Small Motor Axons in Orthopteran Insects: A Reinvestigation of the Innervation of the Femoral Retractor Unguis Muscle in a Stick Insect and two Species of Locust

The innervation of the femoral retractor unguis muscle has been examined by various authors in the locusts Schistocerca gregaria and Locusta migratoria (Hoyle, 1955; Usherwood & Machili, 1968; Rees & Usherwood, 1972; McDonald, Farley & March, 1972) and in the stick insect Carausius morosus by Godden (1972). The previous work in the stick insect suggested that the muscle is innervated by only four motoneurones, while in the locusts there appeared to be only two axons larger than 5 μm controlling this muscle, though some considerably smaller axons had been noticed with the large ones (Rees & Usherwood, 1972). Here, I demonstrate that small axons, which accompany the major axons in all three species, participate in the neuromuscular control of the retractor unguis. The first part of the paper shows, mainly from morphological evidence, that the innervation of the muscle in Carausius is more complex than previously suggested. The second part deals with the morphological and electrophysiological analysis of the innervation of the muscle in the two locusts showing that some small axons are excitatory, some are inhibitory, and some may have a neurosecretory function.

The stick insect Carausius morosus and the locusts Locusta migratoria and Schisto- . cerca gregaria were used. For light microscopy preparations were either fixed with Bouin’s and embedded in paraffin (section thickness 8 μm) or treated as for electronmicroscopy and 0·5–1·0μm sections made. Axons were traced through series of sections using camera lucida drawings. For electron-microscopy, preparations were dissected free, prefixed in 2% glutaraldehyde, postfixed with 2% osmium tetroxide and then embedded in Araldite or Spurr’s resin. Thin sections were contrasted with uranyl acetate and lead citrate and were investigated on an A.E.I. EM6 (Glasgow) or a Zeiss EM 9S (Konstanz) electronmicroscope.

‘Red’ and ‘white’ muscle fibres were distinguished by means of the succinate dehydrogenase reaction (cf. Elder, 1975). Preparations were surrounded by a viscous solution of polyvinyl alcohol and frozen deep in isopentane cooled in liquid nitrogen. ‘Sections, 20 μm thick, were cut on a cryostate, incubated for 10 min in the reaction mixture described by Nolte & Pette (1974) without agarose, fixed in 4% formaldehyde and, after washing in distilled water and drying, embedded in glycerol-gelatine. Photomicrographs from living material were taken either with a Zeiss photomicroscope, using a × 40 water immersion lens and Nomarski interference contrast, or with a Wild ‘Photo-Macroscope’.

During and after dissection the preparations were kept in a saline composed of KC1 10, NaCl 154, CaCl₂ 2 mM, buffered with 2 mM trismaleate at pH 6·8 for locusts, and KC118, NaCl 13, CaCl₂ 7·5, MgCl₂ 50 mM with 63·3 g/1 sucrose added, buffered at 6·6 with 2 mM trismaleate for the stick insect. With the latter sometimes a whole animal preparation was set up by fixing the prothorax and the coxa of one leg to the bath by means of plaster. This allowed motoneurone activity to be evoked via reflexes while simultaneously recording from the muscle. For the investigations on the locust muscles a bath was used which allowed the muscle to be rotated around its longitudinal axis into any desired position. Microelectrodes filled with 2 M K-citrate were used for intracellular recording and current passing. Insertion of more than one micro-electrode into the same preparation was facilitated by supporting the muscle with a little mound of agar (2% made up in locust saline). Glass suction electrodes, manufactured according to Pabst (1968) to give a close fit, were used for stimulating the motor nerve and occasional recording from the muscle. Positive pulses of 0·1–1·0 ms duration were employed with a maximal stimulus intensity of to V. Mechanical responses were monitored visually with the high magnification of a stereo microscope.

Two problems seriously complicated the electrophysiological investigations in the locusts: first, the thresholds of the two ‘fast’ axons (Usherwood & Machili, 1968) quite often became inseparable after some time (with or without experimentation). Second, attempts to block excitation-contraction coupling were not very successful. A détubulation technique, using 1 M ethylene glycol (Sevcik & Narahashi, 1972) was occasionally used, which reduced the membrane resting potential and greatly reduced the electrical response of the non-synaptic membrane to excitatory junction potentials, but contractions were not always completely blocked.

In order to explore the inhibitory innervation in the locusts 2× 10⁻⁸ M Na-glutamate was sometimes added to the saline to block completely excitatory transmission by desensitization of the postsynaptic receptors, or preparations were used which were paralyzed by means of Habrobracon venom (which is known to block solely the excitatory transmission; Walther & Rathmayer, 1974).

The electrophysiological results are based upon some 30 experiments with Schistocerca and about twice as many with Locusta. The ultrastructural findings concerning nerve terminals are based, in the case of Locusta, upon extensive morphometric investigations (Reinecke & Walther, to be published) on more than 20 preparations.

The femoral retractor unguis muscle (Fig. 1a) is composed of some 25 fibres arranged in two bundles (cf. Godden, 1972). These are separately attached to the apodeme (cf. inset at Fig. 1,a) which allows one to split the muscle distally into two functionally different parts. As in locusts (Usherwood & Machili, 1968; see also below) the fibres of the more distal bundle are thicker and have shorter sarcomeres than those inserting more proximally (cf. inset of Fig. 1,a). The muscle, in contrast to the homologous muscle in Locusta and Schistocerca (Fig. 1 b, c), is poorly tracheo-lated. It receives from the major leg nerve (nerve n.cr.; Marquardt, 1939) two branches each containing three large axons. In both the pro- and metathoracic leg, these axons leave nerve n.cr. via the distal branch. After entry into the muscle the largest axon (12–15μm) and the two other large axons (8–12μm) run in two separate nerve branches, the former supplying the distal, the latter the proximal fibre bundle.

The higher magnification of the electron microscope shows unequivocally that the three large axons are accompanied by a number of considerably smaller axons (Fig. 2). The distal nerve contains three small (1–2 μm) and five very small ( < 0·5 μm) axon profiles in addition to the three large ones (Fig. 2f). Possibly all of these small axons are side branches of axons running in nerve n.cr. (see Fig. 2 b, c, g). At least one of the very small axons seen in Fig. 2 f has originated in this fashion, suggesting that it is probably not a sensory axon. A comparable series of sections from another animal gave much the same picture, i.e. 3 large, 3 small and 7 very small axons within the distal nerve branch immediately after separation from nerve n.cr.

More evidence that the innervation of the retractor unguis muscle is more complex than suggested by Godden (1972) was obtained by recording from the muscle with suction electrodes, during spontaneous and reflexly evoked activity. Fig. 3 shows a typical extracellular record for the proximal fibre group, demonstrating the presence of at least five, probably six different types of synaptic responses. Recordings from the distal fibre group usually show only one type of synaptic response, similar to the largest seen in Fig. 3 (but not evoked from the same motoneurone), though occasionally a second much smaller type was recognized. This kind of recording may miss some type of small synaptic response or responses perhaps restricted to some region away from the sucked part of the muscle (cf. results with Locusta, p. 114). Therefore the figure of seven distinct motoneurones - to be inferred from such recordings - probably still represents a conservative estimate of the total population controlling this muscle.

Intracellular recording from fibres of the proximal group clearly indicated the presence of inhibitory innervation, revealed either directly as hyperpolarizing junction potentials (Fig. 4) or as an acceleration of the falling phase of an excitatory junction potential when the intensity of nerve stimulation was increased beyond a certain level. The various kinds of excitatory responses recorded intracellularly were found to be quite similar to those already shown by Godden (1972). There is, however, one type of e.j.p. in the thin fibres which is intermediate between the ‘fast’ e.j.p.s and rather small ‘slow’ e.j.p.s in the same fibres and which is not accompanied in a 1 : 1-manner by e.j.p.s in the tibial parts of the retractor unguis system. Most likely this unit plus the two ‘fast’ units (no. 6 and 7 of Godden) correspond to the three large axons (no. 1–3 ; Fig. 2) leaving nerve n.cr. via the distal branch to the femoral retractor unguis.

The fibres of the femoral retractor unguis muscles in Locusta and Schistocerca (Fig. 1 b, c) are more closely tied together than in the stick insect, but in vivo when viewed and illuminated from a suitable angle two bundles can be distinguished; one markedly striated and the other faintly striated. This corresponds to the classification into ‘red’ and ‘white’ fibres (Usherwood, 1967; Elder, 1975; cf. also below). The different appearance in vivo is due to the larger number of mitochondria at the Z-lines which makes them broader and more prominent in the ‘red’ fibres (Fig. le;Usherwood, 1967). The more quickly fatiguing ‘white’ fibres have fewer mitochondria and their sarcomeres are somewhat shorter (Fig. le;Usherwood & Machili, 1968). The insertion of the fibres on the apodeme, although somewhat variable, basically shows some grouping of the fibres into two (sometimes three) bundles (cf. insets at Figs. 1 b, c). The majority of the faintly striated fibres insert distally but a few of them more proximally, together with the group of intensely striated fibres.

Sections stained for succinate-dehydrogenase (cf. Methods) also distinguish between the two fibre types. ‘Red’ fibres stain more darkly (Fig. id; cf. also Fig. 35 in Elder, 1975, for an example in Schistocerca’). Fig. 5 shows a variety of sections from different preparations of Locusta and Schistocerca. On average there are about 15 fibres per muscle in Locusta and 17 in Schistocerca, the latter having somewhat larger diameters than the former. Although there is considerable variability (particularly in Locusta), about one half of the fibres in Schistocerca but only one third in Locusta belong to the ‘red’ type. If the average crossectional areas are considered, ‘white’ fibres are about 1·5 times as large as ‘red’ fibres in the case of Locusta and almost twice as large as the ‘red’ fibres in the case of Schistocerca.

The preferentially dorso-medial location of the succinate dehydrogenase-positive fibres agrees, as one would expect, with the preferential location of the intensely striated, more proximally ending fibres observed in vivo.

Fig. 6 shows cross-sections through motor nerves immediately after the point of separation from the major leg nerve (sb2 ; Campbell, 1961) which usually sends two branches to the retractor unguis muscle. There are several small diameter axons in addition to the two large ones (Fig. 6b, c, e,f). The same is found on sections of the nerve inside the muscle although sometimes it is difficult to distinguish clearly between the smallest axon profiles and glial structures (cf. Fig. 6f; for a detailed description of the ultrastructure of the peripheral nerves see Lane & Treherne, 1973). A further problem in determining the exact number of small profiles is that counts differ between sections taken at slightly different levels of the same nerve. These innervation of abdominal muscles in a grasshopper. He was able to show that a small axon may follow a tortuous course so that more than one aspect of it may appear within the same section. Possibly this occurs also in the retractor unguis nerve. The axon counts summarized in Table 1 therefore refer to the minimum numbers of small profiles found in different sections of a given nerve. Although there are some uncertainties with regard to the smallest axon profiles these axon counts suggest that there are basically four small axons in Schistocerca and five or six in Locusta. A comparable pattern of axon profiles (two large plus seven small) was found in a specimen from a 1st instar larva of Schistocerca.

Table 2 summarizes axon counts from occasional cross-sections of the nerve deep within the muscle. The minimum numbers of small profiles in Locusta was four (if preparation no. 4 is neglected) compared to three in Schistocerca. This again suggests that in Locusta at least one more axon than in Schistocerca controls the muscle. It should, however, be kept in mind that, because of the frequent nerve branching inside the muscle, such axon counts are probably not strictly representative of the number of neurones controlling the muscle.

Unlike in the stick insect (p. 103) both in Schistocerca (Rees & Usherwood, 1972) and in Locusta the large axons continue to run together within the muscle except for the most distal portion where they separate. Although in Locusta small axons innervate both ‘red’ and ‘white’ fibres, in Schistocerca the ‘white’ fibres seem to be controlled only by the large axons, most of them only by one, some by both of them (see below). It therefore was of interest to examine the number of axons in those branches of the nerve within the muscle to the ‘white’ fibres. In four such branches there were at most three axon profiles (1;2;2;3) which, in view of the neurosecretory innervation (cf. below), is quite compatible with the electrophysiological results.

At the ultrastructural level, it was not possible to distinguish synapses of functionally different motor axons. There were, however, throughout the muscle occasional nerve terminals which had large electron-dense grana amidst small electron-lucent vesicles. Such presumably neurosecretory axons (Osborne, Finlayson & Rice, 1971; Hoyle et al. 1974) tend to be associated with synaptic nerve terminals (Fig. 7). They are found between both types of muscle fibres in Locusta and Schistocerca, but rarely come into close contact with the muscle fibre (cf. Fig. 7). Dense cored vesicles also occur occasionally in synaptic motor nerve terminals of Schistocerca (Rees & Usherwood, 1972). In Locusta, however, these are less clear.

The innervation patterns in Locusta and Schistocerca are different and the electrophysiological results are therefore given separately.

The two large axons (cf. Fig. 6a) are of the ‘fast’ type each innervating predominantly one or other muscle bundle, that supplying the ‘white’ fibres hereafter termed F1 and the other, supplying the ‘red’ fibres, F2. Frequently, but not invariably, the F 1and F2 fields overlap so that some of the fibres, both ‘white’ and ‘red’, receive innervation from F1 as well as from F2. The junctional responses of ‘red’ fibres to stimulation of Fi (cf. example from Schistocerca in Fig. 8) or of ‘white’ fibres to stimulation of F2 range from about 1 to 15 mV (frequently around 10 mV) and thus are smaller than the responses of ‘red’ fibres evoked by F2 or of ‘white’ fibres evoked by Ft (e.j.p.s > 20 mV).

The small motor axons were stimulated selectively by antidromic stimulation of nerve 5b2, distal to the exit of Fi and F2. This is effective because the small axons are side branches of axons running further down nerve 5b2. This procedure consistently demonstrated an inhibitory innervation of all the ‘red’ fibres and of a proportion of the ‘white’ fibres which varied in different preparations from only a few to all ‘white’ fibres. In many, but not all, of the fibres two inhibitory junction potentials with slightly different thresholds can be evoked (Fig. 9). Both inhibitory axons also supply fibres in the flexor tibiae (Fig. 9) and the tibial parts of the retractor unguis system, but not extensor tibiae fibres. The inhibitory innervation was distributed over the whole length of individual ‘red’ fibres which were checked for this point by repeated recording from different regions but this may not be generally true for the inhibitory innervation of the ‘white’ fibres.

In undamaged fibres with resting potentials of about – 58 mV the i.j.p.s are usually quite small (< 1 mV), often depolarizing. They tend to be smaller in the ‘white’ than in the ‘red’ fibres. Demonstration of i.j.p.s in the ‘white’ fibres, in fact, is often only possible if the membrane potential is shifted by several millivolts via current injection and if the nerve is stimulated by a short train of impulses to achieve facilitation and summation of the i.j.p.s. In a variety of preparations in which excitatory transmission had been blocked either by application of 2× 10⁻³ M glutamate or by aralysis with Habrobracon venom (cf. Methods) the inhibitory responses were compared for antidromic and orthodromic stimulation of nerve 562. No difference was noticed for the two modes of stimulation nor did orthodromic stimulation evoke i.j.p.s in fibres where antidromic stimulation failed to evoke them. This rules out the possibility of an additional inhibitor which might leave nerve 5b2 together with axons Fi and F2 and therefore would be missed on antidromic stimulation. In those ‘white’ fibres which are clearly lacking i.j.p.s, GABA receptors can be present: at least in some of them topical application of GABA (5 × 10⁻⁴ M) leads to a reduction of input resistance by 30% or more.

At a usually higher threshold than that required for i.j.p.s an e.j.p. can be evoked by antidromic stimulation of nerve 5b2 in the femur or in the tibia. This e.j.p. is restricted to some of the ‘red’ fibres and within these usually to their distal half. The amplitudes of these ‘slow’ e.j.p.s are maximal in a short region one to several millimetres from the distal end of the fibre (Fig. 10b). Depending on the fibre and on the preparation they vary from a few to some 20 mV, the large ones giving rise to a graded electrically excited membrane response (Fig. 10a). Accordingly many preparations near their distal ends respond to a single nerve stimulation with no or a barely noticeable twitch and with a smooth contraction to repetitive stimulation (e.g. 20 Hz) but some already give a strong twitch upon a single stimulus.

In several cases two e.j.p.s could be evoked antidromically, one at considerably higher threshold than the other. In the example of Fig. 10b the smaller e.j.p. appeared 20 ms after the larger e.j.p., indicating that it is produced by a particularly small axon. This second ‘slow’ excitatory axon innervated only some of the ‘red’ fibres - not necessarily the same as those innervated by the first - and within these only some region in the distal half. In those fibres which exhibited both e.j.p.s the regions of maximal amplitudes did not coincide (Fig. 10b).

In a few cases neither e.j.p.s nor contractions could be detected on antidromic stimulation whilst i.j.p.s were present. Either the ‘slow’ excitatory innervation was lacking or perhaps was restricted to those centrally located ‘red’ fibres (cf. Fig. 1 d) from which recordings were not normally made. On the other hand on rare occasions ‘slow’ e.j.p.s could be antidromically evoked not only in ‘red’ but also in some ‘white’ fibres.

The basic pattern of innervation which emerges from these findings and from the ultrastructural results is summarized schematically in Fig. 11a.

The pattern of retractor unguis innervation in this locust is similar to but somewhat simpler than that in Locusta and seems to be less variable. The response to stimulation of the two large ‘fast’ axons (previously described by Usherwood & Machili, 1968) indicates, as in Locusta, a partial overlap of the fields of innervation of Fi and F2 (Fig. 8). A ‘slow’ excitatory innervation was absent except in one preparation. There are two inhibitors which are common with the flexor tibiae. I.j.p.s, present in all of the ‘red’ fibres, cannot usually be evoked in the ‘white’ fibres by either anti- or orthodromic stimulation. Only one or two ‘white’ fibres did exhibit i.j.p.s on rare occasions. As in Locusta, ‘white’ fibres which clearly failed to give i.j.p.s responded to topical application of 2 × 10⁻⁴ M GABA with an appreciable reduction of the input resistance (which has also been noticed by Clark, Gration & Usherwood, 1979). The basic pattern of innervation of the Schistocerca retractor unguis muscle is shown in Fig. 11 b.

In addition to the two large ‘fast’ axons innervating the metathoracic femoral retractor unguis muscle there are at least four much smaller axons in Schistocerca and five or six in Locusta (Fig. 6; Table 1). Even higher numbers of small axons were demonstrated in the stick insect and at least one inhibitory and one more excitatory axon should be added to the four motoneurons described by Godden (1972, fig. 14). Of the four axons in Schistocerca, one is probably neurosecretory on ultrastructural evidence and two are common inhibitory (possibly branches of two previously described inhibitors to the flexor tibiae; Burrows & Horridge, 1974). In Locusta, one is probably neurosecretory, two are common inhibitory, and there may be two excitatory. This leaves at least one axon more than can be functionally accounted for in both species. This discrepancy could arise if there is more than just one neurosecretory axon or if one neurone escaped the electrophysiological analysis. A further possibility is that there are afferent fibres although there was no indication for sensory structures within the muscle and recordings from the retractor unguis nerve and nerve gb2 failed to demonstrate afferent nervous activity. Ideally one would like to determine the number of efferent neurons by relating the number of axon profiles in the nerve branches to the number of somata backfilled with cobalt. In practice, however, this is not feasible because fibres additional to those innervating this muscle also fill due to the short length of retractor unguis nerve branches (cf. Wilson, 1979).

Small axons (i.e. ⩽2 μm), alongside much bigger ones, have been shown before in motor nerves supplying other muscles in Schistocerca (e.g. Piek & Mantel, 1970; Shepheard, 1973), or other insects, for example a cockroach (Faeder & Salpeter, 1970), a fly (Anderson, 1978) or a beetle (D. Ballantyne, personal communication). Small axon profiles in the locust retractor unguis nerve have been previously overlooked (Locusta:Hoyle, 1955; Schistocerca:McDonald et al. 1972) due to reliance on light microscopical analysis which no longer can be regarded as wholly adequate. It is easy to overlook the inhibitory and ‘slow’ excitatory motoneurones described herd because orthodromically elicited i.j.p.s and ‘slow’ e.j.p.s tend to be masked by very large responses from the ‘fast’ units.

Not only can axon counts be confused by the tortuous course of small axons (Tyrer, 1971), but additional complications may arise from multiterminal innervation. Axons may branch before the nerve branches resulting in dual representation of one neurone in some transverse sections. This possibility has been previously considered in the retractor unguis nerve of Schistocerca by Rees & Usherwood (1972; cf. legend to their fig. 4). In the present investigation evidence for this phenomenon was seen in one series of sections for the light microscope from the stick insect. In another case (in the stick insect) an EM series clearly showed that one of the large axons within a retractor unguis nerve divided into two parts which merged again after a short distance and this could also occur in small axons. In addition to such ‘caprices’ it is possible that neuromuscular mismatching occurs in ontogeny so that in the adult some motoneurones innervate ‘wrong’ muscles but without functional synaptic contacts.

Retractor unguis muscle fibres are long (⩾7 mm in female locusts) whereas their length constant is only 1 to 2 mm and their ‘length constant’ for synaptic potentials only 0·5 to 1 mm (unpublished results; cf. also Gage & McBumey, 1973). Thus it is possible to detect gradients in amplitude of junctional responses along individual muscle fibres indicating discontinuous motor innervation. The ‘slow’ axon in Locusta usually caused e.j.p.s only in the distal half of the ‘red’ fibres, and in cases of double ‘slow’ innervation (cf. Fig. 10) the regions of maximum responses were not quite the same; inhibitory responses, however, were not locally restricted within the same fibres. A proximal to distal gradient of e.j.p.-size has recently been described for single lobster muscle fibres (Meiss & Govind, 1979). The ‘white’ fibres of Schistocerca retractor unguis apparently lack nerve endings distally in the last millimetre (Gration, Clark & Usherwood, 1979). Multiterminal innervation then by no means implies that the synaptic contacts of a given neurone extend over the whole length of a muscle fibre and even when they do, they need not necessarily be equally spaced or act with equal efficiency.

The functional significance of the inhibitory and presumed neurosecretory axons in the two locusts remains to be established. As far as ‘trophic’ effects are concerned, recently it has been demonstrated that in the ‘red’ fibres of the Schistocerca preparation the persistence of inhibitory innervation does not prevent the degenerative changes which follow severance of the excitatory motor axons (Clark et al. 1979).

A possible function of the neurosecretory axon(s) may be illustrated by reference to the extensor tibiae muscle in Schistocerca. In this muscle, a neurosecretory axon from a dorsal median unpaired neurone (DUMETi; Hoyle et al. 1974) has been demonstrated to potentiate, if stimulated, the force of contraction generated by the ‘slow’ excitatory motoneurone and to increase the rate of muscle relaxation (O’Shea & Evans, 1979). This modulation involves probably both the pre- and postsynaptig side, and there is strong evidence that it results from the release of octopamine from the neurosecretory terminals.

Part of these investigations were carried out during a postdoctoral fellowship in the Department of Zoology at the University of Glasgow. I thank Prof. Dr R. Newth and Prof. Dr P. N. R. Usherwood for their hospitality. I wish to thank Mrs C. Dittrich and R. Eberle for their enduring technical assistance, Dr P. M. Nemeth and M. Stieb for help with the histochemistry, Prof. W. Rathmayer, Prof. P. N. R. Usherwood and Dr D. Ballantyne for criticizing the manuscript and the latter also for kindly correcting the English. This work was supported by the Deutsche Forschungsgemeinschaft (postdoctoral research grant and Sonderforschungsbereich 138).