Distributions of Nerve and Muscle Fibre Types in Locust Jumping Muscle

The basic innervation and neuromuscular physiology of the acridid extensor tibiae (ETi), the jumping muscle, was first described by Hoyle (1955a, b). The muscle continues to be a major target for the study of insect neuromuscular physiology (reviews by Usherwood, 1967, 1977; Hoyle, 1975). Recently, it has been shown that there is an intrinsic rhythm of contractility (IR) in the muscle (Hoyle & O’Shea, 1974) and that this rhythm is suppressed by release of a neurohumoral agent from specific terminals within the muscle (Hoyle, 1974), which appears to be octopamine (Hoyle, 1975; Hoyle & Barker, 1975). This is released reflexly before major movements involving the extensor tibiae occur (Hoyle, 1978). The agent is released from a fourth axon that innervates the muscle in addition to the fast excitor (FETi), the slow excitor (SETi) and the common inhibitor (CI) (Hoyle, 1955 a; Usherwood & Grundfest, 1965; Burrows, 1973). The newly discovered axon is termed the dorsal, unpaired median axon supplying the extensor tibiae (DUMETi).

In a recent account of the fine structure of component muscle fibres, it was shown that two kinds of muscle fibre can be distinguished in the muscle, that were designated as ‘phasic’ and ‘tonic’ (Cochrane, Elder & Usherwood, 1972). The experimental basis for this distinction was the mechanical response to immersion in a high potassium saline. If the contraction was quick but fatigued rapidly the fibres were termed ‘phasic’ whereas if it was slower and long-lasting they were termed ‘tonic’. Tonic fibres were described as being confined to a small, proximally located bundle, the remainder, and majority, of the fibres being considered phasic. These muscle fibres were said to be innervated by SETi and CI. An outside, fan-shaped proximal bundle was figured (Cochrane et al. 1972-fig. 1) as being innervated by FETi and SETi but not CI with the bulk of the muscle fibres being innervated by only FETi.

This distribution is somewhat different from that described earlier for Locusta migratoria (Hoyle, 1955a, 1957) in which SETi and also the third axon (CI) were found to innervate fibres located throughout the muscle. The discovery of the fourth axon (DUMETi) raises various functional and anatomical questions. It should be borne in mind that DUMETi neurosecretions could exert other effects than the known one of suppressing IR. The details of its distribution in the muscle need to be known since they might afford additional clues regarding its functions.

The present work was initially undertaken to determine the distribution of DUMETi, but added the additional goal of investigating more thoroughly the distribution of all the axons, muscle fibre types and synaptic events in the muscle and comparing the mechanical responses of different regions to neural stimulation. Fibres innervated only by FETi will be referred to by the letter F, by SETi, S; by CI, I, and by DUMETi, D. Ones receiving both SETi and FETi will be termed SF, and so on.

What correlates are there between ultrastructures and contractile properties of the various combinations? The total possible range characterized by innervation comprises 12 types: F, FD, FS, FSI, FSD, FSID, FID, FI, SD, SID, SI, S. Various other basic features could be combined with innervation patterns besides myofilament architecture: especially biochemical differences. Also, colour tint has been used to characterize some insect muscles (Dresden & Nijenhuis, 1953; Becht, 1959; Usherwood, 1967; Elder, 1975), and crustacean muscles (Hoyle, 1973). The basis of these distinctions is the presence or absence of large numbers of mitochondria which, because of a strong absorption band at green wavelengths create a red colour illusion. In the present work only structural and contractile features were followed.

Cochrane et al. (1972) found distinctive ultrastructures to be associated with two types of fibre they recognized, the features being similar to those associated with functionally slow (= tonic) and fast (= phasic) muscle fibres in crustaceans (e.g Hoyle, & McNeill, 1968). In the latter, a third, or intermediate type is also recognized (Atwood, Hoyle & Smyth, 1965) but this is for descriptive convenience. Intermediates occupy a central position in a continuous spectrum between the slow and fast extremes. There are no abrupt lines of demarcation. It is evident that the ultrastructure is strongly correlated with innervation type. An aim of the present work was to examine further the extent to which structure and innervation are correlated. Altogether, four aspects were examined: general anatomy of the muscle, nerve axon supply to different regions, contractile responses of different regions, and structure of muscle fibres at the light and electron microscope level.

The species studied were the African desert locust, Schistocerca gregaria Forskål (S. americana, Dirsch, 1974), obtained from a stock maintained at the Department of Zoology, University of British Columbia, the American locust S. vaga (S. nitens nitens, Dirsh, 1974), reared in our laboratory from a stock kindly supplied by Professor H. Fraser Rowell, of the University of California, Berkeley, and the Florida lubber grasshopper Romalea microptera collected by a commercial supplier.

Animals were embedded in dental wax in a manner permitting tension recording from anywhere along the ETi apodeme. The metathoracic ganglion was exposed by cutting away overlying cuticle and pinning to a rigid, wax-covered platform that was micromanipulated beneath it. Small hook electrodes were placed as follows: under N5 to excite FETi, N3b to excite SETi, and N3C to excite CI. DUMETi was excited intrasomatically via a microelectrode, from the dorsal surface. Intracellular electrodes filled with M-K acetate were used to record from muscle fibres.

Tension was monitored by clamping a piece of apodeme in fine forceps tips attached directly to the peg of a micromanipulated RCA mechanoelectronic transducer. The clamped position corresponded to a femero-tibial angle of approximately 40°.

Some preparations were examined both in their own haemolymph, as well as after bathing in running standard locust saline (Hoyle, 1953) at 22°C. All results, except where stated, were made after more than 1 h of equilibration. Following physiological testing, single muscle fibres from the regions tested were examined by Nomarski interference microscopy. Fibres from the same regions were fixed in situ in fresh preparations for examination by electron microscopy, using standard procedures.

The branching patterns of the four axons were determined by electrophysiological recording. Selected branches were prepared for electron microscopy and thin sections examined.

A drawing of the femur of S. gregaria, dissected so as to show the component muscle units, is shown in Fig. 1. The corresponding anatomies of S. nitens and R. microptera are not significantly different. A plan view is shown in Fig. 2, with various parts labelled. These drawings also summarize some of the findings. They include a pair of small distal muscle units that are separate from the rest, and which were designated 135c and d by Albrecht (1953). The drawing gives the true number of muscle units, and also represents the proximal region accurately. The drawing presented by Cochrane et al. (1972), their Fig. 1, is a diagrammatic, not an accurate, representational drawing and is misleading because it contains an error. The proximal end of the apodeme is shown curving towards the inside of the femur, not the outside, which it always does. I therefore found it difficult to determine which was the bundle of very short, proximal fibres innervated by SETi but not FETi and having a different ultrastructure that they frequently refer to. It was important to know this because they considered this bundle to comprise only S fibres, and also to be innervated only by SETi.

Their slow bundle appears to be a part of a large, fan-shaped block of fibres that I shall refer to as the ‘fan’ which owes its shape to the proximal asymmetry of the femur, and comprises five muscle units. The fan is more than twice as large as its inner partner, which comprises only two units, one dorsal, one ventral, as do the rest of the subdivisions of 135a and b. The asymmetry causes the apodeme to be pulled towards the outside during contraction.

It proved to be easy to separate one small bundle of very short muscle fibres, the shortest in ETi, from the fan because they are attached to an extension of the main apodeme. This can very easily be seen in S. nitens where the apodeme is bright blue in colour. This little bundle contains fast and intermediate, not slow, muscle fibres. Just distal to this bundle is a cluster of slow fibres innervated by SETi and CI but not FETi. The rest of the fan could not easily be divided, so was tested as a whole. It will be referred to as region outside a, and its counterpart as inside a. The next three pairs of units constitute region b. Outside b units are innervated by a conspicuous nerve branch. Then follows the major fork in the extensor nerve. The first three units beyond the fork are the c units, the next three the d, followed by the e and finally four distinctly smaller units comprising much thinner muscle fibres, the f units.

Rise-times to peak for isometric twitches of the 12 different bundles tested varied over a 2 + fold range, from 45 to 120 ms, with corresponding decay times to baseline of 200−500 ms. The shortest times in all species occurred in outside b units (Fig. 3). Their partners, inside b units, were almost twice as slow. For the remaining units no consistent differences occurred, except that the more distal bundles were slightly slower than proximal ones. No major difference between inside and outside bundles was seen except for that noted in the b units.

All three species were found to contain the two extreme types of fibre described by Cochrane et al. (1972). The easiest way to distinguish them is by examination of individual, whole, live fibres in the phase-contrast or interference microscope (Fig. 4). Fibres innervated by SETi and CI but not FETi which are slow, or ‘tonic’ have well-aligned Z bands with conspicuous dark bands on either side that are due to large mitochondria. Fibres innervated only by FETi have virtually no lateral alignment of myofilament clusters so that it is difficult to discern striations. Weak bands on either side of the Z bands are due to small mitochondria.

Fibres innervated by both FETi and SETi resembled the slow type more closely than the latter when examined by the light microscope. In the electron microscope it could be discerned that their ultrastructural characteristics ally them more with the fast fibres except that there is 30−50% less SR and there are fewer dyads per unit cross-sectional area in these fibres. These will be referred to as intermediate-type fibres.

Slow fibres were not always homogeneous in ultrastructure throughout their cross-section, generally having even less SR and fewer dyads in their cores than at the periphery. In extreme cases, of which an example is shown in Fig. 5, slow fibres had very little SR and almost no dyads in their core regions. This situation should be contrasted with that in a fast fibre (Fig. 6) in which both are abundant. These features could be significant functionally. During a twitch, or brief tetanic excitation by the junctional potential/small graded response combination evoked by SETi, only the peripheral myofilaments are likely to be excited to contract. Since the periphery is relatively well-endowed with SR, it will relax relatively quickly. By contrast, during a prolonged excitation the core of the fibre will be activated also, by inwards diffusion of activator calcium. But relaxation of the core can occur only slowly-probably much more slowly than the periphery. Tests of the relaxation rates of slow-type fibres following different stimulus regimes, to be described below, showed that this is indeed the case.

Unfortunately no bundle of muscle fibres was found that was considered to be comprised entirely of intermediate-type fibres, but outside region d comprises mainly intermediates (see Fig. 2). Both it and bundles of fast fibres, unlike slow fibres, had relaxation rates that were independent of the stimulus regime.

During the investigation it became apparent that fibre diameters are markedly smaller in the most distal bundles. The fibres have irregularly shaped cross-sections, so to accurately estimate their average thickness they were first fixed in situ with the tibia held in the middle of its operating range. Selected bundles were dissected out after fixation and cut transversely to their long axis. The sections were photographed, projected onto a grid of squares equivalent to 1 μm² and the mean areas determined. In any particular region the range of fibre cross-sectional area was large, a fivefold range being common. A region was selected for each in which there was low distortion in a contiguous cluster, and all fibres in the cluster were measured. Although the individual fibre type could usually be recognized from these sections, values for a bundle were not separately measured but lumped together. The mean value was then translated into the diameter of a circle of equivalent area. These diameters are presented in Table i and on Fig. 1 for the various bundles. A preliminary estimate of the relative diameters of slow, fast and intermediate-type fibres was also made, but no significant differences were found when due allowance was made for regional size differences.

The average diameter was found to increase progressively from the proximal border to a peak $13$ distant and then to decline progressively passing distally (Fig. 1). However, the individual diameters vary widely, over a fivefold range, in some regions. Also, some fibres have bizarre, complex shapes, with narrow tongues extending from roughly circular centres. The mean diameters closely follow the leg contours, the greatest being at the point of mean maximum thickness of the femur. As the leg abruptly narrows distally, so do the fibre diameters decline. Also, the mean diameters of outside fibres was 16% larger than inside fibres for the muscle as a whole and 21 % larger for fibres distal to the Y. These differences match size differences associated with the asymmetrical shape of the femur. For proximal regions the muscle fibres in the dorsal set of bundles were of about 40 % smaller average diameter than the corresponding ventral bundles. The largest fibres are in outside region b, the smallest in the distal accessory muscles 135c, d and their nearest neighbours proximally in the main muscle 135a, b. In the fan, slow-, fast-and intermediate-type fibres were found, all about 100 μm in diameter.

No relationship between fibre type and diameter was found in regions containing mixed populations. Independent studies by Kate Skinner (personal communication) on the prothoracic leg, which is much thinner and has little variation in thickness along its length, show that here the fibres are all of nearly constant small diameter. Leg muscle fibres, it seems, enlarge into the space available to them during development.

The extensor excitor nerve contains four axons: FETi, SETi, CI and DUMETi (Fig. 7). It enters the femur dorsally, in contact with the distal rim of the trochanter. From there it passes ventrally to curl into the gap between the flexor and extensor tibiae. It then passes outwards, following the line of the extensor apodeme. Immediately after entering the femur it gives off its first major branch, which enters the large fan-shaped outer bundle. At about the same point it also gives rise to a major branch directly dorsally, to supply the deeper proximal bundles. Next is a branch to the inner counterpart of the fan.

Now the nerve gives off a stout branch that supplies only outer bundles and their dorsal partners. The branchlets from this pass backwards before entering the muscle units from their posterior surfaces. The number of units supplied by this branch is not constant, but is commonly three, rarely two or four. If there are fewer than three, or if the major fork occurs more distally, additional, small branches are given off the main nerve to supply the other units. The matching inner units are commonly supplied by two separate branches, the more proximal of which innervates two units and the distal one or two.

At a point about one-third down the femur, a Y fork occurs. This was common to all the saltatory orthopterans examined. This fork does not occur in the pro-and meso-thoracic legs, or in non-jumping orthopterans, suggesting that it is associated with the functional specialization of jumping. Because of the division it is possible that innervations of inside and outside halves of the muscle are different, so special attention was paid to this possibility.

After the major fork, innervation of each remaining vertical pair of bundles is simple. A single branch occurs just distal to each and passes dorsally to supply the bundles from their posterior faces. There is a major difference after the last pair of bundles of the major part of the muscle (135a, b): only the inside branch continues, moving along the apodeme to form a Y branch just proximal to M135C, d which it innervates by the two branchlets. This feature, like the others described, was common to all specimens of all the acrididae examined. It did not occur in some non-saltatory orthopterans examined, such as crickets, and stick insects.

A summary drawing of the axon distributions, with micrographs of selected regions, is shown in Fig. 7. All branches from the main nerve trunk, except those illustrated, contain FETi, SETi and CI. This does not mean that each axon supplies all muscle fibres of the bundles the branches enter. In mixed bundles, such as the fan, there is a complex local re-distribution, in combinations: F, FS, SCI and FSCI as determined electrophysiologically. By contrast all muscle fibres in 135c, d receive both SETi and CI terminals. Peculiarities of the distributions of individual axons are such that each will be described separately.

FETi is nearly circular in cross-section at its point of entry into the femur and about 18 μm in diameter in an adult male S. gregaria. It does not maintain either a constant shape or a perfectly smoothly tapering profile. It is flattened over short regions and waxes and wanes in thickness as it passes distally. It travels close to SETi and the two large axons twist around each other. Immediately after the major fork FETi is abruptly narrower, but widens again to 11–12μm in mid-femur in both inside and outside nerve branches.

FETi was found in every branch given off by the main trunk except the continuation of the inside branch after the end of 135a. It was found consistently to innervate the proximal outside (a) bundles which in the Cochrane et al. (1972) drawing were shown as not receiving it.

At its point of entry into the femur SETi is not much smaller than FETi, 16 μm in an average male 5. gregaria, and also nearly circular in profile. Like FETi, it is usually oval in section distally. SETi branches were found in all the nerve branches except one. That is the conspicuous branch that supplies outside muscle bundles in the b region. This contained only FETi and DUMETi in S. gregaria and also R. microptera. SETi tapers progressively to about 6μm by the last muscle unit of 135b where, together with CI only, it innervates 135c and d.

As it enters the femur CI diameter is about 6 μm. It branches at all the same places at which SETi branches. After the major Y branch the inner and outer CI branches are about equal in diameter, at 3·5 μm. Thereafter, the outer (a) branch tapers gradually to 1·8 μm at the terminal outer bundle. The inner (b) branch also tapers, to 2·5 μm, before going along with SETi, to innervate both the distal accessory extensors 135c and d. Since these small muscles are outside and inside respectively, it would have been logical to expect continuation of the outside and inside branches to innervate them. But in all of over 100 preparations examined, only the inside branch went on to innervate both muscles.

CI was found to accompany SETi throughout the muscle. These two axons leave the ganglion together in N3b, and only in the coxa join with N5 for a very short distance before emerging, having been joined by FETi and DUMETi, as the extensor tibiae motor nerve. SETi and CI remain associated together, and as far as could be ascertained, CI branches whenever SETi does, entering all the same nerve trunks.

DUMETi enters N5 with FETi. In the coxa they are joined by SETi and CI, which travel together in N3, then the four axons leave, with no others, in a branch that supplies ETi. The DUMETi diameter at its point of entry into the male S. gregaria femur is 3 μm. It tends to travel close to CI, rather than FETi, with which it is associated in N5, but when it emerges it is with FETi. No branches containing SETi and CI were found that also contained DUMETi. DUMETi divides into branches of equal thickness at the Y, but from that point on a difference occurs. In the inside branch it continues, tapering gradually, down to the last bundle. By contrast, in the outside fork it tapers relatively rapidly and could not be seen in sections made distal to the middle of the fork.

The most conspicuous DUMETi nerve twig occurs in the outside branch that supplies the three or four muscle units in the b region accompanying only FETi. It was by following this branch serially at the electron microscope level, that dense-core vesicle-filled DUMETi terminals were located (Hoyle et al. 1974). The DUMETi branch to bundles in the 135 outer, b region is 1·3 μm in diameter. No other branch to a muscle as large as this occurs. However, in the outer fork, DUMETi is conspicuous to the end.

It does not follow that there are functional neuromuscular junctions of all four axons on all muscle fibres to which they travel; even within a small bundle, selective distribution occurs. Knowledge of the distributions of FETi, SETi and CI terminals can therefore only be obtained by intracellular recordings from single muscle fibres that ideally should be by recording from all fibres in a bundle. But regrettably this could not be achieved; sampling was especially difficult for the dorsal bundles. Also, the distribution of DUMETi terminals cannot be determined physiologically: they can be identified only in electron microscope sections. Our knowledge of the locations of these terminals is therefore very incomplete. No DUMETi terminals were found until ventral, middle, outside region b was serially sectioned, after electrophysiological tests had shown that a branch of DUMETi innervates this unit (Hoyle et al. 1974).

The counterpart of the ventral region of the fan, inside region a, responds to both FETi and SETi, though to the latter very weakly. The proximal end of the apodeme curls outwards when SETi is active tetanically as a result of tension imbalance. Inside b units respond weakly to SETi, only partially counteracting the strong outwards pull of the fan. All units except outside b developed some tension in response to SETi. Only a small fraction, estimated to be 5−15% of their component muscle fibres are innervated by SETi. Most of these responded also to FETi. Although fibres innervated by SETi are not common in the main body of the muscle, together they make the major contribution to SETi-developed tension. The two small muscle units 135c, d were found to be exclusively innervated by SETi and CI.

It is important to bear in mind not only that there are different types of muscle fibres, but also that there are likely to be systematic differences in neuromuscular transmission to the different types. These can be due to a combination of differences in synaptic structure and release with differences in postsynaptic response. Details of junctional transmission were found to vary widely, even for the synapses of a given axon on muscle fibres of similar type. Junctions differed in the extent of facilitation, which varied from one extreme, essentially infinite facilitation, i.e. no detectable e.j.p. to a single shock, to the other in which there was no increase in size of e.j.p. with repetition. Secondly, they differed in the rate of facilitation at a given frequency. Some junctions attained maximum j.p. size after as little as 200 ms at 15 Hz, whilst others with similar sizes of initial and post-facilitation j.p.s, were still growing after 5000 ms. Thirdly, there were marked differences in the maximum height of the j.p. after full facilitation, ranging from barely 1 to 44 mV. Fourthly, there was a wide range in the rise-time or decay rate of individual j.p.s, the total durations ranging from 30 to about 300 ms.

Slow muscle fibres were found to be either weakly electrically excitable or electrically inexcitable. Even large e.j.p.s in them gave rise to graded responses of at most 10 mV. Fast muscle fibres gave large graded responses that commonly overshot zero. Intermediate muscle fibres at the fast end of their range, that were innervated by both FETi and SETi, gave large graded responses. In a few, there was overshoot in response to both FETi and SETi. At the other extreme neither showed overshoot.

Even though the general features were common, not only for a given species, but between genera, specific details of the electrical and mechanical responses recorded from different individual animals differed considerably. This was nowhere more apparent than in the mechanical response to SETi. In about half the preparations tested before perfusion with saline, the mechanical response to a single SETi impulse was either nil, or a tiny twitch of less than 10 mg peak tension. By contrast, in a few preparations the mechanical response to SETi changed dramatically upon bathing in saline. There was no increase in the maximum tetanic force, but the rates at which this was developed, and the response to a single impulse, were greatly enhanced (Fig. 9). All responses illustrated after Fig. 9 were from preparations equilibrated for at least 1 h in saline. The mechanical response of F fibres to FETi dominate the overall response of ETi and can therefore be considered relatively well understood. By contrast, the responses to SETi can be considered to be poorly understood because they comprise not only SETi only-innervated slow fibres, but also SETi + FETi-innervated fibres which are of fast or intermediate type. It was deemed particularly necessary to try to understand these differences.

The electrical responses to the fast axon have been extensively studied (del Castillo, Hoyle and Machne, 1953; Hoyle, 19556; Cerf et al. 1959). They consist of large (estimated 30−50 mV) distributed junctional potentials initiated at closely spaced, multiterminal junctions and large, graded, overshooting active membrane responses. These in turn give rise to a large twitch. The distinctions between fast, slow and intermediate muscle fibres have only recently become apparent and the present paper is the first to report attempts to segregate mechanical response of the three types. Most published intracellular records of FETi action potentials must have been from fast type fibres, though several were from fibres in which a SETi junctional potential occurred in the same muscle fibre. The latter are likely to have been from intermediate type fibres, but no author noticed any differences in FETi synaptic transmission, so electrically they behave similarly. Mechanical traces in published records have always been the summed responses of different types of fibre that happened to be innervated by the axon in question. Small, but significant differences between FETi contraction and relaxation times of muscle units in different region were found in the present work. There were also notable differences in mechanical facilitation of twitches, and in rates of fatigue. These differences will be described in relation to location.

During the present work it was found that the distal-most muscle bundles attached to the extensor tibiae, the accessory extensors 135c and d (Albrecht, 1952), are innervated by only SETi and CI. There are 8−12 fibres in each : they do not show the intrinsic rhythm, which can interfere with mechanical studies made on proximal slow fibres. Thus 135c and d are the better models for studying the functioning of this type of muscle fibre. The SETi junctional potentials in 135c and d exhibit a wide range of magnitudes and rates of facilitation, as is characteristic also of SETi junctional potentials in fibres located throughout the body of the muscle.

These two little muscle units always gave a small, brisk twitch in response to a single SETi impulse after equilibration in saline. Their larger j.p.s gave rise to graded responses. Mechanical threshold was tested for single muscle fibres using two intracellular electrodes, one to pass current, the other to record potential. By visual detection, contraction coupling threshold was found to be within 10 mV of resting potential (Hoyle, 1973). The larger j.p.s were about 20 mV, therefore exceeding excitation-contraction coupling threshold, and they elicited tension in such fibres.

In most respects, the relationships of tension to SETi stimulation was as has been described previously for the whole extensor muscle. However, fusion could be detected at lower frequencies than in the whole muscle. Preparations that did not show a twitch developed smooth tonic tension, at a frequency as low as 3 Hz (Fig. 10). Two previously unsuspected aspects of SETi function were encountered: one was a strong facilitation in the mechanical response to SETi when a low-frequency train was repeated after a period of a minute or two, the other was a marked effect of stimulus history on relaxation rate. The first was observed when SETi had been stimulated at 10 Hz for 60 s (Fig. 11 A), then, following relaxation, the excitation was repeated. The initial rate of contraction was doubled (Fig. 11B). The state of facilitation set up by a single burst persisted, with slow decline, for up to about 5 min. The phenomena is large enough to be of behavioural significance if it occurs in the intact animal; however, it may be an artifact of increased synaptic output in the saline.

The time-courses of the twitches shown by 135c and d in response to SETi dispelled the notion that they are intrinsically slow. Cochrane et al. (1972) did not show a figure of a slow muscle fibre twitch, but stated its duration in S. gregaria to be almost 4 s. In the present work it was found to be always less than 1 s.

The relaxation rate of 135c and d, following a train of stimuli, was found to be slower than after a brief train at higher frequency, developing about the same amount of, or greater, tension (Figs. 12, 13). This difference was greatest in S. gregaria where the relaxation time was about 30 times as long after 10 s or more of stimulation at 10 Hz than after 1 s at 20 Hz even though more tension was developed by the former stimulation. The phenomenon was equally evident whether the muscle was bathed in saline or in haemolymph. This is a significant finding that is probably of functional significance. A possible explanation will be offered in the Discussion. The phenomenon is less obvious in the response of the whole extensor, suggesting that it occurs only in slow fibres, not in those of intermediate type.

Single muscle units (as defined by Hoyle, 1957), i.e. natural clusters of fibres wrapped in a common tracheolated membrane and receiving a single nerve twitch, were obtained from different parts of the femur and tested separately. The nerve branches were cut to all units except the one being tested and the force transducer was placed so as to obtain optimal tension recording from that unit. Successive units were tested, from proximal to distal, on the inside, outside, ventral and dorsal surfaces.

There are 17 distinguishable muscle units visible when viewed from the ventral surface after removing the flexor, not including 135c and d. Each of these obscures a similar unit lying dorsal to it. The outlines of the attachments of the bundles can easily be seen on the external cuticle. The units increase in girth as well as in length proximally, in association with increased fibre thickness. The joint is asymmetrical, making the most proximal unit on the outside, i.e. the fan, substantially larger than its inside partner. It comprises five, or possibly more, units compared with the usual number, two, of its partner.

With so many component muscle units (about 74), three basically different types of muscle fibres, and four different types of axons to innervate them, in various combinations, heterogeneity of units was clearly a strong possibility. Various ablations were performed to test the contributions made by different regions to total tension development during SETi stimulation. The functional distribution of SETi was tested by cutting all the nerve branches except one, and recording the tension developed by the region remaining innervated. This is a more reliable procedure than random searching for SETi e.j.p.s with a microelectrode. Also, regions, bundles, and even partial bundles were prepared with the nerve still intact, but the rest of the muscle cut completely away. These were all done in many different combinations. First, the whole muscle tension to SETi was recorded. Then the nerve branches were cut, leaving only a region or bundle still innervated. In this way the relative contributions to total SETi tension of a component unit could readily be made.

The estimated percentages of the total number of extensor tibiae muscle fibres on which SETi endings are present may be obtained from Table 1. In S. gregaria the total percentage of muscle fibres innervated by SETi is 24% ; the earlier estimate, for Locusta, was 30% (Hoyle, 1955b).

When 135c, d only were removed from the whole muscle, final maximum steady tension was reduced by about 10% (Fig. 14A). At a SETi frequency above 50 Hz they are fully activated, so the 16−24 slow-type muscle fibres that comprise them contribute to total tension an amount proportional to their fraction of the slow-fibre population. However, the amount of tension developed in the first 1−2 s was at least 30% less after loss of these bundles. Therefore, they contribute a relatively large proportion of the tension developed by a brief burst.

When nerve branches to all the proximal (a) bundles were cut in addition to those to 135c, d (Fig. 14C), the final steady tension was reduced by an additional amount in the range 15−25 %, depending on the preparation, but the rise-time was little affected. The major portion of the muscle is therefore responsible for generating 65−75 % of the total SETi tension.

The most efficient mechanical contribution must be that made by the most proximal fibres of the fan, which are directly attached to the end of the apodeme. More distal bundles attached to the side of the apodeme must have matched tension in their counterpart to operate optimally. The distal-most bundles, 135c, d are efficient in this regard, since their sizes, innervation and responses are perfectly matched.

It was of interest to compare mechanical responses of fast-type fibres innervated only by FETi with ones of intermediate type innervated also by SETi. A difference could account for differences noted between regions since the relative proportions of such fibres could differ in different regions. Unfortunately no region was found in which the muscle fibres were all of homogeneous intermediate type. However, distal units f are largely of this type. Responses of one of these are shown in Fig. 15 A. The twitches showed a large extent of facilitation with repetition. Maximally, this resulted in the second, facilitated twitch being three times the initial one. This was largely an intramuscle fibre phenomenon rather than a synaptic one since the second action potential was identical to the first. By contrast, fast-type fibres (from outside b), although showing mechanical summation, show no facilitation (Fig. 15 B). The tetanus : twitch ratio in the former fibres was about 12:1 compared with only 4:1 in the latter. This ratio for the whole ETi is about 6:1.

The twitch in response to SETi of a small distal muscle unit comprising a high proportion of intermediate-type fibres innervated by both FETi and SETi, was about twice as long as that to FETi (Fig. 16). This difference was entirely due to a slower relaxation time following a SETi impulse. The difference must be sought either in the different time-courses of the depolarizations or in slow mechanical properties of some fibres innervated by SETi but not by FETi or both combined. The FETi e.j.p./action potential is of greater magnitude than the SETi equivalent, but it is of briefer timecourse. The rapid repolarization phase of the FETi spike obliterates the slowly decaying tail of the junctional potential and may lead into an undershoot. By contrast, the smaller SETi graded responses fail to affect the tail and simply sum with it. The average duration of depolarization past the visually determined excitation/contraction coupling threshold for FETi electrical responses was only 3·7 ms, compared with 7·3 ms for SETi responses in the same fibres. Whilst no specifically slow muscle fibres were found in the bundle in question, it is quite possible that they exist, but the longer depolarization is able to account for the slower SETi twitch.

Major differences emerged when a tetanic stimulation was applied to the dually innervated bundle, principally in regard to fatigue (Fig. 17). Tension continued to build up with continued SETi stimulation for about 20 s at a frequency of 30 Hz, and for longer at lower frequencies. At higher frequencies tension reached a peak within the first 5 s, declined slightly and then rose again. The plateau was maintained for as long as 5·6 h at 30 Hz and, although there was some decline thereafter, examples were seen in which tension was still greater than 50% maximum after 24 h of SETi stimulation at 30 Hz.

The responses to FETi are in marked contrast. There is a strong summation at frequencies above 10 Hz, so it is unlikely that the twitch represents a full development of ‘active state’ as it does in frog fast fibres. Maximum tension increased with increasing frequency up to about 70 Hz, when it was 20−30 times the twitch tension. The tetanus : twitch ratio is less in the whole extensor. From this it may be deduced that in fast fibre bundles the twitch represents a greater degree of activation than it does in the intermediate-type fibres that constitute the FS bundles.

Nevertheless, declining tension is evident during a FETi tetanus at any frequency after only 1−2 s. At 30 Hz, FETi-developed tension declines almost to zero in about 2 min. At 50 Hz, tension declines almost to zero in but 10 s.

Proximal to the Y branch in the nerve there are large mismatches between inside and outside parts of the muscle. There are four distinct types of muscle fibre groups exemplified by the slow 135c, d, the mixed fan bundles (outside a), the fast outer bundles (outside b) and the intermediate (f) bundles. Distal to the Y structural differences between inside and outside are not obvious. The only anatomical difference observed was in the presence of DUMETi, which continues for a greater distance, and is also much thicker, in the inside branch leading one to ask: is there any physiological consequence of there being few DUMETi terminals on outside bundles? Also, since the proximal third of the muscle is innervated from lateral branches of a central nerve trunk, what is the significance of the Y branch, which does not occur in proor meso-thoracic extensor tibiae, or those of non-saltatory orthopterans ?

To compare muscular responses of outside with inside units below the Y, SETi and FETi were independently excited near the ganglion whilst branches of the nerve supplying different regions of the muscle were cut. Branches above the Y were all severed, and the outside branch cut immediately after the Y. The response to FETi of the half whose nerve had been severed was tested after drawing the cut end into a suction electrode. No difference was seen in the size or shape of the twitch of the outside half, compared with the inside half. But consistent differences showed up during tetanic stimulation, or during long series of twitches in a low-frequency tetanus (Fig. 18). The inside bundle showed a continuous gradual facilitation and summation over the first 10 s at 10 Hz. The outside bundle did not summate after the second stimulus, and twitch height was already declining after 10 s (Fig. 18A). At a higher frequency (Fig. 18 B), the outside fatigued appreciably sooner than the inside. In view of the herringbone pattern of the extensor, mismatch of inside to outside units must mean that there is asymmetrical tension development during both the rising and plateau phases of a tetanus on the two sides of the apodeme. This would be sufficient to interfere with optimal functioning if a FETi burst continued beyond 3 s. However, such long bursts may not occur naturally, or asymmetry may be irrelevant for jumping (Heitler & Burrows, 1977). The more rapid fatigue of outside units appears to be due to a more rapid decline in FETi junctional potential height. The only physical feature which could be associated with this difference was the reduced innervation of the outer units by DUMETi.

Corresponding mechanical responses of inside and outside to SETi are shown in Fig. 19. In Fig. 19A(i), the whole muscle response is shown. The outside nerve branches were then cut, and the stimulation repeated, leading to A(ii). There appears to have been a slightly faster summation of tension in the outside fibres in this preparation. In another preparation (Fig. 19 B (i)) the inside branches were cut, leading to B(ii), in which the tension increment is neatly halved, showing the two sides to have been exactly similar in their responsiveness to SETi. Fig. 19C(i) compares the response of both inside and outside regions below the Y with that after cutting all branches above the Y. The result obtained showed both that there is no special character to the proximal region and that SETi tension development is distributed throughout the muscle, as determined earlier for Locusta (Hoyle, 1955b).

Not all preparations showed such a marked twitch in response to SETi as did those in the experiments shown in Fig. 19. Results from a preparation in which there was no detectable response to a single shock are shown in Fig. 20. The outside produced more tension than the inside at lower frequencies. But the inside had both a faster rise-time and developed 15% greater total tension, at higher frequencies. For the whole series of preparations, the halves were remarkably well-matched, but the results clearly showed that there are differences, probably in both the proportions of muscle fibre types, and in synaptic properties, between them.

A major initial purpose of this investigation was to determine the locations of muscle fibres responsible for the spontaneous intrinsic rhythm (IR) of slow contraction (Hoyle & O’Shea, 1974; Hoyle, 1978a). An early IR contraction can be triggered by a single SETi or FETi impulse, or it can be terminated early, once a critical period has elapsed, by a single SETi, FETi or even a CI impulse (Hoyle, 1978b). The latter suggests that IR occurs in muscle fibres innervated by all three axons, yet no FETi electrical response occurs in muscle fibres that appear to be pacemakers for IR.

It was found that the amplitude of IR was not diminished when all of the muscle units were cut away from the apodeme, starting distally, except the fan. If any of the muscle units comprising the fan were removed there was reduction in amplitude. Depolarizing waves of similar time-course to IR were obtained whilst recording intracellularly from units of the fan. These fibres appear to be the pacemakers for IR. Many muscle fibres nearby show smaller depolarizations of similar time-course. The most probable source of these is electrotonic spread from the pacemakers by bridges between muscle fibres, but no direct evidence for such possible bridges has yet been obtained. No comparable waves were seen in fibres elsewhere in the muscle. Fewer than one third of all fibres sampled in the fan showed the waves. Because of the thickness of the fan it proved to be very difficult to sample it satisfactorily.

Tentatively, it is concluded that IR is a property of a small number of muscle fibres entirely situated in the fan region. These are of S type. They are not innervated by FETi and give only very small depolarizations in response to SETi.

The present results provide a broad account of the innervation, muscle fibre composition, and mechanical responses to neural stimulation, of different regions of the jumping muscle of a saltatory orthopterous insect. The results extend the observations by Hoyle (1955 a, b) and Cochrane et al. (1972). They show that the extensor tibiae is highly complex, with marked morphological and physiological heterogeneities.

The locations of specific types of muscle fibre are prescribed and therefore under genetic control. In turn, the major types of synaptic variation also are found repeatedly in the same places in different specimens of all three species. Furthermore, these locations were similar in the different species and genera and are therefore of relatively ancient common origin. May we presume the complexities are responsible for a betterfunctioning leg muscle ? Is the lack of apparent rationale in the design in part the result of evolutionary accidents that combined several synergistic muscles ? The same questions can be asked of the several crustacean muscles that have now been examined, such as the claw closer of Cancer magister (Atwood et al. 1965); the eyestalk levator of Podophthalmus vigil (Hoyle, 1968); the power-stroke swimming muscle of Portunus sanguinolentus (Hoyle, 1973) and the strike muscles of the mantis shrimp Hemisquilla (McNeill, Burrows & Hoyle, 1972). It appears to be also true of the Periplaneta metathoracic extensor tibiae (Atwood, Smyth & Johnson, 1969) and indeed, some aspects of the specialization of the homologous locust and cockroach muscles suggests that they have significant common peculiarities. That would make it likely that these specializations, such as dense proximal and distal SETi innervation, are of very ancient ancestral origin. The complexity raises questions of its achievement during development. Atwood (1973) is developing the theme for crustacean muscles that a combination of developmental factors is what determines the details of innervation, and therefore indirectly, at least in part, muscle fibre and synaptic properties. However, some of the peculiarities of the locust extensor tibiae represent abrupt discontinuities and clearly defy any such simple developmental explanation.

As to what the advantages are that have been bestowed by particular specializations, I regret that I cannot come up with any strong suggestion. The specifics are so bizarre that they must be viewed as the end product of a series of diverse, discontinuous, evolutionary steps.

A major question for the present work concerned the fractions of total tension developed by the slow axon in different parts of the muscle. Cochrane et al. (1972) had attributed all SETi tension to proximal bundles for 5. gregaria. This has turned out to be far from correct. There are two concentrations of purely SETi-innervated muscle fibres, 135c and d, and a portion of the fan. But 70% of SETi-developed tension is produced by fibres in the main body of the muscle, where an estimated 15 % of fibres must receive SETi innervation.

In no preparations were slow-type muscle fibre twitches found to have rise or decay times approximating the values of 790 and 2950 ms, respectively, given for the S. gregaria proximal slow bundle by Cochrane et al. (1972). On the contrary, these fibres were found to have quite brief twitch times. Some of these muscle fibres participate in IR (Hoyle, 1978) and stimulating them sometimes elicits a slow IR contraction lasting one to a few seconds, that might be mistaken for a slow contraction. If the twitch times indicate the proportion of slow and intermediate, compared with fast, muscle fibres in fast axon-innervated bundles we would conclude that they are about equally mixed throughout the muscle except in outside region b, where all are fast, and region f, where there are relatively more intermediates than elsewhere. The results of the survey of the distribution of DUMETi afforded no clues to its possible functions. If its purpose were solely to inhibit IR one would expect its terminals to be confined to fibres participating in, or close to, those that generate IR. DUMETi terminals are indeed relatively concentrated on muscle fibres of the outside b region that abut the proximal fan-shaped region which includes the sites of IR. These are fast-type fibres that do not themselves participate in the IR so the liberated material must diffuse to the targets through circulating haemolymph. However, DUMETi continues part way in the outside branch and in the inside branch to the end of M135a, b. It is, of course, possible that inhibition of IR is not the only function of DUMETi; it could have other physiological roles or a trophic one in relation to maintaining fibre type. The latter is compatible with its distribution because the proximity of DUMETi terminals is closely associated with fast characteristics.

The conclusion of Cochrane et al. (1972) that there are structurally distinct types of fibre in ETi is confirmed. Just precisely what determines differences in muscle fibre structure within a given muscle is not yet established. In vertebrates, innervation has been shown greatly to affect the ultrastructure, physiological and biochemical characteristics of the muscle fibres (Close, 1972). With four physiologically distinct axons innervating the extensor tibiae, several combinations of innervation differences are possible. However, strongly differentiated slow and fast muscle fibres are known in an arthropod muscle that receives but a single axon (Dorai-Ray, 1964), so muscle fibre differentiation can occur independently of innervation differences. There are strong indications that there are reciprocal, trophic interactions between the different axons and the muscle fibres they innervate.

Innervation by the slow axon is associated with reduced amounts of longitudinal sarcoplasmic reticulum (SR) and fewer invaginated tubules and tubule/SR contacts. Innervation by the fast axon is associated with larger amounts of both. If there are trophic controls, muscle fibres innervated by both axons are likely to be of intermediate type.

A large part of the fan was indicated by Cochrane et al. (1972) to comprise fibres innervated by both SETi and FETi. Indeed, this was the only part of the extensor their diagram shows to be dually innervated. In the present work many muscle fibre types were found in the fan, including, on the basis of innervation: S, F, SF, SI and SIF. On the basis of structure, slow, fast and intermediates were all present in all species.

Several special properties of slow muscle fibres are of potential behavioural significance, particularly the variable relaxation rate dependent upon the duration of stimulation. A principal feature of tension rise in response to SETi stimulation is that at low frequencies, up to about 30 Hz, tension rises slowly and continues to build up over a period of tens of seconds, going into a plateau that is maintained indefinitely without fatigue. At higher frequencies initial tension rise is much faster. This is followed by a slight fall in tension followed by a rise to a steady plateau at mid-range frequencies (30−60 Hz). At high frequencies (above 60 Hz) there is a steady drop in tension after the early surge. The explanation of these events appears to be twofold. In part they are due to the different responses of two principal populations of SETi-innervated muscle fibres: slow fibres innervated by SETi but not FETi, and intermediate fibres innervated by both axons. The former, which are concentrated in 135c, d and part of the fan, have low Ec thresholds and large e.j.p.s, some of which do not facilitate. The latter, which are scattered throughout the muscle, have high Ec thresholds and mixed sizes of e.j.p.s, all of which show facilitation, especially the smaller ones which are recruited only after facilitation and summation of e.j.p.s during repetition has led to their exceeding Ec, i.e. whose twitch : tetanus ratio is infinite. By contrast, the former exceed Ec even as single responses, giving twitches. Nevertheless, the twitches are but a tiny fraction of tetanic tension, the twitch: tetanus ratio being about 1:100 (unfacilitated twitch).

The principal functional role of the structurally specialized slow muscle fibres innervated by SETi, but not FETi, appears to be to permit two opposite functions to be performed by a single set of fibres: rapid tension development on one hand, and maintained tonic contraction that is relatively insensitive to small frequency fluctuations, on the other. The first of these functions is achieved by the low Ec threshold and large size of SETi e.j.p.s on these fibres, which result in activation of the peripheral myofilaments by a single impulse. The second is achieved by having a slowly relaxing core that is activated only during prolonged excitation, but that is activated at low frequencies.

Where does the extra tension come from during a tetanus ? This can be due only in small part to recruitment of fibres and is therefore mainly caused by more intense contraction within individual muscle fibres. This may be in part due to progressively stronger activation of contractile machinery at any particular locale, but a major factor could be recruitment of myofilaments in the core of the fibre. It is proposed that the twitch of a slow fibre involves only the periphery, whilst in a tetanus the core is progressively recruited. It is suggested that the progressive decrease in relaxation rate that accompanies increased duration of low-frequency tetanic stimulation reflects a slower intrinsic relaxation rate of the late-recruited contractile material of the core. The most likely explanation is that since there is little SR and few dyads in the core, myofilaments there are farther away from calcium uptake sites. Activator calcium can be expected to reach core myofilaments only during a relatively long period of stimulation and it takes much longer before this calcium is returned to storage areas than is the case at the periphery.

In intermediate fibres the recruitment of myofilaments also presumably starts at the periphery and gradually spreads to the core, but because there is a more extensive SR, their relaxation rate is not greatly affected by duration of stimulation. There appear to be several constraints on the possible diversity of fibre types. Of the twelve possible innervation patterns, four were never seen and two very rarely (Table 2). There is clearly a low probability of I innervating fibres that are also innervated by F, and of D innervating fibres also innervated by S. If a fibre is innervated by S as well as F it is likely to be also innervated by I. This combination is relatively common compared with finding FI, which was seen only twice. There are three possible explanations: I is attracted by S, S suppresses a repulsion of I by F, or both occur. Terminals of D were found on outside b fibres which are otherwise innervated only by F. D is vestigial more distally, in the outside branch after the fork, but prominent in the inside branch. D terminals were seen on only one fibre after the fork, an inside d fibre.

We should ask, based upon the distribution of DUMETi, if DUMETi terminals could have a trophic influence on either innervation by other axons or on the muscle fibre types found in a particular region. There is an abrupt margin between outside regions a and b in both innervation and fibre type. In a there are many fibres of slow or intermediate type, as well as some fast ones. SETi, FETi and CI are all present, but not DUMETi. In b there are only fast-type fibres and both SETi and FETi are missing. This is where DUMETi terminals occur. SETi and CI do not begin to supply more than an occasional muscle fibre, passing distally, until the most distal bundles. These do not receive branches from DUMETi. Thus it is possible that when DUMETi is present it suppresses innervation by SETi and CI. Also perhaps because of the absence of the latter, the qualities associated with fast contraction and relaxation are developed either directly or indirectly. Since the peripheral distributions of other DUM neurones is not known we cannot yet decide to what extent this might be a general principle. Comparable studies to the ones reported are under way in our laboratory on the non-specialized pro- and mesothoracic extensor tibiae. Kate Skinner (personal communication) has found that there are no specialized slow and fast muscle fibres in these serial homologues. All are of relatively fast type ultrastructurally regardless of innervation, which includes slow only, dual and fast only. In the homologous extensor muscles of the cockroach there is a complex pattern of innervation that is basically similar to that described in the present paper, with dense foci of slow axon innervation proximally and distally (Atwood et al. 1969), although no ultrastructural differences in the muscle fibres were noted. There are functionally specialized slow and fast portions in trochanteral depressor muscles of cockroach (Becht, 1959), which show colour and biochemical differentiation (Smit, Becht & Beenakkers, 1967), as well as subtle differences in innervation (Pearson & Iles, 1971). The latter include parts innervated by a fast, but not a slow, axon, parts innervated by both a fast and a slow axon, and other parts that receive this same slow axon plus three inhibitory axons.

A detailed study of the morphologies of SETi and FETi of all three extensor tibiae has recently been made (Wilson & Hoyle, 1978). This study revealed that in the metathorax, the morphological homologues of the pro- and meso-SETi and FETi have switched functional roles, the original slow becoming fast functionally, and vice versa. For this to have occurred, the starting point for specialization in innervation must be genetic determination of muscle fibre properties. There must also be genetic determination of the major branching pattern of the whole nerve trunk. Innervation details and synaptic transmission properties must follow genetic predetermination of muscle fibre type rather than vice versa, but there is mutual interaction during development.

This research was supported by grant No. BNS 75-00463 from the National Science Foundation. I am grateful to Melissa Williams, Betty Moberly, Margaret Titmus, Connie Huston and W. Colquhoun for technical assistance.