ABSTRACT
The direct flight muscles of Drosophila melanogaster are innervated by the anterior dorsal mesothoracic (ADM) nerve and the mesothoracic accessory (MAC) nerve. Each of the four conspicuously large axons in the ADM nerve serves one of the muscles designated pal, pa3, pa4 and pa5. Muscle pa4 is additionally innervated by a very small neurosecretory axon. Muscle pa6, also innervated by the ADM nerve, receives at least one small nerve fibre but no large axon. Musclepa2 is innervated by a large axon from the MAC nerve. Large motor axons, identified by serial section tracing from their respective muscles, are consistent among different individuals in both relative positions and relative diameters within the ADM nerve.
INTRODUCTION
Several neurones involved in escape behaviour and flight have been relatively well-studied in Drosophila. The giant fibre pathway, which triggers the initial events of escape, has been described anatomically (King & Wyman, 1980; Koto et al. 1981)and physiologically (Tanouye & Wyman, 1980). Motor neurones to the indirect flight muscles, which provide the main power for flight, are also known both anatomically (Coggshall, 1978; Ikeda, Koenig & Tsuruhara, 1980) and physiologically (Harcombe & Wyman, 1977, 1978; Koenig & Ikeda, 1980; Tanouye & Wyman, 1981). Recent experiments have examined the genetics of these interconnected motor systems (Thomas, 1980, 1981; Tanouye, Ferrus & Fujita, 1981; Thomas, Costello & King, 1982; Thomas & Wyman, 1982). Genetic study has been facilitated by the large size and simple synaptic relationships of neurones in the giant fibre and flight motor pathways.
This report and the one that follows (Tanouye & King, 1983) extend the analysis of associated motor systems in Drosophila by providing initial anatomical and physiological descriptions of neurones controlling direct flight muscles. In contrast to the large fibrillar muscles, which indirectly generate powerful wing beats by distorting the thoracic walls, the direct flight muscles are small tubular muscles which act direct on wing-base sclerites to generate wing movements. The direct flight muscles control wing position for wing opening and closing and for directional guidance during flight. The present report establishes that the motor axons of several direct flight muscles are among the largest and most easily identified nerve fibres in the fly. The motor neurones are accessible for indirect recording of their electrical activity. At least one of these neurones is activated by the giant fibre pathway (Tanouye & King, 1983). This information provides a basis for further physiological and neurogenetic investigation into the neuromuscular mechanisms controlling flight in Drosophila.
MATERIALS AND METHODS
Adult female Drosophila melanogaster (wild-type strain Hochi, from the collection of Dr R. Wyman, Department of Biology, Yale University, New Haven, Connecticut, U.S.A.), aged 4–10 days post-eclosion, were immersed in cacodylate-buffered glutaraldehyde. Fixative could penetrate the thorax through holes cut by removing the head, abdomen, legs and wings. Specimens were then post-fixed in osmium tetroxide and dehydrated in ethanol. Thoraces embedded in Epon were cut into serial sections (0·5–5 μm) for light microscopy or ultrathin sections for electron microscopy. Ultrathin sections could be cut at selected locations by re-embedding thick sections that had previously been studied with the light microscope (King, Kammlade & Murphy, 1982). Individual motor axons within peripheral nerves were identified by tracing the axons through serial sections to the target muscles. Specimens for scanning electron microscopy were fixed, critical-point dried, and then dissected. For additional details on the histological processing, see King & Wyman (1980).
OBSERVATIONS
Our studies have confirmed Zalokar’s (1947) gross anatomical description of direct flight muscles in the thorax of Drosophila melanogaster. The present account of motor innervation focuses on muscles which originate on the pleural body wall anterior to the pleural apophysis and insert via tendons or apodemes directly onto wing base sclerites, muscles which Zalokar called anterior pleural muscles (Fig. 1A). Zalokar’s nomenclature, in which anterior pleural muscles are designated by the abbreviation pa followed by a number (e.g. pa2, pa3) will be followed in this report. Synonyms for these anterior pleural muscles and for homologous muscles in other species are listed in Table 1.
The anterior pleural muscles are innervated by two separate nerves named in Power’s (1948) description: the anterior dorsal mesothoracic nerve (ADM nerve) and the mesothoracic accessory nerve (MAC nerve).
The ADM nerve is a large mixed nerve arising from the dorsolateral aspect of the mesothoracic neuromere. This nerve contains about 500 fibres, including four axons which are much larger than the others and are easily resolved at low power by light microscopy. These large axons, with diameters of 3 μm or greater, are motor fibres innervating muscles pal, pa3, pa4 and pa5. Among the smaller fibres are additional axons innervating pa4; at least one of the muscles pa3, pa5 and pa6; and several posterior pleural muscles. Other small axons form a large and coherent bundle of sensory fibres originating from receptors on or near the wing. After leaving the ganglion the ADM nerve passes laterally across the front of the anterior dorso-ventral (fibrillar) muscle. Here the large motor axons begin to separate from the nerve trunk to form three separate branches, as indicated in Fig. 1B. Cross sections of these nerve branches are shown beside the anterior dorso-ventral muscle on the left side of Fig. 2A, and enlarged in Fig. 2B, C and D. The first of these branches (Fig. 2D) arises along the ventral aspect of the ADM nerve and curves posteriorly. This posterior branch contains one large motor axon that innervates muscle pal, and five smaller axons which continue posteriorly, eventually crossing the pleural apophysis to innervate posterior pleural muscles. The next branch (Fig. 2C) also arises from the ventral aspect of the ADM nerve; it contains two large axons and three small axons. The more ventral of the two large axons soon separates to innervate muscle pa3. The other large axon innervates muscle pa5. The three small axons are difficult to trace reliably through serial sections; at least one must continue to innervate muscle pa6 since no other nerve approaches this muscle. The third branch (Fig. 2B) arises dorsally; it contains one large axon and one very small axon, both of which innervate musclepa4. The remainder of the ADM nerve, called the wing nerve, consists of sensory axons from the wing. The arrangement of large motor axons in the distal ADM nerve is quite consistent from individual to individual. The conspicuous components of the nerve are arrayed, from dorsal to ventral, as follows (Fig. 2A; also see Fig. 2B in Tanouye & King, 1983): pa4 axon, wing nerve sensory axons, pa5 axon, pa3 axon and pal axon. The axons to musclespa3 mápa5 may occasionally be reversed; but may be distinguished since thepaJ axon is the larger of the two. The motor axon to muscle pa4 is most readily identified, since it is separated from the other motor axons by the sensory bundle and is also consistently the largest axon in the nerve.
The MAC nerve appears to contain motor axons only. This nerve originates from the posterior aspect of the mesothoracic neuromere, passes beneath the arm of the sternal apophysis (mesofurca), then curves laterally behind the sternopleural muscle. Proximal branches of the MAC nerve innervate the sternopleural muscle as well as more posterior muscles not identified in this study. Distal to these branches the MAC nerve contains four large axons only, one noticeably larger than the rest (Fig. 2E). One of the three similar-sized axons passes anteriorly across the pleural apophysis to innervate muscle pa2; muscle pa2 is the only anterior pleural muscle not innervated by the ADM nerve. The other MAC axons innervate posterior pleural muscles.
It may be noted from this description that both the ADM nerve and the MAC nerve innervate both anterior and posterior pleural muscles. Five small axons from the posterior branch of the ADM nerve cross the pleural apophysis posteriorly while one axon from the MAC nerve crosses the pleural apophysis anteriorly, with the curious result that at this point a single nerve bundle includes motor axons coursing in opposite directions. Heide (1971) first pointed out this ‘nervenbrucke’ connecting the ADM nerve and the MAC nerve in Calliphora, whose gross pattern of flight motor nerve branches appears virtually identical to that described here.
The axons innervating muscles pa1, pa2, pa3, pa4 andpa5 are all relatively large for Drosophila, with diameters typically ranging from 3–10μm (Fig. 2). The motor axon to muscle pa4 is consistently the largest of these, occasionally reaching a diameter of 15 μm. The pa4 axon is thus one of the largest nerve fibres in the fly, comparable only to the cervical giant fibre and the giant motor axon of the tergotro-chanteral muscle, both elements in a specialized giant fibre pathway (King & Wyman, 1980). The axon to muscle pa3 is noticeably smaller than that topa4; the pa5 and pal axons are slightly smaller still; and the pa2 axon is the smallest of these identified nerve fibres. Absolute axon diameters vary rather widely, perhaps in relation to size or age of the specimen or to specific location along the nerve; these factors were not carefully controlled. It is the relative size among these large motor axons that is consistent among individual flies. The large diameters of these direct flight muscle axons together with the small size of Drosophila greatly facilitates serial section tracing of individual axons from ganglion to muscle. This should allow routine identification of motor processes within the thoracic ganglion and permit future analysis of synaptic contacts involving direct flight muscle motor neurones.
Each of the anterior pleural muscles (with one exception) receives one large motor axon. The motor axon enters its target muscle at a consistent location (Fig. IB) by passing between adjacent muscle fibres. From within the muscle the axon sends fine terminal branches between muscle fibres where neuromuscular junctions are established (Fig. 3A, B). The only exception to this pattern is muscle pa6. This muscle receives no large axons; the small axon(s) apparently destined for this muscle (Fig. 2C) have not yet been reliably traced through serial sections. Since muscle pa6 is basically a planar fan of muscle fibres rather than a solid mass, its motor nerve cannot penetrate into the muscle but rather is distributed across its medial surface (Fig. 1).
Neuromuscular junctions have been observed in muscles pal, pa3 and pa4. The presynaptic terminals contain synaptic vesicles, some of which are intimately associated with specialized T-bar membrane structures (Fig. 3B). These endings thus display morphology typical of dipteran chemical synapses (see Toh & Kuwabara, 1975; Osborne, 1975). The subsynaptic cytoplasm of the muscle fibres appears denser beneath each junctional contact than in surrounding regions (Fig. 3B). Neuromuscular junctions are frequently encountered in most sections of these muscles, but their size, distribution and ultrastructure have not yet been quantitatively studied. In addition to such standard neuromuscular junctions, muscle pa4 was found to include small neural processes, distinct from the motor axon terminals, which contain large dense vesicles (Fig. 3C). These processes resemble neurosecretory endings found in other insects. They probably arise from a very small axon that accompanies the large pa4 motor axon (Fig. 2B). Whether any other direct flight muscles receive similar neurosecretory endings has not been determined. Since five axons (two large and three small) are associated with the three muscles pa3,pa5 and pa6, at least one of these muscles must also be multiply innervated, although this has not yet been directly confirmed. Muscle pal appears to receive one large axon only, since the other five axons in the posterior branch of the ADM nerve continue across the pleural apophysis.
DISCUSSION
The identified motor axons described above are consistent not only in their peripheral nerve courses and destinations but also in their relative positions and diameters within the peripheral nerve. These observations extend the results of Ikeda et al. (1980) who found consistent relative positions of dorsal longitudinal muscle motor axons in the posterior dorsal mesothoracic nerve of Drosophila. These consistent morphological features offer an opportunity to identify specific motor neurones without the laborious tracing of axons from muscles or the use of intracellular labels injected into somata. Preliminary observations within the central nervous system (D. G. King, unpublished) suggest that here also these motor axons display consistency in the positions of their major postsynaptic processes. Thus, these identified motor neurones may become routinely accessible for histological analysis of both synaptic morphology and peripheral axonal form.
One anterior pleural muscle (pa4) receives a fine neurosecretory fibre in addition to its single large motor axon. This is apparently the first report of neurosecretory endings in Drosophila flight muscles, although similar endings are found in a variety of insect and vertebrate muscles (Osborne, Finlayson & Rice, 1971). A previous description of neuromuscular endings in Drosophila did not indicate such neurosecretory terminals in the dorsal longitudinal muscle (Shafiq, 1964). The function of neurosecretory axons in muscle is still obscure. Even the distribution of such axons to different muscles within an organism has never been determined. The direct flight muscles of Drosophila may provide a good model for studying the distribution of such fibres, since several distinct muscles can be included in single sections prepared for electron microscopy. The existence of small midline neurones with fine axons to flight motor nerves (Coggshall, 1978), together with the current observation of neurosecretory endings in flight muscle, suggests that Drosophila may possess neurones similar to the neurosecretory DUM (dorsal, unpaired, median) neurones found in locusts (Hoyle, Colquhoun & Williams, 1980). Genetic manipulation of muscle development in Drosophila (Costello & Thomas, 1981) may eventually offer some clues to the functional role of neurosecretion in muscle.
A major challenge for developmental biology is presented by the morphogenesis of specific features that characterize identifiable neurones. The motor neurones described in this report are characterized by consistent diameters and positions within their respective nerves as well as by specific peripheral pathways and target muscles. A growing axon may reach its destination along pathways established by pioneer fibres early in development when the distances to be traversed are still quite small (Goodman & Bate, 1981). This of course does not explain how a growing motor axon recognizes its appropriate target muscle, nor why axons destined for adjacent muscles follow divergent nerve courses in the ADM and MAC nerves. The anatomical site from which a muscle originates (Lawrence, 1982) does not seem to determine the course of its motor axon, since both ADM and MAC nerves innervate muscles from both dorsal and ventral regions.
Mechanisms that determine consistent axon diameter are also unknown. The hypothesis that axon diameter is precisely adapted to provide a behaviourally appropriate conduction velocity is appealing. If this were true, the ontogenetic mechanisms which determine axon diameter must allow for differential modification of individual axons through evolution. This hypothesis is supported by the existence of giant-axon pathways mediating rapid escape reflexes in a variety of invertebrate and vertebrate animals. Whether this concept will include axons which are not so extremely specialized is not yet clear. It was somewhat surprising that the largest axon to a direct flight muscle (pa4) in Drosophila is not linked to the giant fibre pathway while a smaller axon (to muscle pa3) within the same nerve is associated with this pathway (Tanouye & King, 1983). Perhaps musclepa3, although driven by the cervical giant were, need not be activated immediately in the reflex initiation of flight. By contrast, the tergotrochanteral muscle, also activated by the cervical giant fibre (Tanouye J Wyman, 1980) receives one of the shortest and largest-diameter axons in the fly (King & Wyman, 1980). The large axon to muscle pa4, although apparently not needed in the giant fibre reflex, may participate in some other very rapid reflex, perhaps involved in flight guidance. The occurrence of motor axons as small as those apparently innervating muscle pa6 is difficult to interpret in view of a total lack of information about the function of this muscle. (See Tanouye & King, 1983 for additional discussion of the functions for direct flight muscles in Drosophila.) Such functional rationalization for specific axon diameter is unconvincing, however, without additional supporting data. An alternative explanation for consistent differences in diameter among different identified axons is that such differences reflect incidental consequences of developmental processes and have no specific adaptive significance with respect to conduction velocity.
Additional physiological investigation together with comparative study of homologous axons in other flies, whose relative axonal size distribution can differ from that in Drosophila (D. G. King, unpublished observations), may help resolve the significance of specific determination of motor axon size. Such studies, which complement and extend those analysing genetically modified motor pathways in Drosophila (Thomas & Wyman, 1982; Thomas, 1980, 1981; Thomas et al. 1982), can contribute toward understanding the mechanisms that regulate cellular morphogenesis in the nervous system.
ACKNOWLEDGEMENTS
We thank Nancy Kammlade for assisting with histological specimen preparation. We thank Dr Seymour Benzer for his support. M. A. T. was supported by National Institutes of Health Training Grant GM 7401-02. This work was supported in part by National Science Foundation Grant PCM 7911771 to S. Benzer and the Pew Memorial Trust and in part by Southern Illinois University School of Medicine.