ABSTRACT
In the hemimetabolous insect Locusta migratoria, fundamental restructuring occurs at the transition from flightless nymph to flight-capable adult. This transition involves all components of the flight circuit, which is present but not used for flight in nymphs. The meso- and metathoracic pleuroaxillary muscles, M85 and M114 respectively, constitute one component of this circuit. In the adult locust, these are flight-steering muscles, but their function in the nymph is as yet unknown. Our study reveals that adult and nymphal metathoracic pleuroaxillary muscles M114 differ profoundly. The nymphal muscle contains the distinct part M114c in addition to parts M114a and M114b characteristic of the adult. The contractions of M114c are slow and long-lasting, corresponding to its long sarcomeres and slow form of ATPase, and contrast with the adult muscle parts M114a and M114b in all of these features. We demonstrate a hormone-dependent degeneration of M114c after the adult moult. This degeneration can be blocked by actinomycin D and cycloheximide. It may thus be termed genetically programmed cell death, triggered after the adult moult and, as demonstrated here, functioning via the ATP-dependent ubiquitin pathway. Given the defined onset of degeneration after the adult moult, it is possible that M114c may fulfil a specific function in nymphs, during or shortly after moulting.
Introduction
During development from larva to adult, insects undergo changes in anatomy and physiology that finally result in structures subserving adult behaviour. Holometabolous insects such as Manduca sexta go through a pupal stage in which major restructuring and reorganisation of neuronal and muscular structures occurs prior to eclosion of the adult moth (Weeks and Levine, 1990). Among other restructurings, the larval prolegs are lost during metamorphosis. This involves the degeneration of proleg muscles and the regression of proleg motor neurones. At the end of the larval stage, a subset of these motor neurones dies, whereas the central and peripheral dendrites of the remaining subset expand again (Weeks and Truman, 1986; Weeks and Levine, 1990). Similarly, certain body wall muscles (external intersegmental muscles) degenerate early in the pupal stage and are then rebuilt from myoblasts to function in adult motor circuits. Strikingly, some larval muscles (internal intersegmental muscles) are retained and used in eclosion before they finally degenerate (Weeks and Truman, 1986). In hemimetabolous insects, the transition from nymph to adult seems to be less dramatic at first sight, since nymphs closely resemble the adult insect and a pupal stage is omitted. However, a closer look reveals profound restructuring in neuronal and muscular structures. In the flightless grasshopper Barytettix psolus, it has been shown that muscles specifically used during moults degenerate before the final moult (Arbas and Tolbert, 1986). In addition nymph muscles indispensable to the adult, such as the flight muscles of Locusta migratoria, change in their physiological properties and fibre number, and tracheolisation increases (Van den Hondelfranken et al. 1980; Riddiford and Truman, 1993).
Thereby, the relative titres of certain hormones direct developmental processes (Weeks and Truman, 1986; Riddiford and Truman, 1993; Tanaka and Pener, 1994). In Manduca sexta, the developmental history of an entire behavioural circuit from sensory input to motor output has been identified and analysed (Weeks and Truman, 1986; Weeks and Levine, 1990; Fahrbach et al. 1994). However, besides hormonal influences, activity-dependent processes have been shown to be involved in rebuilding the motor synapses of newly formed muscles in the pupal stage of M. sexta (Weeks and Davidson, 1994). Another important factor in restructuring developing neuronal systems is programmed cell death, which in M. sexta is dependent on both mRNA and protein synthesis, and is triggered by steroid hormones via nuclear receptors (Lockshin and Williams, 1964; Fahrbach et al. 1994).
As in M. sexta, the components of a mechanosensory pathway for flight steering have been well characterised in the hemimetabolous insect Locusta migratoria. This circuit consists of wind-sensitive filiform hairs (Pflüger et al. 1994), a projection interneurone in the ventral nerve cord (Pflüger, 1984) and a special pair of direct flight muscles, the pleuroaxillary (PA) muscles (Snodgrass, 1929). This pathway functions during adult flight-steering behaviour (Elson and Pflüger, 1986; Wolf, 1990; Burrows and Pflüger, 1992); however, its components are already present in the nymph. Since a developmental activity-dependent structural dynamic has been found for the receptors of the filiform hairs on the input side (Pflüger et al. 1994), we were interested in exploring the development and underlying mechanisms of the target side of this mechanosensory pathway, the PA muscles. These muscles, which function as flight-steering muscles in the adult locust, have also been identified in the nymph (Snodgrass, 1929; Wiesend, 1957). In this report, we compare the structure and physiology of the nymphal and adult PA muscle of L. migratoria and describe the sequence of developmental restructuring this muscle undergoes. We then elucidate the probable mechanisms driving these changes. Our findings suggest that the restructuring of the PA muscle involves hormone-dependent programmed cell death.
Materials and methods
Experimental animals
Male and female fifth-instar nymphs and adult Locusta migratoria were used. For hormonal treatment experiments, third- and fourth-instar nymphs of L. migratoria were also used. All experimental animals were taken from an established crowded colony at the Freie Universität Berlin, maintained under standard conditions with a 12 h:12 h L:D photoperiod and at a constant temperature of 28 °C. Specimens of defined age were obtained by separating groups of newly moulted animals. Since there was no observable difference between the genders, the results for both were pooled.
Microscopy
For scanning electron microscopy and transmission electron microscopy, M114 was removed together with its cuticular insertions (Fig. 1) and was pinned out in isotonic saline. The samples were fixed either in 4 % neutral formaldehyde (10 ml of 37 % formaldehyde, 1 g of CaCl2, 80 ml of distilled water, excess CaCO3; scanning electron microscopy) or in GPA fixative (5 ml of 25 % glutaraldehyde, 15 ml of saturated picric acid, 0.1 ml of acetic acid; scanning electron microscopy/transmission electron microscopy) and treated with 1 % osmium tetroxide (OsO4) in distilled water for 30 min. Following dehydration through an ethanol series and final immersion in 100 % acetone, they were subsequently critical-point-dried, and a gold layer was evaporated onto the sample surfaces. Samples were assessed using a Philips SEM 515. For transmission electron microscopy, the samples were incubated in 2 % aqueous uranyl acetate after treatment with OsO4, dehydrated in an ethanol series, incubated in propylene oxide and embedded in Araldite (Fischer) (Watson, 1986). Transmission electron microscopy specimens were sectioned and viewed using a Philips electron microscope at the University of Wales, Cardiff, by Dr A. H. D. Watson.
Evaluation of actin band labelling
To obtain PA muscle samples with fibres at their resting lengths, the muscles were excised with their cuticular insertions and immediately pinned out and fixed in 4 % neutral formaldehyde, imitating in vivo stretching. Subsequently, they were stained in a 2.5 % solution of TRITC-labelled phalloidin in phosphate buffer (0.1 mol l−1; 8 g of NaCl, 0.2 g of KCl, 1.44 g of Na2HPO4.2H2O, 0.24 g of KH2PO4, in 1 l of distilled water, pH 7.2; 1 h at room temperature or 12 h at 4 °C). The preparations were cleared in glycerol and viewed with a Rhodamine filter in a Zeiss Axiophot microscope. Since the muscles of different individuals of one age category differed in size, the relative sarcomere length was calculated for each muscle part of individual muscles: 10 sequences of 10 actin bands were measured, and the median sarcomere length for each muscle unit was calculated. The sum of all absolute medians was taken as 100 %, and the proportion of each individual muscle unit median was then calculated and designated as relative sarcomere length. For each muscle unit, the results for muscles of defined age categories were pooled. The same technique was used to assess the effects of actinomycin D and cycloheximide on the development of M114a+b and M114c.
ATPase labelling
To classify the ATPase of M114, excised muscles were frozen in embedding medium for frozen specimens (Tissue-Tek, O.C.T. compound 4583, Sakura). Alternating cryo cross sections 8 μm thick were tested for three variables: (a) ATPase activity, pH 8.4, (b) pH stability at acid pH (pH 4.6) and (c) pH stability at basic pH (pH 10.2). Staining of the myosin ATPase was performed according to the method of Maier et al. (1984). The staining intensity of each individual muscle part was evaluated for comparison with the other muscle parts of the same muscle. The staining intensities of muscle fibres tested for ATPase activity correspond directly with enzyme activity (dark staining reflects high activity); the staining intensities of fibres tested for pH stability correspond directly with inhibition (no staining) or non-inhibition (dark staining) of the enzyme.
Force measurements
Contractions of M114 nerve–muscle preparations were measured using a force transducer for isometric tension measurements (Cambridge Instruments, model 400A) at room temperature (21–24 °C) as previously described (Stevenson and Meuser, 1997). All preparations were stimulated at their motor nerve using single stimuli of adequate amplitude and 0.2 ms stimulus duration generated by a Grass SD9 stimulator. Contraction signals were recorded on DAT tape and evaluated using Spike2 (Cambridge Electronic Design CED 4001).
Retrograde and anterograde staining of the motor nerve
Retrograde and anterograde staining of the motor nerve, N4D4, was carried out as previously described (Pflüger et al. 1986). Staining solutions were 2.5 % Neurobiotin (Vector Laboratories) or biocytin (Sigma) in distilled water. For dye diffusion, the animals were cooled at 4 °C for 2–4 h and subsequently left for another 32–48 h at room temperature in a humidity chamber. The staining was developed as previously described (Stevenson and Meuser, 1997) or by application of Cy3–streptavidin (Dianova; 1:1000 at 4 °C overnight or at room temperature for 2.5 h) (Mesce et al. 1993). Anterograde and retrograde cobalt backfills were obtained using a 2.5 % aqueous solution of cobalt hexammine chloride (Pflüger et al. 1986). Bidirectional backfills of a few neurones were obtained by perforating nerve N4D4 with a microelectrode. Ideally, only one axon was severed and filled with Neurobiotin or biocytin. The staining was detected using the Cy3 method. All stained material was dehydrated in an ethanol series and cleared in methyl salicylate for microscopic viewing (Zeiss Axiophot or confocal microscope, Leica TCS40). Drawings were obtained with the aid of a camera lucida.
Immunocytochemistry
γ-Hydroxybutyric acid (GABA) immunocytochemistry was employed to reveal the projection of the common inhibitor neurone (CI1) on nymphal PA muscles. PA muscles were fixed in GPA fixative for 2 h. The preparations were further treated according to established immunotechniques as described by Stevenson et al. (1994). The immunostaining was achieved using a commercial GABA antibody (Sigma) and the ABC method (Stevenson et al. 1994). The stained samples were dehydrated in an ethanol series and cleared in methyl salicylate. Drawings were obtained with the aid of a camera lucida.
Hormone treatments
Newly moulted fifth instars (±6 h after the moult) were treated with the juvenile hormone analogues epofenonane (Sigma) (50 μl of 1 % epofenonane in n-heptane, resulting in supernumary nymphs) and fenoxycarb (CIBA) (5 μl of 150 μg μl−1 fenoxycarb in 100 % ethanol, resulting in overaged nymphs). Newly moulted fourth or fifth instars were treated with precocene (5 μl of 44 μg μl−1 precocene acetone, resulting in precocious adults) (Kutsch and Stevenson, 1984; Nair and Prabhu, 1985; Wang et al. 1993). To generate ‘chimeras’, fifth instars were supplied with wax pellets, each containing 10 μl of 1 % epofenonane (Fluka), on the epimeron of one body half only, and a pure wax pellet on the other body half as a control. Following treatment, animals of one treatment category were kept in groups of no more than 30 in separate cages under strictly standard conditions. Solvent-treated specimens of corresponding age were kept separately as controls. In all treatment groups, M114c or M114c–rest body (adipose tissue in place of the nymphal M114c) were assessed and compared with the structure, colour and size of M114a and M114b of the same animal and with M114a, M114b and M114c of control animals. Body size, the structure of the wings and cuticle colour were also assessed.
Application of actinomycin D and cycloheximide
To inhibit RNA synthesis (actinomycin D, Sigma) and protein synthesis (cycloheximide, Sigma), groups of fifth instars (1–2 days before the imaginal moult) and groups of newly moulted adults (0–3 h after the adult moult) were treated with (a) 3 μl of actinomycin D solution (2 μg μl−1 actinomycin D in isotonic locust saline), (b) 1 μl of cycloheximide solution (2.5 μg μl−1 cycloheximide in isotonic locust saline) or (c) 3 μl of isotonic locust saline. The solutions were injected ventrally into the last segments of the abdomen. The drug concentration was chosen to allow the adult moult to proceed. The PA muscles were labelled with fluorescent phalloidin and assessed as described above.
Detection of ubiquitin during cell degradation in M114
Ubiquitin was quantified by enzyme-linked immunosorbent assay (ELISA) in frozen muscle preparations. The trachea, the fat body other than the remains of M114c and the cuticle were removed from PA muscles of defined age. The entire preparation was performed on ice. Either entire muscles (M114a+b+c) or separated parts (M114a+b and M114c) were frozen in 100 μl of sample buffer (50 mmol l−1 Tris–HCl, pH 7.8, 2 mmol l−1 EDTA, 2 mmol l−1 EGTA, 1 mmol l−1 β-mercaptoethanol) at −20 °C. Each sample was thawed on ice and then mixed with 100 μl of sample buffer, homogenised, vortexed and stored on ice. Protein was measured according to the method of Bradford (1976) using bovine serum albumin (BSA) as standard. To determine the amount of ubiquitin in the homogenates, the protein concentration of each probe was adjusted to 0.2 μg μl−1 using sample buffer. A concentration gradient of each calibrated sample and of ubiquitin (Sigma), as a standard, was then pipetted over the microtitre plate. Ubiquitin was detected according to a standard immuno protocol (Müller, 1997) using anti-ubiquitin (rabbit, 1:1000 in blocking buffer) (Sigma) as the primary antibody. The staining reaction was measured with an ELISA reader in equal time steps at 405 nm. To determine the absolute ubiquitin content of each sample, the increase in staining per time unit was calculated for each sample at each concentration. Using the ubiquitin samples as standard, the concentration of ubiquitin could be determined for each muscle sample.
Results
Anatomy
Nymphs and adult L. migratoria both possess a peripherally located direct flight muscle, termed the pleuroaxillary (PA) muscle M85 in the mesothorax and M114 in the metathorax (Snodgrass, 1929). Here, we focused on the metathoracic PA muscle M114 (Figs 1, 2). This muscle originates on the posterior face of the metathoracic pleural ridge and stretches in a triangular shape dorsoventrally to insert at the third axillary sclerite, part of the pteral joint (wing hinge). Light microscopy and scanning electron microscopy studies revealed a profound difference between nymphal and adult PA muscle. The adult muscle is divided into two parts, M114a and M114b (Fig. 2A), M114a being the most dorsal part (Snodgrass, 1929; Pflüger et al. 1986; Elson and Pflüger, 1986; Wolf, 1990; this study). In contrast to the description of Wiesend (1957), the nymphal muscle consists of three distinct anatomical parts. This additional third part was termed as M114c, according to the established nomenclature (Snodgrass, 1929; Pflüger et al. 1986). M114c is the most ventral part of the nymphal PA muscle and inserts at a distinctively different point from M114a and M114b on the third pteral wing hinge. M114c inserts at a cuticular hinge that covers the insertions of M114a and M114b (Fig. 2A, inset). In the adult animal, where M114c is replaced by adipose tissue, this hinge lies more dorsally, exposing the insertions of M114a and M114b (Fig. 2B).
Sarcomere length and constitution
Conclusions can be drawn about the functional characteristics from the sarcomere lengths and the condition of the actin bands. We compared the sarcomere lengths of the different parts of M114 and followed changes in sarcomere constitution during development by labelling actin filaments with fluorescent phalloidin. As these labellings of muscles (Fig. 3A,B) and their quantification (Fig. 3C) in defined age groups show, nymphal muscles and muscles of young adult animals up to 20 h after the adult moult can be divided into two fibre classes. One class consists of short sarcomeres (muscle parts M114a and M114b) with narrow actin-band spacing, whereas the second class consists of long sarcomeres with wide actin-band spacing (muscle part M114c). This difference is significant (Fig. 3C). At 16 h after the adult moult, the actin bands in M114c show signs of decomposition. In animals 22–24 h after the adult moult (Fig. 3B), no distinct actin register could be observed in M114c; the tissue was thus termed M114c–rest body (24 h, Fig. 3B). With advancing age, the decomposition in M114c eventually results in adipose tissue lacking filamentous actin. M114a and M114b show no signs of decomposition during metamorphosis and keep their narrow actin-band spacing (Fig. 3; Table 1). The values of sarcomere lengths in Fig. 3 are given as ‘relative sarcomere length’ (see Material and methods) since muscles within one age category differ greatly in absolute muscle size (selected examples of absolute sarcomere lengths: nymph, M114a, 5 μm per sarcomere, M114b, 6 μm per sarcomere, M114c, 9.5 μm per sarcomere; adult 5 h, M114a, 4 μm per sarcomere, M114b, 4.8 μm per sarcomere, M114c, 8 μm per sarcomere; adult 16 h, M114a, 4.7 μm per sarcomere, M114b, 5.3 μm per sarcomere, M114c, decomposition beginning in 50 % of the animals, 9.4 μm per sarcomere; adults older than 48 h, M114a, 6.8 μm per sarcomere, M114b, 6 μm per sarcomere).
Histological ATPase staining reveals two ATPase isoforms
In invertebrates, three different forms of ATPase have been described according to their reactivity at defined pH and to their pH stability. The three isoforms are graded as slow, medium and fast ATPases, each displaying a different staining intensity when investigated according to the protocol of Maier et al. (1984). We applied these staining methods to nymphal and adult PA muscles; two types of ATPase were found (Fig. 4). The slow isoform stains lightly at pH 8.4, is base-stable according to its dark staining after preincubation at pH 10.2 and is restricted to M114c. The fast isoform stains intensely at pH 8.4, is neither base-nor acid-stable, and is restricted to nymphal and adult M114a and M114b (Table 1).
Contraction properties of M114
Muscle part M114c degenerates early after the adult moult. The muscle loses fibres that differ considerably from the remaining ones. To study eventual differences in contraction kinetics in nymphal and adult PA muscles, we employed nerve–muscle preparations. The adult muscle (Fig. 5A) can be stimulated to contract in short twitches (time to peak 40–50 ms, twitch duration 300–400 ms). Using graded stimulus amplitudes, the total twitch can be separated into two independent twitch amplitudes, each with peaks at 45–50 ms. In comparison, the nymphal muscle (Fig. 5B) shows a long-lasting total contraction (maximal duration of contraction 1400 ms) that displays a prominent ‘shoulder’ and can be separated into two fast twitch contractions (time to peak 60–80 ms) and one rather slow and long-lasting contraction (time to peak approximately 300 ms). This third ‘slow’ contraction could only be elicited in nymphal preparations and adult preparations up to 16 h after the adult moult. Furthermore, following experimental ablation of M114c or nerve branch N4D4c, the contraction lacked the shoulder and could only be separated into two fast twitch contractions. This corresponds to the results of anatomical and histochemical studies, which revealed that M114c successively degenerates 16 h after the adult moult (Figs 2–4). Therefore, there is a very good correlation between anatomical changes in M114 and its contraction properties (Table 1).
Innervation of M114
Both nymphal and adult M114 are innervated by the most distal branch of nerve 4, N4D4. Light microscopic examination suggested that there is no major change to the branching pattern on M114 during metamorphosis. Thus, we employed retrograde and anterograde staining of nerve N4D4 to reveal the number of motor neurones and the pattern of innervation on M114. Retrograde stainings of N4D4 revealed that nymphal as well as adult M114 are innervated by four different types of neurones, the number of which remains constant from nymph to adult (Fig. 6). (i) There are two motor neurones with large somata (30–50 μm in diameter in nymphs and 40–60 μm in adults) and thick axons. (ii) There are four motor neurones with small somata (15–18 μm in diameter in nymphs and 20–30 μm in adults) and thin axons. Both types of motor neurones lie in a ventral posterior cell cluster. Their primary neurites project to the dorsal surface of the ganglion, where they ramify in parallel. (iii) The third type of neurone innervating M114 is the common inhibitory neurone CI1 (Fig. 6), which has a large GABAergic (Watson, 1986; Wolf, 1990) ventral midline soma. It projects unilaterally in nerves 3, 4 and 5 and innervates M114a via nerve N4D4a (Fig. 7). (iv) As the fourth class of neurones innervating M114, we identified a single dorsal unpaired median (DUM) neurone. This single large dorsal midline cell projects bilaterally in nerves 3, 4 and 5, and is the first DUM neurone of this category with an identified target (PA muscles M114 and M85). We thus termed this cell DUM3,4,5(a) (Meuser et al. 1995; Stevenson and Meuser, 1997). The number of preparations assessed was 125 larval samples (biotin, N=63; cobalt, N=62) and 82 adult samples of different age after the adult moult (biotin, N=43; cobalt, N=39). Neurobiotin/biocytin was found to be the more reliable tracer for staining neurones with thin axons such as the CI, the DUM cell (Stevenson and Meuser, 1997) and the small motor neurones.
Employing ultra-thin sections of N4D4 just before it enters M114, we showed that the number of retrogradely stained neurones remains constant, although the muscle undergoes profound changes. In all nymphal and adult preparations, eight axon profiles were consistently observed (Fig. 6C,D), two large-diameter axons (9–12 μm in larvae, 15 μm in adults) and six smaller-diameter axons (3–5 μm in larvae, 2.5–6 μm in adults). Given the axon diameters observed in the backfills (Fig. 6A,B), the large diameters in the transmission electron micrographs can be assigned to the somata of the large motor neurones.
To reveal the axonal projections on M114c, we anterogradely stained M114 with Neurobiotin/biocytin and developed the stainings using Cy3–streptavidin. Fig. 7A–E shows camera lucida drawings of the anterograde staining of nymphal muscles. One of the two motor neurones with thick axon innervates M114a exclusively (Fig. 7A), whereas the second neurone innervates both M114a and M114b (Fig. 7B). There is an exception to this rule in a small percentage of animals: in approximately 15 % of the nymphs tested (and in 19 % of mature adult animals older than 24 h), both neurones with large somata innervated both M114a and M114b. In nymphs and recently moulted adults, M114c is innervated by thin axons only, some of which also project onto M114a and M114b (Fig. 7C,F). Within the limits of resolution of this method, this pattern of innervation remains unchanged. The remains of M114c in adults still receives arborizations of thin axons (Fig. 7G). Since Neurobiotin/biocytin is actively transported in neurones, we assume that these endings are still intact. By perforating nerve N4D4 and then staining the nerve bidirectionally, we found that the DUM cell innervates the entire PA muscle (Fig. 7D). GABA immunohistochemistry revealed that CI1 projects only onto M114a in nymphs (Fig. 7E), as it does in adult locusts (Wolf, 1990). Thus, our investigations show that, throughout postembryonic development, the innervation of M114 remains unchanged in its distribution and assignment to specific muscle parts and that, in particular, the innervation of M114c survives the degeneration of this muscle part.
The degeneration of the PA muscle is hormone-dependent
The pattern of innervation and the number of neurones innervating M114 is developmentally conserved. Nevertheless, the muscle shows profound restructuring after the adult moult. A comparison of this phenomenon with that in other insect species, such as M. sexta (Weeks and Truman, 1986), the flesh fly Sarcophaga bullata (Bothe and Rathmayer, 1994) and the fruit fly Drosophila melanogaster (Truman, 1990), suggests that the degeneration of M114c may be driven hormonally. We tested the influence of hormones on the degeneration of M114c by changing the hormonal equilibrium either by applying juvenile hormone analogues or by chemical allatectomy.
Animals treated with fenoxycarb, a potent juvenile hormone analogue, never moulted. They survived up to 40 days as overaged nymphs. The yellow patches of the nymphal cuticle turned bright green 24 h after treatment. The wing buds remained unmovable and black. The PA muscle consisted of three parts and, in structure and colouring, resembled the PA muscles of very old fifth instars just before moulting (Table 2). Treating only one body side of fifth-instar nymphs with wax pellets containing the juvenile hormone analogue epofenonane resulted in chimerism after the final moult. The epofenonane-treated body half showed the nymphal cuticle colour, and the wings were malformed, black and immobile. The PA muscles consisted of three parts, as in normal nymphs. The untreated body half, however, was normally developed, adult in cuticular colour and wing shape, and its wings were flapped in response to wind stimuli. Correspondingly, the PA muscles consisted of only two intact parts, with adipose tissue in the place of M114c (Table 2).
This result was reversed when third or fourth instars were treated with precocene, a drug that chemically ablates the juvenile-hormone-producing corpora allata (Kutsch and Stevenson, 1984). This treatment resulted in L4 or L5 adultiforms (ADF4, ADF5) respectively (Fig. 8B). These adultiforms were adult in cuticle colouring and had miniature wings that were flapped in response to wind stimuli. The PA muscles consistently contained only the adult-specific muscle parts M114a and M114b (Table 2).
Effect of actinomycin D and cycloheximide on the degeneration of M114c
Actinomycin D (a mRNA inhibitor, slowly metabolised) and cycloheximide (a protein synthesis inhibitor, quickly metabolised) are established drugs for interfering with genetically induced developmental changes (Fahrbach et al. 1994). We applied this technique to study the implication of genetically driven cell death in the degeneration of M114c and to reveal the time window during which the degeneration is triggered, by injecting the drugs 24 h before or 0–3 h after the final moult.
Injection of actinomycin D, 24 h before the adult moult significantly inhibits degeneration in M114c (Fig. 9). In contrast, injection of cycloheximide before the adult moult has no effect on the degeneration. As in control animals, muscle part M114c had degenerated after 36 h in approximately 90 % of the animals tested. Clearly, injection of actinomycin D as well as injection of cycloheximide 0–3 h after the final moult inhibit degeneration of M114c (Fig. 9). Surviving animals were killed 28–36 h after injection. Approximately 80 % of the animals tested had an intact M114c, whereas M114c was well into degeneration in 90 % of the control animals. These results implicate a genetic pathway in the degeneration of muscle part M114c. In contrast, the maturation of muscle parts M114a and M114b is uninfluenced by the application of actinomycin D and cycloheximide, as shown by labelling of actin bands (data not shown) and comparisons of the anatomical appearance with that of other thorax muscles.
Role of the ATP-dependent ubiquitin pathway
One pathway underlying the constant protein turnover in eukaryotic cells is the ATP-dependent ubiquitin pathway (Haas et al. 1995). The concentration of free ubiquitin rises as the transcription of genes that are responsible for the degeneration of cells is triggered by hormones (Haas et al. 1995). Utilizing an enzyme-linked immunosorbent assay (ELISA), we measured a considerable increase in the concentration of ubiquitin at the onset of muscle degeneration (Fig. 10). This increase reached its maximum 16–22 h after the adult moult and could be ascribed to the rise in ubiquitin concentration in M114c (Fig. 10B) by separately measuring ubiquitin in M114c and in M114a+b. Whereas the ubiquitin content in the combined parts M114a+b remains at a low level after the adult moult, M114c contains significantly more free ubiquitin 22 h after the adult moult than before the moult. The minor differences in ubiquitin concentration observed in the experiment with whole muscle homogenates compared with the experiment with homogenates of separated muscle parts (M114a,b compared with M114c) probably result from slight differences in experimental procedures used in these independently run experiments.
In agreement with the time course of degeneration in part M114c of the PA muscle (measurements of actin bands, ATPase and contraction kinetics), the maximum level of free ubiquitin (Fig. 10) coincides with the switch from a functional to a non-functional muscle part (Figs 3–5). This suggests that the hormone-induced programmed cell death in M114c requires the ATP-dependent ubiquitin pathway.
Discussion
Selective degeneration of M114c after the adult moult
In this study, we have shown that the adult PA muscle plays an important role in flight steering and consists of two distinct muscle parts, M114a and M114b, whereas the nymphal muscle includes one additional muscle part, M114c. This third muscle part undergoes hormone-dependent degeneration soon after the adult moult. A similar observation was made for the mesothoracic homologue, M85, which was not examined further.
Although M114c inserts at a different point of the wing hinge from M114a and M114b, the entire muscle is innervated by the most distal branch of nerve 4, N4D4. Also, at least one of the neurones with a thin axon innervates all three muscle parts. Thus, we unequivocally define M114c as a part of M114. M114c degenerates increasingly following the final moult, beginning at 16 h of adult life. Adipose tissue successively replaces M114c and, finally, the mature adult has only M114a and M114b. Ewer (1953) described a muscle, the accessory posterior tergopleural muscle (a.p.t.p.m.), that resembles M114c in its position and is similarly only found in newly moulted adult animals, yet never in older specimens. This evident analogy between anatomical position and development suggests that the muscle described by Ewer (1953) as the accessory pleural muscle actually corresponds to M114c.
Interestingly, a functionally homologous muscle to M114c has been described in adult M. sexta (Wendler et al. 1993). This pleurodorsal (PD) muscle consists of three parts in the meso- and metathorax (IIPD2u,m,l in the mesothorax; IIIPD2a,b,c in the metathorax) of newly hatched adult moths. The mesothoracic IIPD2l degenerates after the adult moult. Provided that the accessory pleural muscle and M114 correspond to each other, and provided that the PD muscle of M. sexta and M114 are homologous, the function and development of these muscles may be evolutionarily conserved. Degeneration after eclosion has been described for the PD muscle in M. sexta and for eclosion muscles in the flesh fly Sarcophaga bullata (Bothe and Rathmayer, 1994), the tsetse fly (cited in Bothe and Rathmayer, 1994) and Drosophila melanogaster (Truman, 1990).
Considering the supposed homology between the PD muscle in M. sexta and M114 and recalling the studies of Wiesend (1957) and Bernays (1972), a specific function for M114/M114c during moulting may be suggested. Wiesend (1957) and Bernays (1972) mapped nymphal muscles in the thorax and abdomen of Schistocerca gregaria and L. migratoria, which they reported to function only during moulting. These muscles are absent from nymphal stages other than the ones in which they are used and are never found in adult locusts.
Changes in anatomy bring about changes in physiology
Among others, Günzel et al. (1993) showed that a high ATPase activity (fast ATPase) and narrow sarcomeres coincide with fast contraction kinetics, whereas a low ATPase activity (slow ATPase) and wide sarcomeres coincide with slow contraction kinetics. This is also true for the pleuroaxillary muscle of Locusta migratoria. We show that M114a and M114b are fast-contracting muscle parts because of their narrow sarcomeres and their fast type-II ATPase. In contrast, muscle part M114c in the nymph and in the very young adult is a slow-contracting muscle part because of its wide sarcomeres and its slow type-I ATPase. This slowly contracting unit is lost when M114c degenerates, beginning 16 h after the adult moult. Thus, we assume that this muscle serves a special function during or shortly after the adult moult. On account of its physiology and its anatomical location, two functions for the nymphal PA muscle M114 are feasible: (1) M114 may function during and shortly after moulting, and (2) it may have a function in circulating haemolymph through the wing bud (Weeks, 1995).
Moulting behaviour has been described in detail in the locust Schistocerca gregaria (Bernays, 1972). That description of rhythmic muscle movement during shedding of the old cuticle matches our observations for the wing joint at which M114 inserts: it is slowly and firmly twisted at regular intervals. The position and the contraction features of M114, and especially of M114c, support the hypothesis that M114c may cause this observed twisting movement and thus aid in the shedding of the cuticle. The hypothesis that the predominant role of M114c is as an aid to moulting is favoured by the finding that M114c degenerates selectively after the final moult. The phenomenon of post-moult degeneration of moult-aiding muscles also occurs in other insect species, such as the flightless grasshopper Barytettix psolus (Arbas and Tolbert, 1986), the fire bug Dysdercus cingulatus (Nair and Prabhu, 1985), the flesh fly Sarcophaga bullata (Bothe and Rathmayer, 1994) and the tobacco hornworm Manduca sexta (Fahrbach et al. 1994).
It has also been reported for locust thoracic muscles other than M114 and for muscles in the locust abdomen (Ewer, 1953; Wiesend, 1957; Bernays, 1972). However, these muscle degenerations always affect an entire muscle, unlike the selective degeneration of M114c.
Like the abdominal muscles in pupal M. sexta, M114 may also function as a haemolymph pump throughout the larval stages. As in these muscles in M. sexta, the PA muscles may, because of their position and contraction features, act as an accessory heart-like organ that pumps haemolymph and may thus help to circulate nursing lymph components in the developing wing bud. In M. sexta, the abdominal muscles seem to function as a haemolymph-pumping heart substitute as long as the accessory heart structures at the bases of the wings are not fully developed (Weeks, 1995). In L. migratoria, this pumping function may also be extended to the first few hours after the adult moult, when the wings need to be expanded and inflated. This would match the role of the functionally homologous pleurodorsal (PD) muscle in M. sexta (Wendler et al. 1993). Regardless of the function of M114/M114c in the nymph, the changes in the muscle properties after the adult moult switch the nymphal function of the PA muscle to the adult function as a fast flight-steering muscle.
The pattern of innervation remains constant
Both nymphal and adult PA muscle are innervated by four types of neurones: two excitatory motor neurones with large cell bodies and thick axons; four neurones with small cell bodies and thin axons; the common inhibitor 1 (CI1) (nymphs, this study; adult locusts, Pflüger et al. 1986; Kutsch and Schneider, 1987; Wolf, 1990) and DUM3,4,5(a) (Stevenson and Meuser, 1997). The number of these listed cells corresponds to the number of axonal profiles in N4D4 revealed by transmission electron microscopy. Furthermore, our transmission electron microscopy results are in agreement with those of earlier transmission electron microscopy investigations (Pflüger et al. 1986) and with the investigations of Kutsch and Schneider (1987). Both groups also found axons with eight different diameters in nerve sections, but were not able to cobalt-stain the DUM3,4,5(a) cell. We succeeded in staining this octopaminergic neurone using the tracers Neurobiotin and biocytin (Meuser, 1996; Stevenson and Meuser, 1997). This diverse innervation pattern again suggests that the PA muscle could be multifunctional during the nymphal and adult moults. Since the staining pattern remained unchanged at the transition from nymph to adult locust, the degenerating muscle part M114c seems to play a unique role that is restricted to the nymphal stages and the moults, including the final imaginal moult.
Hormones, gene expression and ubiquitin
From the unchanged pattern of innervation, it may be deduced that the degeneration of muscle part M114c is not activity-dependent. For the flightless grasshopper Barytettix psolus, Arbas and Tolbert (1986) described presynaptic endings that survive the degeneration of a dorsolongitudinal muscle (DLM) during sexual maturation. Also, the biphasic degeneration of an eclosion muscle in the flesh fly Sarcophaga bullata (Bothe and Rathmayer, 1994) is neither triggered by retraction of motor neurones nor does it trigger neuronal degeneration. These authors therefore assume that the moulting hormones are the key signal for muscle degeneration. Hormone-dependent degeneration of muscles after the larval–adult/nymphal–adult transition have been described in a variety of insect species (e.g. Nair and Prabhu, 1985; Levine and Weeks, 1990). The present study clearly shows that the degeneration of M114c is hormone-dependent. Treatment with juvenile hormone analogues delayed or inhibited degeneration, whereas deprivation of juvenile hormone induced an earlier degeneration of muscle part M114c. According to other studies (e.g. Lockshin and Williams, 1964; Fahrbach et al. 1994) on neuromuscular restructuring during metamorphosis, the degeneration of M114c may therefore be defined as programmed cell death.
Steroid hormones have been shown to bind nuclear receptors, for example the superfamily of ecdysteroid receptors (ECRs) (Koelle et al. 1991), that are responsible for triggering the genetic machinery of cell death. In M. sexta, the death of thoracic central nervous system neurones can be observed at the pupal transformation. This cell death has been classified as genetically driven programmed cell death. Inhibition of protein synthesis by the application of cycloheximide or inhibition of mRNA synthesis by the application of actinomycin D resulted in a blocking or delaying of the death of the chosen indicator neurones in M. sexta (Fahrbach et al. 1994). By inhibiting protein synthesis with injections of cycloheximide and mRNA synthesis with injections of actinomycin D, we have shown that genetic processes are also involved in the degeneration of M114c. Both treatments inhibited the degeneration of M114c. Although we cannot determine how much of the injected drug was active at what time after injection (actinomycin D, for example, is a highly metabolism-resistant drug), it is still possible to deduce that moulting and muscle degeneration must be independent processes. The concentration used was sufficient to inhibit the degeneration of M114, but still allowed moulting. Thus, these two different processes may have a different susceptibility to disturbances of metabolism and therefore may involve partially different gene cascades. This deduction is warranted by the observation that injection of actinomycin D and cycloheximide after the adult moult has been completed still inhibited the degeneration of M114c. The genetic cascade inducing degeneration must therefore be triggered after or at the adult moult, since injection of cycloheximide before the moult is ineffective. The effectiveness of actinomycin D injected prior to moulting may be due to the drug’s metabolic longevity.
The programmed cell death observed in M114c is dependent on a decline in the titre of juvenile hormone but, since this decline occurs well before the moult, we assume that additional signals that set off the genetic clock are involved, ensuring the survival of M114a and M114b and triggering the degeneration of M114c. This additional signal may be the ecdysteroid-induced expression of early genes, a model that has been described for the ECR in D. melanogaster (Koelle et al. 1991) and M. sexta (Fahrbach et al. 1994). In these insects, the involvement of ecdysteroids in programmed cell death invokes a two-step process in which the binding of ecdysteroids to the ECR before moulting activates ‘early’ genes. The products of the latter are used to activate ‘late’ genes as soon as the titre of ecdysteroids declines after the imaginal moult (Fahrbach et al. 1994). A similar model may hold for the degeneration of muscle part M114c in the locust.
Role of the ubiquitin pathway in the degeneration of M114c
Ubiquitin serves as a polypeptide marker to label proteins for degradation (Yu et al. 1996). In M. sexta, the degradation of dorsolongitudinal muscles coincides with an increase in the level of ubiquitin (Haas et al. 1995). We have shown that this is also true for the degradation of M114c. The concentration of ubiquitin in M114c rises significantly 16–22 h after the adult moult. Recalling the course of degeneration and the contraction kinetics of M114, this coincides exactly with the time at which M114c becomes non-functional. This again confirms the argument that the hormonal signal, in combination with a second signal, triggers different genetic pathways in M114: the rise in ubiquitin level is only observed in M114c. Thus, ubiquitin may be a product of ‘late’ genes that control the end of a genetic cascade.
Our findings clearly show that M114c degenerates after the adult moult. This degeneration is hormone-dependent and can be defined as programmed cell death because of its dependence upon mRNA and protein synthesis. The cell death in M114c is mediated via ubiquitin-dependent protein degradation. Since the restructuring of the muscle occurs after the adult moult, we suggest that M114, and especially M114c, may play an important role during and after moulting. Further studies should aim to define the function of larval M114 and to identify the genetic pathways underlying the developmental restructuring of M114.
ACKNOWLEDGEMENTS
The authors would like to thank Drs G. Bicker and U. Müller, Berlin, for reading the manuscript critically and for helpful discussions. We are indebted to Dr P. A. Stevenson, Leipzig, for advice, discussions and proof-reading, and to Catherine Hall, Houston, for proof-reading the manuscript. We would especially like to thank Dr A. H. D. Watson, Cardiff, for carrying out transmission electron microscopy and Professor Dr K. Hausmann, Berlin, for generously allowing the use of the scanning electron microscope belonging to his group. We thank D. Bucher and C. Duch for helpful criticism. We are also grateful to CIBA Geigy, Basel, for the gift of fenoxycarb. This study was supported by the DFG, SFB 515 (project B6) and a graduate grant from Berlin (NaFöG) to S. Meuser.