An Isolated Insect Ganglion-Nerve-Muscle Preparation^*

Previous studies on neuromuscular transmission in insects have all been made using preparations in which the muscle is left in situ and a minimum amount of damage done to the tracheal system (Hoyle, 1965 a). Success has also been achieved in the study of an isolated muscle without its nerve supply—the dorsal longitudinal flight muscle of Locusta migratoria (Weis-Fogh, 1956). Attempts have been made to produce fully isolated preparations (e.g. Hoyle, 1953) but none has so far been fully successful. There are several advantages, in principle, to the use of a completely isolated preparation. The muscle fibres may be much more thoroughly bathed in saline than can be achieved in situ, even with perfusion, and changes in ionic composition, addition of drugs, etc., can be accomplished more efficiently.

A major advantage is also offered by the fact that isolation permits examination by transmitted light, whereas in situ work usually demands the use of reflected light. In view of the increasing awareness of diversity in the nature of individual fibres within the same muscle in arthropods (Atwood & Dorai-Raj, 1964; Atwood, Hoyle & Smyth, 1965) the need to visualize individual fibres clearly during intracellular penetration has increased and will be demanded of the most critical work. Visualization is desirable for penetration by two electrodes at different separations for determination of electrical characteristics of the membrane. A further, related point, of great relevance where other than the thinnest muscles are concerned, is that an isolated muscle can be turned around and examined from all sides. If a muscle has a heterogeneous composition this becomes of vital importance, because unusual fibres may be inaccessible to selective puncture by micro-electrodes when operations are restricted, as in in situ studies.

An insect muscle which is particularly favourable for isolation was recognized as a result of studies (Hoyle, 1965b) aimed at understanding the neurological basis underlying learning to keep a leg in the raised position, which can be accomplished by the thoracic ganglia of certain insects Horridge (1963). In these studies it was shown that the motor nerve output to the anterior coxal adductor (a.c.a.) muscles of grasshoppers and locusts carries a constant tonic discharge at about 12/sec. This rate may be raised three-or fourfold by giving electric shocks to the tibia of the leg on which the adductor operates, every time the spontaneous background discharge rate falls by more than 15 % over a 10 sec. interval.

The studies described above were done with minimum dissection, but interference from many sensory inputs was evidently occurring and could have been influencing the results adversely. Progressive sectioning of nerve trunks was made in the present work to see how much isolation could be tolerated without loss of spontaneous activity in the motor nerve. It was found that all the nerves leaving the metathoracic ganglion could be severed but the discharge in nerve branch 3 c (cf. Hoyle 1955) which contains the motor axon supplying the a.c.a. still continued in many instances. This led to attempts to make a fully isolated ganglion-nerve-muscle preparation retaining the capability of spontaneous activity, and of reflex and integrative functioning. The results were successful and are described in this paper.

The anterior coxal adductor is a flattened, fan-shaped sheet of muscle fibres attached solely to the sternal apophysis on its inner margin and via a short, half-disk apodeme to the coxal rim at its outer end. The muscle can be fully isolated quite easily and can be mounted so that its force may be measured accurately (see details below). Nerve branch 3 c together with the whole mesothoracic ganglion may be freed from the rest of the animal and removed along with the muscle. It was found that the preparation thus isolated would remain alive for more than 8 hr. in standard locust saline (Hoyle, 1953) whilst giving good muscle-fibre membrane potentials. Stimulation of the nerve branch always leads to good contractions in the muscle.

The contraction is effected by a single excitatory axon. In addition to this axon a second efferent axon, also located in nerve branch 3 c, was found which when active gives rise to polarizing potentials in some of the muscle fibres and, occasionally, partial inhibition of the contraction caused by the excitor. This type of axon has been regarded by Hoyle (1965a) as a ‘conditioning’ axon preparing the muscle for strong contractions, but by Usherwood & Grundfest (1964, 1965) as a peripheral inhibitor comparable to those found in crustaceans (Wiersma, 1960).

Three matters were considered to be worth studying with the preparation, and the report which follows gives a preliminary account of them. First, in some preparations the excitatory axon still fires spontaneously at rates similar to those found in the intact animal. The rate may reflect a previous ‘learning’ experience aimed at leg raising, by being unusually high. It can be modified by stimulation applied to various of the cut stumps. Thus the preparation may serve to permit further elucidation of the ‘learning ‘process. Secondly, the preparation may be directly stimulated or ‘driven ‘by electric stimulation applied to the cut ends of nerves attached to the ganglion; i.e. the motor neurons fire only during the stimulation and cease firing when it is stopped. This capability permits a study of integrative processes in the ganglion. Thirdly, the second axon gives peripheral effects similar to those of the third axon supplying the jumping muscles, whose functions are currently the subject of controversy (Hoyle, 1965 a; Usherwood & Grundfest, 1964, 1965). In the isolated preparation its properties may be studied under conditions favourable for experimentation.

The studies have been made principally on males of the African locust Schistocerca gregaria Forskål, together with some Ramalea microptera, Melanoplus differentialis and S. vaga of both sexes. All the above are orthopterans, but it is considered probable that large insects from other orders, especially Hemiptera and Coleóptera, can yield Comparable preparations. The physiological properties of the muscles from the grasshopper and the locust species were remarkably similar.

The dissections were made on the metathoracic legs. The insect is first laid on its back in soft wax with all the legs spread out whilst the metathoracic legs are particularly widely spread, rotated in the clockwise direction, and firmly fixed in wax. This extends the anterior coxal adductors (a.c.a.) and brings their attachments to the coxae uppermost. Next, the ventral thoracic cuticle is cut away in the central region so as to expose the ganglion ; all nerves attached to the ganglion except the third branch of the third nerve are severed, leaving stumps about 1 mm, long or more. The ganglion is subsequently washed with locust saline (Hoyle, 1953), which is frequently changed during the dissection.

Attention is now focused on the sternal apophyseal pits. The sternal apophyses are severed cleanly just below the invaginations and the remains of the basisternal and first abdominal sclerite are removed. The anterior adductor, which emerges from under the posterior rotator, which is also attached to the apophysis, is located (Fig. 1). The attachments of the rotator are next cut away close to their bases until the whole of the anterior adductor is exposed. This uncovers the posterior coxal adductor, which is also cut away. The whole of the anterior adductor can now be seen, together with the nerve, which enters over the anterior ventral margin and immediately branches profusely as it enters the fan-shaped bundle.

The dorsal attachments of the sternal apophysis are now cut through, taking great care not to damage the muscle. After this, the apodeme is freed by cutting the distal attachment. A small piece of coxal rim may be taken along with the apodeme, providing a greater area for attachment later and serving also as a handling point; or the severed apophysis may be used. It remains only to lift the muscle gently up and cut tracheal tubes and connective tissue away. Likewise, it is necessary to see that the ganglion is free from attachments. At this point the preparation may be lifted out into fresh saline.

It is held by clamping or micro-pinning the triangular apophysis in such a manner that the muscle overhangs a small block of hard wax (Fig. 2). The apodeme may be conveniently seized by the tips of a forceps transducer (Hoyle & Smyth, 1963) to permit registration of the force of contractions. The ganglion is gently drawn away from the muscle and fixed to the wax by a small Perspex bridge. The muscle is illuminated by transmitted light and an intracellular glass capillary micro-electro de is introduced into a single muscle fibre. A pair of small, tapered, chlorided-silver hook electrodes is micro-manipulated into place under the nerve trunk 3 c. Similar electrodes are manipulated under cut nerve stumps attached to the ganglion as needed.

Recordings were made with a six-channel pen-oscillograph (Offner Dynograph) in conjunction with a Tektronix dual-beam cathode-ray oscilloscope.

When first penetrated the muscle fibres have resting potentials from 46 to 73 mV. in standard locust saline (Hoyle, 1953). Fibres showing a low resting potential at the start of the experiment often show a gradual increase over the first half-hour. The cause of this has not been determined, but since the muscle is under continuous neural excitation in these experiments at least up to the moment of severing the nerves, it may represent a slow recovery process. A high frequency of excitatory nerve stimulation leads to a slow progressive depolarization (d.c. shift) in addition to the brief junctional potentials, and this has a slow time-constant of recovery. Not all the fibres show this property, and the differences between them were sufficient to suggest the possibility that two different kinds of muscle fibre are present. An electron-microscopic investigation of the muscle fibres has been started; the fibres all have similar, long (6μ) sarcomere lengths but their sarcoplasmic reticulae are of two distinct kinds, supporting this possibility.

The electric responsiveness of the membrane is of the moderately graded kind (c.f. Cerf, Grundfest, Hoyle & McCann, 1959). The largest of the neurally evoked membrane responses almost reach the zero membrane potential level, but overshooting action potentials have not been found.

Facilitation of the junctional potentials has not been found in the Schistocerca preparations, but was quite marked in some Romalea preparations. In the latter, however, it could be eliminated simply by raising the calcium ion level in the saline to 5 mM/1. The junctional potential then had about the same height as previously after full facilitation. The single excitatory axon thus fails to possess, in Schistocerca, one of the most significant criteria of ‘slow ‘axons. However, it cannot readily be assigned to the ‘fast’ type because many of the junctional potentials are small (less than 20 mV.). It may be termed ‘intermediate’, or grouped with axons of the ‘slow’ type.

The single excitatory axon has a lower threshold than the ‘inhibitory’ one and hence its responses may be studied alone by stimulating nerve 3 c with just threshold shocks. The responses are of various magnitudes in different fibres, ranging from 5 to 60 mV. The smaller ones are pure junctional potentials (Fig. 3 a). The largest ones are compound, having a large junctional component together with a large graded response (Fig. 3e) (cf. Cerf et al. 1959).

The durations of the junctional potentials vary over a ten-fold range between different fibres of the same muscle, ranging from about 5 msec, to more than 150 msec. These differences are partly the result of differences in the electrical time-constants of the membranes of the fibres, for the short duration responses have faster rise times (Fig. 3b). They also appear to reflect markedly different junctional events, for there are large differences in the extent to which graded responses initiated in the membrane by the depolarizing action associated with the junctional potentials in turn cause faster rates of decay of their falling phases. In some fibres the decay is greatly increased by even a small graded response. In others it is partially accelerated at first but the late phase is slow (Fig. 3,h). In extreme cases, the slow decay rate of the junctional potential remains totally unaffected by the occurrence in the membrane of even a large graded response (Fig. 3m). These differences are not easy to account for in muscle fibres of short length (2·4 mm.) having multiterminal innervation. Where the decay phase of the junctional potential is obliterated by a graded response it means that the transmitter action is brief and the falling phase a purely passive, electrotonic effect. The failure to reduce it in other fibres could be due to their having a much longer transmitter action, perhaps due to a slower enzymic destruction. The effect can otherwise be explained only on the basis of a very large area of chemically, but not electrically excitable post-synaptic membrane. The view that the transmitter substance is destroyed at different rates is favoured. It is significant that the fibres whose junctional potential decay is not reduced are also the ones with the longer time constant; they may represent fibres specialized for slow following of tension upon depolarization and hence contribute only to slow and prolonged contractions of the whole muscle.

The mechanical response to a single excitatory impulse is a small twitch. This rises to a peak force of about 0 · 4 g./cm.² in 40 msec, and decays to zero in 600 − 1400 msec, at 23°C. The slow decay, compared with the rapid rise suggests that the muscle, like crustacean muscles (Dorai-Raj, 1964; Atwood & Dorai-Raj, 1964; Atwood et al. 1965), is not homogeneous but comprises a mixture of slow, fast and possibly intermediate muscle fibres. Fusion of twitches begins at about 8 sec. and is complete at 30/sec. but this may well represent only the action of the fastest fibres. Subsequently tension still rises progressively with increasing stimulus frequency, up to a maximum at 120/sec., where it is at least 4 K g./cm.². Considering the junctional potential nature of the excitatory action this is a remarkably high value. Full tetanic force is not developed until after at least 4 sec. of stimulation at a high frequency. The late accumulation of tension may be developed by specialized slow muscle fibres. Even larger values may be obtained by raising the calcium ion concentration in the bathing medium or adding Ba++ ions, which render the membrane capable of producing spikes. Full relaxation after a tetanus takes about 4 or 5 sec. The muscle can maintain its full tetanic force for as long as an hour without fatigue.

The second axon could occasionally be stimulated selectively by adjustment of the position of the stimulating leads applied to N 3 c, in combination with the stimulus strength and duration. More commonly it was excited by preganglionic stimulation of a nerve stump. It gives rise to small junctional potentials in most of the fibres; these are principally polarizing (Fig. 5), but some are depolarizing (Fig. 6). Their maximum magnitudes ranged from 16 mV. polarizing to 14 mV. depolarizing, with many fibres giving ones so small as to be difficult to detect; all possible intermediates were encountered. The smallest I-C junctional potentials were seldom increased greatly by polarizing the membrane with a second electrode and so cannot be simply due to the membrane potential being close to the equilibrium potential for the ‘inhibitory’ synaptic conductance change (c.f. Grundfest, 1962).

Electron-microscopic examination of the neuromuscular junctions of these fibres has shown that in addition to large neuromuscular synapses, presumed excitatory, there are terminals which are extremely thin and therefore contain but a few vesicles. These could represent inhibitory terminals and explain the small size of inhibitory potentials in some fibres.

The inhibitory potentials are always longer in duration than excitatory ones, ranging from 60 to 260 msec. This is the case whether they are polarizing or depolarizing and indicates that the duration of I-C synaptic conductance change is 3-4 times as long as the excitatory conductance change.

Stimulation of the I-C axon alone causes a slight contraction of not more than 200 mg. force in some preparations (Fig. 5). In others it causes a slight relaxation of resting tension, even when the excitatory axon has been completely silent for some time and therefore cannot be causing a background tonic contraction (Fig. 7). In most muscles there is no direct mechanical effect. Combinations of ‘inhibitory’ with excitatory effects lead to complex interactions (Hoyle, 1966). The contractions are presumably caused by depolarizing I-C axon junctional responses sufficiently large to exceed the threshold for contraction coupling. The relaxations must be the result of a preponderance of polarizing I-C axon junctional potentials, but another requirement is that in some muscle fibres the threshold for contraction is exceeded in the resting, slightly stretched muscle.

When first isolated many preparations show no activity in the motor axons supplying the a.c.a. In some of these excitatory axon activity starts up spontaneously after several minutes. This may represent the cessation of an inhibitory effect brought on by the violent stimulation associated with extirpation. The basic frequency of the pacemaker is 8−12 per sec. in the intact animal before extirpation, and similar frequencies are encountered in the isolated preparation (Fig. 8). Since the activity in the excitatory axon appears to continue relatively unchanged in frequency and pattern following extirpation, it must be located in the same ganglion as the motomeuron, possibly in the neuron itself. However, the frequency may be much higher than this, or it may fluctuate between wide limits, as is also the case in the intact animal. In some preparations spontaneous firing of the excitatory axon never occurs, but bursts of impulses may be elicited by preganglionic stimulation (see below). The effects could be caused by prolonged central inhibition, or they may be a result of deterioration.

The I-C axon is usually silent in these isolated preparations. However, it may fire impulses spontaneously in bursts, sometimes with a slow rhythm, at about one burst every 30 sec., with a frequency of impulses within the bursts of 5−8/sec. ; but such activity is sporadic. It also may fire for prolonged periods, especially when the excitatory axon is firing uninterruptedly at a relatively high (7−15/sec.) frequency. The functional role of the I-C axon will be considered in the second paper.

Sometimes the smooth, regular spontaneous discharge of the E axon breaks up into a series of slowly changing, patterned bursts. The change may be triggered by a shock applied to a cut nerve trunk, or it may occur without stimulation. The I-C axon may also enter the picture and fire in bursts. Some of the patterns are reminiscent of those seen in the intact animal (Hoyle, 1966). Examples of such discharges are given in Fig. 8.

The potential value of such activity lies in the fact that it may afford clues as to the storage capacity of the ganglion for complex information related to behaviour. The activity may represent the spontaneous expression of neural ‘centres ‘involved in some kinds of complex motor acts. However, so far this activity has been bewildering in its complexity, as the few examples given in Fig. 8 will show, and it has not so far been possible to assign any well-defined meaning to any of them.

In a silent preparation the excitatory axon may usually be induced to fire by stimulating some of the cut nerve stumps. It is most readily driven by exciting the stump of N 5 on the same side, but all of the major nerves have at some time been found effective. It is possible that owing to the small size of the ganglion, stimulus escape could be causing the excitation by affecting the motorneuron directly. However, several points argue against this being the case. For example, electric stimulation applied to the ventral nerve cord, no matter how strong, rarely evokes a response and usually suppresses a spontaneously occurring one (Fig. 8). Since this stimulation is as likely to lead to stimulus escape as any other, and since it causes a negative response compatible with natural behaviour, one is justified in considering the stimulation to be indirect. The response to stimulation of the whole nerve cord, mediated by the giant axons in the cord, is activation of leg extensors accompanied by central suppression of flexors and of the coxal adductors. This is comparable to the suppression found in the isolated preparation.

Suppression of the spontaneous excitatory discharge is also the commonest result of stimulating the anterior connectives, though another common result is a combination of initial suppression followed by late excitation in a continued burst of stimuli.

Interesting results have been obtained by combining stimuli to various nerves, but a full appraisal of such effects needs a more extensive study than has yet been made.

Stimulation of the connectives with the mesothoracic ganglion, especially that on the ipsilateral side, not only suppresses spontaneous excitatory activity, but also often causes the I-C axon to fire. The rate of firing may be only loosely related to the stimulus frequency ; sometimes a single shock will evoke a long burst. At other times a similar stimulus will give rise to a single impulse and the fibre can be driven i : i up to frequencies as high as 150/sec. Later the same fibre may be completely unresponsive to similar treatment. It is apparent that either the intervening synapses are very labile, or else spontaneous inhibitory and excitatory effects within the ganglion, and perhaps some effects of deterioration, combine to alter the properties. When the I-C axon is responding to stimulation of the connectives between meso- and metathoracic ganglia in a 1:1 manner, if the frequency of stimulation is raised to above about 50/sec the E axon also fires, giving a burst of impulses (Fig. 9). A few seconds of stimulation must be applied before the burst occurs, and the E axon frequency is not related to the stimulus frequency.

The existence of the spontaneous firing in the E axon makes it possible to carry out conditioning experiments aimed at altering the maintained mean frequency of the discharge. Provided such changes persist over long periods they constitute a simple form of learning. In view of the critical nature of this kind of experiment, and the need for careful statistical controls, no definitive statements may yet be made. However, some encouraging indications of the possibility of effecting long-term changes under conditions which may permit the physiological analysis of underlying events have been obtained.

Single electric shocks of increasing strength were applied to the cut stump of the crural nerve (N 5) on the same side until they just called forth a small reflex increase in the excitatory axon output to a.c.a. Shocks of this strength were then given every time the mean spontaneous activity fell over a 10 sec. period by 20% from the mean value over the immediately preceding 10 sec. period. This experiment is comparable to that which was often successful in raising the output frequency in the semi-intact preparation where stimuli were applied to the whole tibia (Hoyle, 1965 b).

Unfortunately, most experiments of this kind have so far given less satisfactory results than those done on the semi-intact preparations. There are larger variations in mean frequency and prolonged inhibition frequently occurs. Only two successful examples in which persistent changes occurred have been experienced so far (example in Fig. 10) and the possibility that in both cases the rise in frequency obtained could have been due to chance alone cannot be ruled out.

Little can usefully be said in discussion about the above results at the present time since most of them are of a preliminary nature. However, the preparation described is of potential value for research in several fields: neuromuscular transmission, muscular contraction, central integration and the physiology of learning. Its use in the latter field results from its spontaneous activity, which may be modifiable by simple conditioning. By some criteria this kind of physiological alteration can be regarded as constituting a form of learning. The physiological changes underlying the process must be located in the small piece of isolated—and therefore accessible—nervous system constituting the metathoracic ganglion.

Learning changes occurring in this preparation may arise as a result of an inherent tendency of the motorneuron pacemaker system to adjust its frequency of firing automatically following the receipt of certain sorts of input, provided these are timed with respect to its own output in a regular and characteristic way. The association of the input with the output may be provided for by inherent neuronal connexions forming a coupled feed-back relationship serving as a regulatory device in relation to posture. This regulatory system could, perhaps, account for the remarkably rapid adjustments made by insects to loss of, or damage to, limbs, which is evident so quickly in their locomotory patterns. An understanding of the physiological basis of this adjustment mechanism will be of intrinsic interest, whether or not it is eventually found to be comparable to other forms of learning.

Of some interest in connexion with general neuromuscular physiology is the strong suggestion afforded by the present work that an insect muscle can have a heterogeneous composition, comprising a mixture of muscle fibres of different types, having large ranges in membrane electrical, mechanical contractile, and excitation-contraction coupling properties. That this is true of some crustacean muscles has recently been conclusively shown by measuring various parameters of single muscle fibres of the same muscles (Atwood et al. 1965). A comparable analysis of insect muscle would be much more difficult to achieve on account of the smaller diameter of the fibres, but should be technically feasible.