water). The designations of the muscles and nerve trunks were based on those given by Eaton (1974) for Manduca sexta.

The number and sizes of axons in the nerve trunks and the cell bodies of the flight motoneurones within the thoracic ganglia were determined by examining serial sections in the light microscope. The tissue was fixed for 12 h in 2% glutaraldehyde in phosphate buffer, washed for 12 h in buffer containing glucose (3·6 g per 100 ml), dehydrated, and then embedded in epoxy resin through n-butyl glycidyl ether. Serial sections 1·5–2 μm thick were cut and stained with toluidine blue.

The location of cell bodies and the arborization patterns of flight motoneurones were determined by the cobalt axonal filling method (Sandeman & Okajima, 1973; Tyrer & Altman, 1974). The pro- and pterothoracic ganglia and their nerves were dissected out in a Bombyx saline (H. Tsujimura, personal communication) which was based on Naoumoff & Jeuniaux’s (1970) analysis of haemolymph. Axons were exposed to a 400 mm cobalt chloride solution containing 13 mg of bovine serum albumen per. 100 ml (Strausfeld & Obermayer, 1976), for 24–72 h at 4 °C. The tissue was rinsed in saline, developed in ammonium sulphide (1 drop in 3 ml saline), fixed, dehydrated, cleared in methyl benzoate, and then mounted in Canada balsam. On some preparations, Timm’s silver-intensification (Bacon & Altman, 1977) was performed to enhance details of the dendritic processes before dehydration. The stained neurones were drawn using a camera lucida drawing tube attached to a microscope. A total of 103 successful preparations were examined.

The forewing indirect flight muscles in the mesothorax are shown in Fig. 1 and the sclerites to which they attach are shown in Table 1. These muscles are morphologically divided into two groups. The first, the dorsal longitudinal muscles (DLM), attach to the notum, or the sclerites derived from it, at both ends. The DLMs are further subdivided into three muscles, DLM1, DLM2 and DLM3. Of these, the DLM1, consisting of five bundles (DLM1a – e), is the largest. All run longitudinally in the dorsal portion of the mesothorax. The DLM2, consisting of two muscle bundles, runs lateral to DLM1. Although both clearly belong morphologically to the DLMs, they run obliquely, or nearly dorsoventrally, with their posterior end ventral. The DLM3 is a small muscle bundle which is dorsal to the DLM1 and runs longitudinally.

The second group of indirect flight muscles, the dorsoventral muscles (DVM), consists of six muscle bundles, all of which run dorsoventrally. Three of them (DVM1 a, b and DVM2) are anterior to the coxa, while the other two (DVM4 and DVM5) are posterior to it. Another thin DVM (DVM3) arises at the mesonotum and attaches to the trochanter

The function of some of these indirect flight muscles was determined by pressing the sclerites to which they attach along their long axis. Shortening the mesonotum along the axis of DLM1 resulted in depression of the forewing, while pressing down the mesonotum ventrally along the DVM long axis resulted in wing elevation (Table 1). These results indicate that DLM1 functions as a forewing depressor and DVM as a forewing elevator. The same kind of mechanical movement of the sclerites to which the DLM2 and DLM3 attach did not cause any observable movement of the wing, and thus their function in moving wings could not yet be established.

The thoracic ganglia from which nerve trunks emerge to innervate the relevant indirect flight muscles are shown diagrammatically in Figs. 1 and 2. There are two thoracic ganglia, a pro- and a pterothoracic ganglion. The latter is composed of the mesothoracic ganglion and a series of ganglia shown ontogenetically to be the metathoracic, and the first and the second abdominal ganglia (H. Tsujimura, personal communication).

All the DLMs are innervated by the nerve IIN1c. It arises from the dorsal side of the pro-mesothoracic connective, together with the thick nerve trunk IIN1 b extending to the wing base. It then deviates from the latter and runs dorsolaterally. After branching to the mesoscutum, the IIN1 c splits into a further two nerve trunks. One runs postero-dorsally and laterally to DLM1 and innervates DLM2, while the other extends dorsally medially to DLM1 and innervates DLM1e and DLM3. The main nerve trunk of IIN1 c enters the DLM1 a and then runs dorsally while dividing into many branches that innervate DLM1a – d.

The DVMs are innervated by three nerve trunks, IIN4, IIN5b and IIN6. Nerve IIN4 arises together with IIN2 on the lateral side of the mesothoracic ganglion. It immediately runs between DVM1 b and DVM2 and then divides into many branches that innervate DVM1a, b and DVM2. Nerve IIN5b is the second branch of the median nerve trunk, IIN5, which arises from the posterolateral portion of the mesothoracic ganglion. It extends posterolaterally between DVM2 and DVM3 and divides into a few branches to innervate DVM3 and the pleuroventral muscle (PVM3). Nerve IIN6 arises on the posterolateral portion of the fused mesothoracic ganglion, runs posteriorly on the medial surface of the DVM4 and DVM 5 and sends a ventral branch to coxal (CXM), sternopedal (STM), and pleuroventral muscles (PVM). The main nerve trunk of IIN6 passes to the posterior side of DVM4 and DVM5 and innervates them with a second branch. It then innervates one of the subalar muscles (PVM5) and the axillary muscles (PDM).

The number and diameter of axons in the nerve trunks innervating the indirect flight muscles is shown in Table 2. These measurements were obtained from examination of sections taken from each nerve trunk where they enter each muscle. It was, however, impossible to determine the number and diameter of axons innervating the muscle bundles of DLM1a – b, DVM1a, b, DVM2, DVM4 and DVM5, because the nerve trunks do not branch until they have entered these groups of muscle bundles. Table 2, therefore, gives only an indication of the minimum number of axons which innervate these groups of muscle bundles.

All but two of the axons innervating DLMs were relatively thick, ranging from 7 to 15 μm in diameter (Table 2 and Fig. 3). The two thin axons were between 2 – 6 pm in diameter. One innervates DLM3 and the other, which was presumed to be an unpaired median neurone (UM-neurone), innervates DLM1 a – d muscle group. The DVM muscle groups are also innervated by relatively thick axons ranging from 6 to 13 μm in diameter. From these results and the muscle anatomy described above (Fig. 1), it was found that the larger muscles were innervated by the thicker axons. All these axons including that of the UM-neurone were ensheathed by a tissue stained heavily with toluidine blue, which seemed to be glial cells (Fig. 3).

Axonal filling with cobalt chloride through the cut end of each of the peripheral nerves revealed the number, location and dendritic arborization of the flight motoneurones (Fig. 4). Fig. 5 summarizes the positions of all motoneurones and unpaired median neurones innervating mesothoracic indirect flight muscles. When nerve IIN1c was filled unilaterally, 9 cell bodies appeared stained: 7 of them are located ipsilateral in the prothoracic ganglion and 2 in the mesothoracic ganglion (N = 21). One of the latter 2 cell is located contralateral and the other near the midline of the boundary between the meso- and metathoracic ganglia. It was shown by selectively filling each of the branches of nerve IIN1c (N = 9) that 4 large cells of the 7 cells in the prothoracic ganglion innervate DLM1a – b, and 2 intermediate and one small cell innervate DLM2 and DLM3, respectively. The contralateral cell in the mesothoracic ganglion innervates DLM1e. The median cell, a UM-neurone, has a single neurite in the anterior portion of the ganglion which then bifurcates to send paired right and left axon to the main trunk of IIN1c innervating DLM1a – d.

Filling nerve IIN4 resulted in the staining of five cell bodies (N = 13). Four of these are located ipsilaterally near the ventral surface of the mesothoracic ganglion. The fifth has a T-shaped axon and is located near the dorsal median surface of the ganglion. Filling a branch of IIN5b which extends to DVM3 (N = 24), and a branch of IIN6 to DVM4 and DVM5 (N = 4), revealed that these muscles are innervated by three and two motoneurones, respectively. Their cell bodies are ipsilateral, in the mesothoracic ganglion. These results agree well with the number of axons seen in sections of the relevant nerve trunks.

The cell bodies of these motoneurones are located in the peripheral layer or cortex of the ganglia. All of the paired motoneurones, except those of DVM3, which are located on the lateral side of the mesothoracic ganglion, have their cell bodies on the ventral surface of the ganglia. This was also confirmed cross sections of pro- and mesothoracic ganglia. Most cell bodies are seen to be located in the ventral or lateral portion of the ganglia, with only a small number on the dorsal median surface. Some cell bodies are strikingly large ranging from about 40–60 μm in diameter and occur in the regions of the ganglia where the cell bodies of indirect flight motoneurones and UM-neurones were found to be located by cobalt filling.

The cell body of the UM-neurone sending an axon to DLM1a – d is located in either the ventral or dorsal surface of the mesothoracic ganglion. In 25 preparations filled unilaterally, the cell body appeared on the ventral surface in 15 and on the dorsal surface in 10. Only one such cell body was seen in 9 preparations filled bilaterally. Another UM-neurone with its axon in the IIN4 innervating DVM1a, b and DVM2 has its cell body on the dorsal median cortex of the mesothoracic ganglion (N = 14).

The cell bodies of the four motoneurones innervating DLM1a – d appear clustered on the ventral and posterior portion of the prothoracic ganglion (Fig. 6 A, B). They are among the largest (40 – 60 μm in diameter) in the ganglion. The main neurites from their cell bodies extend into the dorsal neuropile where they branch into many secondary processes. Most of the processes extend towards the dorsal portion of the ganglion and some of them cross the midline. There appears to be a few processes extending ventrally (Fig. 6B). The main neurites then run posteriorly into pro- and mesothoracic connectives.

DLM1e motoneurone has a large contralateral cell body, which is 30 – 45 μm in diameter and located close to the root of the anterior connective in the ventrolateral region of the mesothoracic ganglion (Figs. 4 A, 6C). Its neurite runs postero-dorsally the dorsal surface of the ganglion and then turns to travel across the midline to e ipsilateral neuropile, where most dendritic branches are located. Short fine dendritic processes are also seen near the midline. In one preparation, some branches were observed contralateral to the muscle (Fig. 6C).

Two DLM2 motoneurones have cell bodies of an intermediate size (30 – 40 μm in diameter) and occur slightly anterolateral to DLM 1 a – d motoneurones in the prothoracic ganglion (Fig. 5). Their main dendritic processes lie ipsilateral to the muscles, although some fine dendritic branches cross the midline (Fig. 6D). The processes, however, are restricted to the posterior half of the ganglion and it seems that the area occupied by them overlaps that of DLM 1 a – d motoneurones.

DLM3 motoneurone has the smallest cell body (15 – 20 μm in diameter) of DLM motoneurones and is near the midline in the posteroventral region of the prothoracic ganglion (Fig. 6E). This motoneurone is characterized by having less dense dendritic processes, compared with those of other DLM motoneurones. A few secondary branches travel across the midline of the ganglion and there give rise to numerous finer branches.

The DVM 1 a, b and DVM2 muscle bundles are innervated by 4 motoneurones with cell bodies 40 – 50 μm in diameter (Fig. 7 A, B). These neurones are in a close cluster on the ventrolateral surface of the mesothoracic ganglion near the root of the anterior connective. All four neurites run together in a close bundle, extend dorsally and then turn sharply ventrolaterally to the IIN4 nerve trunk. These motoneurones have dense branches at the point where their neurites turn ventrolaterally. These dendritic branches are confined to the dorsal neuropile ipsilateral to their cell bodies, and have not been observed to travel across the midline of the ganglion. By selectively filling the branch of nerve IIN4 innervating DVM2, it was possible to stain the DVM2 motoneurone alone, so allowing its fine dendritic processes to be observed in detail (Fig. 7B).

The DVM3 is innervated by three motoneurones which have cell bodies, 20 – 30 μm in diameter, in the anterolateral surface of the mesothoracic ganglion (Figs. 4B, 7C). The neurites of each of the DVM3 motoneurones run horizontally to the lateral region of the ganglion, where most of the secondary processes arise dorsally, a few ventrally.

The cell bodies of DVM4 and DVM5 motoneurones are 40 – 50 μm in diameter and are located near the boundary between meso- and metathoracic ganglion (Fig. 7D). Their neurites run dorsally and slightly anteriorly to the dorsal region of the ganglion and there run ventrally. All the dendritic processes are in the posterior two thirds of the ipsilateral side of the ganglion.

Two UM-neurones were found in the mesothoracic ganglion (Fig. 8). Both of those UM-neurones are characterized by these morphological features. (1) their single median neurite bifurcates to form symmetrical left and right axons, (2) den-litic processes are considerably shorter and sparser as compared with paired moto-neurones examined in the present study, (3) they have relatively large cell bodies (30 – 40 μm in diameter) in the posteromedial surface of the ganglion, and (4) their axons are thin (3 – 6 μm) and this seems to be the reason why their cell bodies could not always be filled, and why their axons could not be recognized in some of the microscopical preparations.

In spite of the drastic variance in the location of the cell body of the UM-neurone innervating DLM1a – d, there is no observable difference in its basic pattern of dendritic arborization and orientation of neurite. In Fig. 8 A a neurone with a dorsal cell body is shown. Its neurite runs ventrally up to a depth of 200 – 300 μm from the dorsal surface and then turns anterodorsally to the dorsal neuropile where it bifurcates to form paired right and left axons. A small number of short and fine secondary processes arise near the T-shaped region of the median neurite and also in the paired neurites.

Mesothoracic indirect flight motoneurones have been described in cricket (Bentley, 1973), locust (Bentley, 1970; Burrows, 1973 ; Altman & Tyrer, 1974; Tyrer & Altman, 1974), dragonfly (Simmons, 1977), Heteropteran (Davis, 1976), and Dipteran insects (Ikeda et al. 1975, 1976; Coggshall, 1980). In Lepidopterous insects only the DLM motoneurones in Manduca sexta have been described (Casaday & Camhi, 1976). The present study provides the first data on DVM neurones as well as DLM neurones, which will be discussed in terms of their arrangement in the ganglion and morphological features of their arborization.

A comparison of the arrangement of DLM and DVM motoneurones in B. mori with that of other insects may allow us to imagine the general neural organization of the insect flight system. Thus, the number and location of indirect flight motoneurones in the mesothorax shows some common features among the studied species (Table 3). First, the location of the DLM neurones in the ganglia is the same throughout these insects. Motoneurones innervating the four bundles of DLM (DLM1 a – d in B. mori), the dorsal oblique muscle, and another DLM (DLM3 in B. mori) have their cell bodies in the ipsilateral and posterior portion of the next anterior ganglion, while the motoneurone of one DLM (DLM1 e in B. mori) is in the anterior of the contralateral half of the ganglion in the same segment as the muscle. In the locust (Neville, 1963; Bentley, 1970) and in Drosophila (Coggshall, 1980), the contralateral neurone innervates the dorsal most bundle of the major dorsal longitudinal muscles (DLM1e in B. mori), while the four ipsilateral motoneurones in the prothoracic ganglion innervate the more ventral bundles. Secondly, the number of motoneurones to the DLM is the same in all the insects examined except for DOM in Dysdercus which has three motoneurones. Thirdly, the dorsoventral muscles are all innervated by motoneurones of the same segment, whose cell bodies are ipsilateral to the muscles they innervate. Of the DVM motoneurones, those of tergosternal, anterior tergocoxal and tergotro-chanteral muscles are at the anterior end of the mesothoracic ganglion, while those to posterior tergocoxal muscles are at the posterior end.

This basic architecture of motoneurones innervating dorsal longitudinal muscles and dorsoventral muscles is similar to that in the abdominal ganglion of the moth (Taylor & Truman, 1974), supporting the idea that insect segmental ganglia might have a basic plan for the arrangement of motoneurones (Taylor & Truman, 1974).

One remarkable feature is that the dendritic processes of dorsal longitudinal motoneurones, including that to the dorsal oblique muscles, extend to the contralateral neuropile ; whereas dorsoventral motoneurones have a dendritic field restricted to the ipsilateral half of the ganglion. This is also the case in locust (Tyrer & Altman, 1974). The functional meaning of the difference in the extent of arborization between dorsal longitudinal and dorsoventral motoneurones is uncertain. The dorsal longi-tudinal motoneurones on each side of the ganglion, however, may morphologicalls and consequently physiologically have contact with each other through their branches extending over the contralateral hemiganglion.

By filling the dorsal nerve of the 4th abdominal ganglion which is probably homologous to IIN1c innervating DLMs, Truman & Reiss (1976) revealed one contralateral motoneurone in the ganglion of the same segment as the nerve filled in the pharate adult of Manduca. On the basis of homology of the nerve filled and its innervating muscles (H. Tsujimura, personal communication), the 4th abdominal contralateral neurone can be safely judged homologous to the DLM1e motoneurone. This abdominal motoneurone has two dendritic fields on each side of the midline which are nearly of the same size. This is in sharp contrast to the homologous neurone (DLM1 e) in the mesothoracic ganglion in B. mori which has scarce dendritic processes on the side contralateral to the axon. It is, however, not known whether this difference results from the difference in segment, developmental stage, or species.

In the present study, two unpaired median neurones were revealed. One of them is in either the ventral or dorsal median cortex of the mesothoracic ganglion, while the other is always in the dorsal median cortex. These neurones were shown to be unpaired by the results of bilateral filling. This is also supported by their morphological similarity to DUM neurones in locust (Plotnikova, 1969; Crossman et al. 1971, 1972; Hoyle, 1974, 1978).

DUM neurones associated with the dorsal longitudinal muscles (DUMDL) have been reported for Orthopterans (Bentley, 1973 ; Clark, 1976; Hoyle, 1978), a Heter-opteran (Davis, 1978), fruit fly, Drosophila melanogaster (Coggshall, 1980) and a satumiid moth, Antherae polyphemus (Heinertz, 1976). Taylor & Truman (1974) have also described six ventral and four dorsal UM-neurones in the 4th abdominal ganglion of Manduca. Alternative location, in the dorsal or ventral surface of ganglion, has not been previously described for a UM-neurone. It may arise in B. mori simply by developmental accidents.

The UM-neurone associated with the DVM has not been previously observed.

We are grateful to Dr H. Tsujimura for his technical advice and Dr M. Burrows for his critically reading and improving the manuscript. Our thanks are also due to Dr S. Ohshiki and Mr H. Kishi for their kindly providing us with pigment-free silkmoths which proved to be very useful in the morphological study of motoneurones.

This work was partially supported by Grants-in-Aid (Nos. 411801, 510901) for Special Project Research on Mechanisms of Animal Behaviour from the Ministry of Education, Science and Culture to YO.