Common Inhibitory Motoneurones in Insects

Motor nerve fibres to insect muscles are usually classified as either fast, slow or inhibitory, depending on the mechanical and electrical responses they elicit in the muscle (Hoyle, 1953a; Usherwood, 1962; Usherwood & Grundfest, 1964), although some motor axons cannot be satisfactorily placed within any of these categories (Hoyle, 1959, 1966b). Histologically, single insect muscles have been seen to be innervated by up to eight different axons (Dresden & Nijenhuis, 1958). Electrophysiological experiments, however, have not yet revealed that any simple muscle or branch of a compound muscle receives more than five axons. All insect muscles so far described receive at least one fast axon and they may or may not be innervated by slow excitatory axons and/or an inhibitory axon. Unlike crustacean muscle, no insect muscle has been found having dual inhibitory innervation. Single fibres of some crustacean muscles receive up to six different axons (Kennedy & Takeda, 1965), whereas only one insect muscle has been described in which single fibres are more than triply innervated; this being the basalar fibrillar muscle of the beetle, Oryctes rhinoceros, where a small number of fibres receive four excitatory axons (Ikeda & Boettiger, 1965). The usual innervation pattern of single insect muscle fibres is either single or dual excitatory.

The number of peripheral inhibitory nerve fibres so far found in insects is few compared to those for Crustacea, and there is some doubt in insects as to whether they have any real functional role (Hoyle, 1966b). Usherwood & Grundfest (1965) have described an inhibitory axon innervating the extensor tibiae muscle of the locust and grasshopper. Although stimulation of this axon always produced mechanical inhibition of the slow excitatory response, its normal function was not clear. Recently, however, Runion & Usherwood (1968), recording from the motor nerve to the extensor tibiae muscle in freely moving locusts, have shown that for regular walking the inhibitory axon fires just before the discharge of excitatory axons to the flexor tibiae muscle. This suggests that the function of this particular inhibitory axon may be to facilitate relaxation in the extensor tibiae muscle, thus allowing a more rapid and stronger flexion of the tibia.

A nerve fibre that can produce mechanical inhibition in the anterior coxal adductor muscle of the locust and grasshopper has been described by Hoyle (1966b). Hoyle found, however, that activity in this axon produced mechanical inhibition of the excitatory response only in about one third of his preparations. In another third it produced mechanical facilitation, while in the remainder it had no effect. Because of these diverse effects he preferred to call this axon an ‘inhibitory-conditioning’ (I-C) axon. Records of the spontaneous activity of both the 1–C axon and the single excitatory axon to the anterior coxal adductor showed that the I-C axon activity generally preceded the excitatory axon activity. In some cases contractions preceded by I-C activity were larger than those not preceded by I-C activity. The I-C axon does not then function as a simple peripheral inhibitor. Unfortunately the results to date do not permit a clear understanding of the normal function, if any, of this axon.

The only inhibitory fibre that has been found in cockroach muscle was briefly reported by Usherwood & Grundfest (1965) to innervate the extensor tibiae muscles of the two species, Blaberus giganteas and B. craniifer. Hyperpolarizing junctional potentials in the basalar fibrillar flight muscle of the beetle, Oryctes rhinoceros, were described by Ikeda & Boettiger (1965). These workers presented no results to suggest a possible function for the fibre producing these potentials.

The results presented in the first part of this paper show that the metathoracic posterior coxal levator muscle of the cockroach, Periplaneta americana, is sextuply innervated by two inhibitory and four excitatory axons. The electrical and mechanical responses in the muscle to activity in the three smaller excitatory axons allow these axons to be classified as slow, whereas the largest excitatory axon may be classified as fast. One inhibitory axon is shown to be a branch of a common inhibitory neurone. Common inhibitory neurones, although well known in Crustacea (Wiersma, 1961), have not been reported before in insects. The result presented in the latter part of this paper demonstrate not only the existence of common inhibitory neurones in the cockroach but also in both the locust species, Locusta migratoria and Schistocerca gregaria. Two axon branches of a common inhibitory neurone in the locust have been found to correspond to the inhibitory axon innervating the extensor tibiae muscle and to the I-C axon innervating the anterior coxal adductor muscle. These results thus extend the current picture of the possible types of innervation in insect muscle, and the similarities with the neuromuscular organization in Crustacea become more marked. A preliminary report of our findings in the cockroach has been published elsewhere (Bergman & Pearson, 1968).

The musculature of the cockroach, Periplaneta americana, has been described by Carbonell (1947) and the nervous system by Pipa & Cook (1959). The numbering system for the muscles and nerves specified by these workers has been used throughout this paper. The location of the different muscles and nerve trunks relevant to this investigation is shown in Fig. 1. The numbering system of the nerve trunks leaving the metathoracic ganglion is shown in Fig. 2.

The posterior coxal levator muscle (182) has four different origins on the dorsal wall of the coxa and its four branches (A–D) converge to an apodemal tendon attached to the distal part of the trochanter. Branches A and B are attached to the dorsal wall of the coxa and branches C and D to the dorsal part of the coxal rim. The posterior coxal levator muscle is innervated by axons contained in nerves 6Br4 and 5r3. Nerve 6Br4 contains twelve motor axons which have been numbered from 1 to 12 with increasing size (Pearson & Stein, 1969).

The nerves 6A, 6B and 4r2a are easily exposed by rotating the coxa so as to expose its dorsal side, and removing the soft cuticle connecting the dorsal rim to the abdomen (we have extended Pipa & Cook’s numbering system to specify the nerve to the sternal remotor muscle of the coxa (172) as 4T2a). Apart from containing nerve fibres that innervate muscle 182, nerve 6Br4 also contains all the nerve fibres innervating the main coxal levator muscle (181). Nerve 6A contains nerve fibres that innervate the tergal remotor muscles (174, 175, 176), wing muscles (158, 169) and tergopleural muscles (157, 159)-There are no sensory fibres in nerve 4r2a.

The second coxal branch of the main depressor muscle of the leg (177E) has its origin on the anterior wall of the coxa and anterior part of the coxal rim, and attaches to a broad apodemal tendon connected to the proximal angle of the trochanter. This muscle is innervated by axons contained in two fine branches coming off nerve 5r1a. One of these branches is easily exposed by removing muscle 179 (anterior coxal depressor) and part of muscle 180 (anterior coxal levator).

The positions and innervation of the extensor tibiae and the anterior coxal adductor muscles of the locust have been well described by Hoyle (1955, 1966a). The anterior coxal adductor is dually innervated by an excitatory axon and what Hoyle calls an ‘inhibitory conditioning’ axon, both of which run in nerve trunk 3c. The extensor tibiae muscle is innervated by three axons : one fast excitatory, one slow excitatory and one inhibitory. The slow excitatory and inhibitory axons are contained in nerve 3b and the fast excitatory axon in nerve 5. All three axons come together in the coxa to form the motor nerve to the extensor tibiae which runs the length of the femur. This nerve is readily exposed on removing the ventral cuticle of the femur together with the underlying flexor tibiae and retractor unguis muscles.

All cockroach experiments were carried out on the metathoracic segment of adult male Periplaneta americana. The muscles most intensively studied were two branches (C and D) of the posterior coxal levator muscle (182).

The animals were lightly anaesthetized with carbon dioxide, decapitated and pinned ventral side up on a cork board. To enable investigation of the metathoracic posterior coxal levator muscles the coxa had first to be rotated until its dorsal surface was exposed. The pro-and metathoracic legs were pinned so as not to interfere with the rotated metathoracic leg. The two branches C and D of muscle 182 were exposed for intracellular recording by removing a small section of cuticle above them from the dorsal wall of the coxa. A part of the depressor muscle 178 had also to be removed to fully expose branch 182D. The apodeme to which all four branches converge is easily visible on removing the cuticle from the dorsal coxal wall near the trochanter. Muscle branches A and B were removed from this apodeme and an RCA 5734 tension transducer was attached after cutting the apodeme close to the trochanter. This allowed tension measurements to be made from branches C and D.

The soft cuticle connecting the dorsal coxal rim to the abdomen was removed and the tergal coxal remotors (174, 175, 176) were detached from the dorsal coxal rim. This enabled bipolar 75 μ silver-wire electrodes to be manipulated under nerves 66r4 and 4r2a. Each electrode pair could be used for either stimulating or recording. The nerves were lifted clear of the haemolymph to give good signal-to-noise ratios and coated with petroleum jelly to prevent drying. Records could be achieved which remained stable over many hours using this technique. Records from nerve 6Br4 allowed the activity of various identifiable motor axons within this nerve to be correlated with the different junctional potentials recorded from single fibres of muscles 182C and 182D. Stimulation of nerve 4x2a gave selective activation of the inhibitory axon to muscle 182. (This was possible because the inhibitory axon to muscle 182 is in fact one branch of a common inhibitory neurone, another branch of which is contained in nerve 4x2a (see below). Stimulation of 4x2a antidxomically excites one axon branch of this common inhibitory neurone which in turn orthodromically excites all the other axon branches.) The experimental arrangement for investigating the innervation patterns of muscles 182C and 182D, and the electrical and mechanical effects of the inhibitory axon to these muscles is shown in Fig. 2.

In the experiments where intersegmental effects were not being investigated the metathoracic ganglion was isolated by cutting the connectives between the mesothoracic and first abdominal ganglia. This procedure led to a preparation in which there was less movement and one with increased activity of the small excitatory axons to the posterior coxal levator muscle. This increase in excitability was of use when investigating the innervation patterns of this muscle. Larger excitatory axons could be readily activated by light mechanical stimulation of the tarsus and the tibial spines.

Intracellular recordings from single muscle fibres of the promotor (168), remotors (172, 174) and depressor (177E) muscles were also made. These muscles were exposed by removing a small piece of cuticle above them.

Usherwood & Grundfest (1965) showed that chloride ions were involved in the generation of the inhibitory junctional potentials in the locust extensor tibiae muscle. For this reason micropipettes filled with 0·6 M-K₂SO₄ have always been used. The electrodes were mounted on a silver-silver chloride wire whose flexibility allowed stable records to be obtained during small movements. Muscles exposed for intracellular recording were kept moist by the application of small drops of saline. These were the only places where saline was added. The composition of the saline was that given by Becht, Hoyle & Usherwood (1960).

Adult male and female locusts of the species Locusta migratoria and Schistocerca gregaria were used. The animals were lightly anaesthetized with carbon dioxide, decapitated and pinned ventral side up on a cork board. The metathoracic legs were fixed with plasticine at right angles to the body and ventral side up. The ventral thoracic cuticle was removed to expose the metathoracic ganglion and the nerve trunks leaving it. The ipsilateral connectives and nerves 3a, 3b, 3c, 4 and 5 were exposed sufficiently in order that they could be either recorded from or stimulated. Further dissection of the soft cuticle between the rim of the coxa and thorax exposed the anterior coxal adductor muscle for intracellular recording. The ventral cuticle of the femur was removed together with the underlying flexor tibiae and retractor unguis muscles. Silver-wire extracellular recording electrodes were placed under the exposed motor nerve to the extensor tibiae muscle. This nerve was lifted clear of the haemolymph and coated with petroleum jelly.

The inhibitory axon to the extensor tibiae muscle was identified in the record from the motor nerve by stimulating nerve 3c. This gave selective activation of the inhibitory axon (Usherwood & Grundfest, 1965). Stimulation of nerve 3b allowed the slow excitatory axon to be identified and confirmed that the axon activated by stimulation of nerve 3c was in fact the inhibitory axon. The slow excitatory axon has the lowest excitation threshold when the nerve 3b is stimulated (see Fig. 17).

This preparation allowed the activity in the inhibitory axon to the extensor tibiae and in the I-C axon to the anterior coxal adductor to be monitored simultaneously. The effect of stimulating the different nerves on the activity of these two axons could also be investigated.

Any attempt to understand fully the function of the different motor nerve fibres in invertebrates normally requires a reasonably detailed knowledge of the innervation patterns of the muscles. For this reason the innervation patterns of the posterior coxal levator muscles, 182C and 182D, have been investigated. The results of this work are presented before any details are given of the effects produced by the inhibitory axon to these muscles.

Normally in the headless animal only four (1,2, 3 and 4) of the twelve motor axons in nerve 6Br4 were continuously active. Large spontaneous burst of activity occurred periodically during which up to ten axons have been seen to be active. The intensity of these bursts resulted in interaction and summation of the extracellular potentials thus preventing any confident identification of the different axons. Because of this problem of identification, and also because considerable movement made microelectrode studies difficult, the isolated ganglion preparation was used exclusively to study the innervation patterns of muscles 182C and 182D.

In the resting isolated ganglion preparations axons 1–4 were usually continuously active, while axons 5 and 6 fired in bursts. Axons 5 and 6 were readily activated by mechanical stimulation to the ipsilateral tibia and tarsus. The axons larger than 6 could rarely be reflexly activated after ganglion isolation. Pearson & Stein (1969) showed that the amplitudes of the nerve impulses recorded from axons 1 to 6 were remarkably constant from animal to animal. This measurement, together with a knowledge of the discharge patterns of axons 3 and 4, allowed identification of the motor axons 1-6 in any preparation. The amplitudes recorded from axons 3 and 4 were very similar but their discharge patterns differed. Axon 3 was rarely silent between bursts and fired very regularly with frequencies from 1 to 5 impulses/sec., whereas the discharge of axon 4 was usually less regular, and at times it was completely silent. The firing rate of axon 3 increased only for a short time at the beginning of a spontaneous burst of activity in axons 5 and 6 while axon 4 generally fired at a much higher and steadier rate throughout the burst. The activity of axon 3 was often inhibited at the end of a burst. If there remained any doubt about the identification of the impulses recorded from axons 3 and 4, records from nerve 4rza removed this by showing a one-to-one correspondence between the firing of a small axon in this nerve and axon 3 in nerve 6Br4. In the top trace of Fig. 3 a is shown a small burst of activity recorded from nerve 6Br4 while the corresponding record from nerve 4r2a is shown in the bottom trace. Note the one-to-one correspondence between the firing of axon 3 and a single axon in 4r2a. The gain of the top record is insufficient to allow the impulses from axons 1 and 2 to be seen. These are shown in the high-grain record of Fig. 3b.

Identification of the impulses from motor axons 1–6 allowed correlations to be made between the activity of these axons and junctional potentials recorded intracellularly from single muscle fibres of 182C and 182D. In no preparation were any junctional potentials correlated with activity of axons 1 and 2. The muscles that these two axons innervate have not as yet been identified. All four axons, 3-6, innervate muscles 182C and 182D. Figures 4a and b show a hyperpolarizing, or inhibitory, junctional potential (I.J.P.), and a small depolarizing, or excitatory, junctional potential (E.J.P.), recorded from a single muscle fibre of muscle 182D. These junctional potentials are correlated with the firing of axons 3 and 4 respectively. Larger E.J.P’S from another muscle fibre of the same preparation correlated with activity in axons 5 and 6 are shown in Fig. 4c and d. Many muscle fibres were innervated by all four of these axons as shown in Fig. 5. The patterns of innervation of muscles 182C and 182D by the axons 3-6 are summarized in Table 1. There are a number of interesting features about these patterns.

• (1) No muscle fibre in either muscle was found to be innervated by axon 3 or by axon 4 alone.

• (2) The major difference between the two muscles was that in 182C many fibres (24%) were innervated by axon 6 alone and not by any of the axons 3, 4 or 5. Also, a small number of fibres (3 %) received axon 5 and not any of the axons 3, 4 or 6. In muscle 182D these innervation patterns were not seen.

• (3) In both muscles axons 3 and 4 always innervated the same fibres.

• (4) With few exceptions axons 5 and 6 innervated the same fibres of 182D. This was not so for 182C.

• (5) A large number of fibres in both muscles (32% in 182C and 62% in 182D) were innervated by all four axons.

From these patterns the general features emerge that, apart from the fibres in 182C innervated by axon 6 (and not by any of axons 3, 4 or 5), single muscle fibres are either innervated by axons 3 and 4 together or by axons 5 and 6 together, or by all four axons.

The observed amplitudes of the junctional potentials corresponding to the axons 3-6 are shown in Fig. 6. The amplitudes in these histograms have been determined only from cells in which the resting potential was greater than –40 mV. For these cells no significant correlation between junctional potential size and resting potential has been found. In over 150 cells analysed in forty preparations no depolarizing junctional potentials corresponding to activity in axon 3 have been seen. In Crustacea it is well known that inhibitory axons can produce either depolarizing or hyperpolarizing junctional potentials. Hoyle (1966b) reported that the I-C axon to the locust anterior coxal adductor muscle may produce depolarizing junctional potentials. Hoyle used KCl-filled micropipettes (personal communication) and a possible explanation for the occurrence of these depolarizing junctional potentials was that the internal chloride concentration was increased sufficiently by diffusion from the electrode tip to shift the equilibrium potential for the I.J.P. from the resting level in a depolarizing direction. No evidence is available to show that chloride ions are involved in the generation of the junctional potentials in response to I-C axon activity, but this is very likely since they are involved in the generation of the I.J.P. in the extensor tibiae muscle of the locust (Usherwood & Grundfest, 1965) and the cockroach posterior levator muscle (unpublished observations). Thus Hoyle’s observation that the I-C axon may sometimes produce depolarizing junctional potentials must be treated with some reservation. In over twenty preparations using KjSO-filled electrodes to record the response in single fibres of the anterior coxal adductor muscle, no depolarizing junctional potentials have been seen in response to activity in the I-C axon. (See below.)

Because of the distributed nature of the nerve axon endings on a single muscle fibre, the amplitude of the junctional potentials will vary with the position of the recording electrode relative to the nerve endings. The magnitude of this variation will depend on the length constant of the muscle fibres and the distance between nerve endings. If such variation occurs it may partly explain the considerable range of amplitudes observed for the different junctional potentials shown in Fig. 6. It may also explain the common finding that, in a fibre innervated by the four axons 3–6, when the amplitudes of the I.J.P. and E.J.P. produced by axons 3 and 4 respectively were small, then the E.J.P.’S produced by axons 5 and 6 were large, and vice versa. This finding suggests that the terminals for axons 3 and 4 are close together and removed from those of axons 5 and 6 which are also close together. Figure 6 also shows that for axons 4-6 the amplitude of the corresponding E.J.P. increases with increasing fibre size.

For reasons mentioned earlier it was difficult to determine from records of spontaneous activity whether axons larger than 6 also innervated muscles 182C and 182D. To investigate this possibility nerve 6B was stimulated while recording intracellularly from single muscle fibres. Figure 7 shows a large E.J.P. recorded from a fibre of muscle 182D as a result of this stimulation (the stimulus strength was carefully adjusted to be just threshold). These large E.J.P.’S could be recorded from many fibres in both 182C and 182D. The amplitude of these E.J.P.’S could be as high as 50 mV. and their rise time and duration was less than those of the E.J.P.’S produced by axons 4-6. These large EJ.P.’S were typical of those produced by fast axons. Records taken from nerve 6Br4 distal to the stimulating electrodes showed that, compared to the other nerve fibres in this nerve, the axon giving rise to these large EJ.P.’S had one of the lowest activation thresholds, indicating that this axon was one of the largest within nerve 6Br4. In nerve 6Br4 of the mesothoracic segment Dresden & Nijenhuis (1958) have reported that there are seven axons with diameters greater than 10 μ. These workers also showed that the third and seventh largest of these axons innervated all branches of the posterior coxal levator muscle and the sixth largest innervated only branch D of this muscle. The other four axons innervate the main levator muscle. For nerve 6Br4 in the metathoracic segment Pearson & Stein (1969) also reported seven axons with diameters larger than 10 μ which they numbered from 6 to 12 with increasing size. If a similar organization for the distribution of the motor axons to the different levator muscles exists in the metathoracic segment as for the mesothoracic, then axon 10 (axon 10 is the third largest in the trunk and axon 6 is the seventh largest) would be the axon giving rise to the large E.J.P.’S. Although no direct evidence has been obtained that activity in axon 10 does give rise to the large E.J.P.’S, from now on the axon giving these responses will be referred to as axon 10. It remains to be determined whether axon 7 (6th largest) innervates only muscle 182D and if so what effect activity in this axon has on the muscle. The possibility exists that some of the other large fibres also innervate branches of the posterior coxal levator muscle. Small twitches when axon 10 was stimulated, and large graded contractions when axons 5 and 6 were active, tended to displace the microelectrode. For this reason it was usually not possible, after finding whether axon 10 innervated a particular muscle fibre, to determine whether axons 5 and 6 also innervated this fibre, or vice versa. The determination of the innervation patterns of single muscle fibres by the five axons 3-6 and 10 was therefore very difficult. Insufficient stable records were obtained to estimate the fraction of muscle fibres innervated by the different combinations of these five axons.

In a small number of preparations two different-sized I.J.P’S, with similar time courses and effects on the E.J.P.’S, have been recorded from the same muscle fibre. One of these I.J.P.’S corresponds to activity in axon 3 while the other is not correlated with the activity in any of the axons in nerve 6Br4. This is shown in Figs. 5 b-c and 8. The uncorrelated I.J.P. could be the result of either spontaneous transmitter release or activity of an axon in nerve trunk 5r3 (this nerve joins a branch of 6Bt4 and contains motor fibres that may terminate in muscles 182C and 182D). The possibility of spontaneous transmitter release as the cause of these uncorrelated I.J.P’S was rejected because often their amplitude was greater than the I.J.P. produced by activity in axon 3. Records have not yet been made from nerve 5r3 to determine whether the second I.J.P. is in fact correlated with the activity of a motor axon within this trunk. If the uncorrelated I.J.P. is the response of a specific inhibitor to the levators, whose activity is positively correlated with the excitatory activity to the depressors, the fact that in an isolated ganglion preparation the excitatory axons to the depressors are rarely active may explain the small number of preparations in which this I.J.P. was observed.

These studies on the innervation pattern of the muscles 182C and 182D have shown that they are innervated by at least five of the motor axons (one inhibitory and four excitatory) contained in nerve 6Br4, and indicate that they receive an inhibitory axon contained in nerve 5r3. Some single fibres of these muscles are quintuply innervated by the two inhibitory axons and axons 4-6 (Fig. 5), while a few fibres may be innervated by all six axons.

The mechanical effects produced by the four excitatory axons to muscles 182C and 182D are shown in Figs. 9 and 10. Usually no mechanical effect is observable in response to single impulses in axons 4–6, whereas repetitive activity in these axons produces a steady increase in tension. The minimum frequency at which axon 4 begins to produce an increase in tension depends on the activity in the inhibitory axon 3. When axon 3 was firing at about 4 impulses/sec. the minimum frequency in axon 4 to produce tension was about 8 impulses/sec. (Fig. 9 a). A lower minimum frequency to produce tension resulted when axon 3 was discharging at a lower rate. The minimum firing rate of axons 5 and 6 to produce a tension increase was less sensitive to the inhibitory axon activity, and an increase in tension was noticeable when these axons were firing between 20 and 30 impulses/sec. (Fig. 9b). Occasionally single impulses in axons 5 and 6 produced very small twitch-like contractions. The response to single stimulation of axon 10 resulted in a twitch response (Fig. 9c). Fusion of these twitch responses occurred with stimulating frequencies greater than about 50/sec., and the twitch-tetanus force ratio was between 1:2 and 1:3.Figures 10 a and b show the increase in tension produced by spontaneous bursts of activity in axon 4 and in axons 5 and 6 respectively. Note the faster relaxation rate on cessation of activity in axons 5 and 6 than that when axon 4 ceases firing.

Selective activation of the inhibitory axon 3 was possible by stimulation of nerve 4T2a. This is shown in Fig. 11, where the I.J.P. in the record on the right was the response in a single muscle fibre of 182D to stimulation of 4r2a. This I.J.P. is seen to be identical with the I.J.P. on the left record which resulted from a spontaneous firing of axon 3. The latency of the stimulated response was about 4 msec.

To study the effect of the I.J.P. on the various E.J.P.’S axon 3 was stimulated regularly at i/sec. and the interaction between the stimulated I.J.P. and spontaneous and reflex activated E.J.P.’S was recorded. The series of records in Fig. 12 shows the interactions of the I.J.P. with E.J.P.’S produced by axon 4. In the top row of the drawings on the right these interactions, together with some not shown in the records on the left, have been redrawn and superimposed. The generally observed effect was a considerable reduction in the level of depolarization reached by the E.J.P.’S. The second row in this figure was obtained by subtracting the I.J.P. and shows the variation in amplitude of the E.J.P.’S. The graph in this figure is a plot of the relative amplitude of the E.J.P.’S against the time interval between their arrival and the arrival of the I.J.P. Initially, in all but 3 % of the fibres, there was a considerable reduction in amplitude of the E.J.P. This reduction lasted about 16 msec. In 87% of the cells the reduction in amplitude was followed by a facilitation. The reduction in amplitude is simply explained by the conductance change produced by the inhibitory transmitter tending to clamp the membrane potential at the equilibrium potential of the I.J.P. During the tail of the I.J.P., where the inhibitory conductance change has presumably declined to a small value, there will be a larger driving force to generate the E.J.P.’S owing to the membrane being hyperpolarized. (This facilitation effect may also be seen from the records of Usherwood & Grundfest (1965) for the interaction of the I.J.P. and E.J.P. produced by the inhibitory and slow excitatory axons in a single fibre of the locust extensor tibiae muscle.) the I.J.P. had a smaller but qualitatively similar effect on the E.J.P. produced by axon 5 and very little effect on the E.J.P. produced by axon 6. No results from these experiments suggested that axon 3 gives rise to anything other than post-synaptic inhibition in the muscle fibres. However, the possibility of a small pre-synaptic effect cannot be ruled out until experiments similar to those of Dudel & Kuffler (1961) have been carried out.

In a few preparations stimulation of axon 3 produced a slight relaxation in the resting muscle. Generally, however, if the excitatory axons were not active, no mechanical effect was seen. Contractions of the muscles were either spontaneously or reflexly evoked, and Fig. 13 shows the effect of stimulating the inhibitory axon at 50/sec. All three records are from different preparations and the obvious effect was a marked tension decrease. The mechanical inhibition was most marked when the muscles were contracting under the influence of the small excitatory axon 4 alone, usually producing complete mechanical inhibition. The decrease in tension was smaller when the muscles were contracting under the influence of the larger excitatory axons. This is shown at the first stimulation period of the middle record in Fig. 13. Sometimes during very strong contractions, when axons 5 and 6 were discharging at high frequencies, no noticeable mechanical inhibition was produced on stimulation of the inhibitory axon. However, as the activity in axons 5 and 6 decreased the amount of mechanical inhibition increased.

As already mentioned, stimulation of nerve trunk 4r2a gives selective activation of the inhibitory axon 3 in nerve 6Br4 Since there are no sensory fibres in 4r2a (Pipa & Cook, 1959) this means that there is either a tight synaptic junction between a motor nerve fibre in 4T2a and axon 3 or a single motoneurone sending branches down both nerves 412a and 6Br4, the branch in 6Br4 corresponding to axon 3. Evidence for the latter of these possibilities comes from the following observations. Simultaneous recordings from nerve 6Br4 and various other ipsilateral nerves (3A, 3B, 4, 5r1a and 6A) show a one-to-one correspondence of the firing of the inhibitory axon 3 in 6Br4 and a single axon in the other nerves. An example of this is shown in the top set of records of Fig. 14. Stimulation at each recording site in turn selectively activated, with the same latency, the corresponding axon seen at the other site (Fig. 14, left set of records). This selective activation would continue on a one-to-one basis at frequencies of stimulation of 1oo/sec. and higher (Fig. 14, right set of records). The simplest explanation for these findings is the existence of a single motoneurone sending axon branches down most of the ipsilateral nerve trunks, the branch in 6Br4 corresponding to the inhibitory axon 3. However, they do not exclude the possibility that there exists a group of neurones each sending its axons down different nerve trunks and firing synchronously because of tight electrical coupling. The synaptic junctions between these neurones would have to be capable of transmitting impulses on a one-to-one basis in both directions at frequencies greater than 100/sec. The septal and commissural electrical synapses of the crayfish lateral giant axons show these properties, and electrical connections between inhibitory axons have been seen in the lobster abdominal ganglia (Otsuka, Kravitz & Potter, 1967). Even if such a group of nerve cells does exist, functionally the group operates as a single motoneurone with many axon branches.

The question now arises as to whether the other axon branches apart from axon 3 of this motoneurone have inhibitory effects in the muscles they innervate. Nerve filaments containing these branches have been traced to various muscles (see Fig. 19 a) and intracellular records have been taken from single fibres of these muscles. Hyperpolarizing junctional potentials correlated to the firing of axon 3 have been recorded from the sternal and tergal remotor muscles (172 and 174) and the coxal depressor muscle (177E). Figure 15 shows an I.J.P. recorded in a single fibre of 177E correlated with the firing of axon 3, top trace. The remotor muscles are very lightly innervated by the inhibitory axons (about 1%) while about 50% of the fibres of 177E receive inhibitory innervation. All these results thus demonstrate the functional existence of a common peripheral inhibitory neurone.

The distribution of the axons of the common inhibitor that has so far been found is shown in Fig. 19 a. Owing to the large size of the connectives and nerve trunk 5 it was not possible to record impulses from axons of the same size as the branches of the common inhibitor in these nerve trunks. For this reason the only evidence for the existence of branches of the common inhibitor in the ipsilateral abdominal and thoracic connectives and nerve trunk 5 (after branch 2) is that stimulation of these nerve trunks gave short-latency selective activation of axon 3 in nerve 6Br4, which would continue on a one-to-one basis at frequencies higher than 100/sec. The fact that stimulation of the ipsilateral connective between the prothoracic and suboesophageal ganglia gave similar selective activation of the common inhibitor indicates that the branch leaving the ipsilateral thoracic connective possibly terminates in the brain of the animal. The branch in the ipsilateral abdominal connective does not go further than the first abdominal ganglion. No evidence has been found to suggest that the common inhibitor sends branches to the contralateral nerve trunks, nor to any lateral nerve trunks of the meso- or prothoracic ganglia.

These patterns were investigated in four different preparations: (1) in the intact animal, (2) in the headless animal, (3) in animals with the connectives cut between the meso- and metathoracic ganglia, and (4) in animals with the metathoracic ganglion isolated. The aim of these experiments was to determine when the common inhibitor fired relative to activity in the excitatory axons to both levator and depressor muscles. In preparations ( 1 )–(3) the discharge rate of the common inhibitor i ncreases considerably at the beginning of a spontaneous burst of activity in the excitatory levators motoneurones, with interspike intervals as low as 12 msec. This increase in firing rate is maintained throughout the burst. At the end of such a burst the discharge rate of the common inhibitor is reduced to less than that before the burst. In the isolated ganglion preparation the discharge rate increases only slightly at the beginning of a burst of activity and is not always maintained throughout the burst. Figure 16 shows simultaneous records of spontaneous activity in nerve 5r1a to the depressor muscle 177E, and in nerve 6Br4-These records show a strong reciprocal relationship between the activity of axons 5 and 6 and a single axon in 5r1a (extracellular recordings from the extensor trochanteris muscle show this axon to be the slow excitatory axon described by Pringle, 1940). During these periods of reciprocal activity the firing rate of the common inhibitory neurone increases at the beginning of the burst of activity in levator excitatory motoneurones concomitant with the complete inhibition of activity of at least one excitatory axon to the coxal depressor muscles.

Usherwood & Grundfest (1965) reported that selective activation of the inhibitory axon to the extensor tibiae muscle was possible by stimulation of nerve trunk 3c. To explain this they postulated a monosynaptic connection from sensory fibres in 3c to the inhibitory neurone. Another possibility, which they did not consider, was that the inhibitory neurone sent branches down both nerve trunks 3c and 3b. If the inhibitory neurone sent a single axon from the ganglion which branched outside the ganglion, then cutting nerve 3 close to the ganglion would not prevent this selective activation. Figure 17a demonstrates that with nerve trunk 3 cut close to the ganglion the inhibitory axon to the extensor tibiae is still selectively activated by stimulating nerve 3c. The postulated monosynaptic reflex pathway must therefore be rejected.

Simultaneous recordings from nerve 3c and the motor nerve to the extensor tibiae showed that there was a one-to-one correspondence between the firing of the inhibitory axon in 3b and a single axon in nerve 3c. Tracing the axon in 3c whose activity corresponded to that of the inhibitory axon in 3b, indicated that it innervated the anterior coxal adductor muscle, and suggested that the inhibitory axon to the extensor tibiae and the I-C axon to the anterior coxal adductor were in fact axon branches of a common inhibitory neurone. Direct evidence for this was obtained by simultaneously recording the activity of the inhibitory axon from the motor nerve to the extensor tibiae and intracellularly from single muscle fibres of the anterior coxal adductor. Figure 18 shows I.J.P.’S recorded from the anterior coxal adductor correlated with the firing of the inhibitory axon of the extensor tibiae. This common inhibitory neurone exists in both the species Locust migratoria and Schistocerca gregaria.

Using the above method (see Fig. 14), branches of the common inhibitory neurone were found to be sent out in nerves 3a and 4 also but not in nerve 1 and 2. Stimulation of the ipsilateral thoracic connective and nerve trunk 5 gave short-latency activation of the common inhibitor. One-to-one activation was possible at frequencies greater than 100/sec. Furthermore, stimulation of the ipsilateral connective between the prothoracic and sub-oesophageal ganglia also activated the common inhibitory neurone on a one-to-one basis at frequencies greater than 100/sec. The finding that the common inhibitory neurone could be selectively activated by stimulating the ipsilateral thoracic connective is consistent with Hoyle’s finding that stimulation of this connective would produce selective activation of the I-C axon to the anterior coxal adductor muscle (Hoyle, 1966 a). These results suggest that the common inhibitor neurone in the locust is similar to that in the cockroach in that it may send out axon branches in both the ipsilateral thoracic connective and nerve trunk 5. The distribution of the common inhibitory neurone that has so far been found in the locust is shown in ‘Fig. 19 b.

A feature of the innervation pattern of single fibres of the posterior coxal levator muscles 182C and 182D is the pairing of axons 3 and 4 and axons 5 and 6, and it is of interest to know whether this pairing is of any functional significance. In the intact animal axon 4 is generally continuously spontaneously active and produces graded contractions in the posterior levator muscle when firing at frequencies as low as 5/sec., provided that axon 3 is silent. The firing rate of axon 4 can change slowly, resulting in slow tension fluctuations in the muscle. Thus axon 4 is probably involved in the maintenance of leg position when the animal is standing, and the inhibitory axon 3 may act to selectively inhibit this tonic response. The pairing of the small excitatory axon 4 with the inhibitory axon 3 is similar to that in the extensor tibiae muscle of the locust where Usherwood & Grundfest (1965) observed that the inhibitory axon only innervated those muscle fibres receiving the slow excitatory axon. The discharge patterns of axons 5 and 6 are very similar, and different from that of axon 4. Both axons 5 and 6 fire in bursts of activity, axon 5 beginning to discharge earlier and ceasing to fire later than axon 6. The mechanical effects produced by axons 5 and 6 are considerably larger than those produced by axon 4 and are not apparent unless they are firing at frequencies greater than 20–30 impulses/sec. When active, these axons rarely discharge at rates less than 30/sec., and the mechanical responses produced are rapid graded contractions. Thus functionally axons 5 and 6 are very similar and are probably involved in leg levation when the animal is walking. This possibility is supported by the observation of the very strong reciprocal relationship between the firing of axons 5 and 6 and an excitatory axon to the coxal depressor muscle 177E (Fig. 16).

In the intact animal fast axon 10 sometimes fires during very large bursts of activity of levator motoneurones (indicated by twitch responses in tension records corresponding to the firing of a large axon in nerve trunk 6Br4). Thus this axon probably becomes active to produce extremely rapid levation of the leg when the animal is running, or removing its leg from a noxious stimulus.

When considering the normal functioning of the coxal levator muscles in the cockroach a comparison with the abdominal flexor system of the crayfish is useful. Kennedy & Takeda (1965) showed that the twitch and tonic abdominal flexor muscles of the crayfish are anatomically separated. The superficial muscles are innervated by one inhibitory and five excitatory axons. Tonic activity in the excitatory axons gives rise to slow contractions of these muscles and they are involved in the steady maintenance of tail position. The deeper flexor muscles on the other hand are innervated by nine large excitatory axons all of which produce twitch contractions. A similar anatomical separation exists to some extent in the metathoracic coxal levator muscles of the cockroach, the large main levator muscle (181) being innervated by at least four fast axons whose activity gives rise to twitch contractions (unpublished observations), and the posterior coxal levator muscle which is involved in the tonic maintenance of posture and relatively slow leg levation.

A potentially enormous variety of control over the rate and degree of leg levation by the coxal levator muscles is available to the animal. This variety ranges from very slow tonic responses produced by slow changes in activity of axon 4 to extremely strong and rapid responses produced by high-frequency ( > 50/sec.) activity in the fast axons. Activity in the inhibitory axon 3 and one other inhibitory axon (discussed below) enrich this variety.

Intracellular recordings from some single muscle fibres showed two different-sized I.J.P.’S, one being correlated to the firing of axon 3 (Fig. 8). As discussed earlier, the most likely explanation for the cause of the uncorrelated I.J.P.’S was activity in a motor axon contained in nerve trunk 5. The verification of this must await the results from experiments in which simultaneous recordings are made from 5r3 and single muscle fibres of 182C and 182D. Since the two inhibitory axons that have been described in the locust (Usherwood & Grundfest, 1965; Hoyle, 1966b) are branches of a single common inhibitory neurone, there is now no definite example in insects, with the possible exception of the inhibitory axon to the basalar fibrillar muscle of the beetle, of specific inhibitory neurones. The axon giving rise to uncorrelated I.J.P.’S in the posterior coxal levator muscle of the cockroach may be a specific inhibitory neurone. Another example of a possible specific inhibitory axon is that supplying the coxal depressor muscle 177E where intracellular recordings from some single muscle fibres again show two different sized I.J.P.’S one of which corresponds to activity in the common inhibitory neurone (unpublished observations). It is of interest that recordings from nerve trunks 5r1a and 5r3 show that a small axon in each becomes active when the excitatory axons to the levators and to the depressors are active respectively (unpublished observations). This is what would be expected for specific inhibitory neurones to both sets of muscles if these neurones operated in a similar way to those supplying the abdominal muscles of the crayfish (Kennedy, Evoy & Fields, 1966), i.e. their firing being negatively correlated with the firing of the excitatory axons to the same muscle.

The findings described above on the neuromuscular organization of the posterior coxal levator muscles, 182C and 182D, of the cockroach considerably extend the current picture of the possible types of innervation of insect muscle, and show that the neuromuscular organization of these muscles is as complex as that for any crustacean muscle (cf. Atwood, 1967).

Figure 19 shows the distribution of the common inhibitory neurones that has so far been determined for the metathoracic segment in the cockroach (Fig. 19 a) and in the locust (Fig. 196). In both animals the general properties of these neurones show several similarities. First the various branches innervate functionally different muscles (synergic and antagonistic); secondly, only muscles involved in walking appear to be innervated as no branch is sent out in nerve trunk 2 in the cockroach or in nerve trunks 1 and 2 of the locust (the axons in these nerve trunks only innervate thoracic and wing muscles); and thirdly, one branch terminates in either the suboesophageal ganglion or brain of the animal.

Because the first three abdominal ganglia in the locust are fused with the third metathoracic, it is not possible to determine whether there is a branch of the common inhibitory neurone in the locust that corresponds to that in the cockroach, which terminates in the first abdominal ganglion. Stimulation of the abdominal nerve trunks leaving the first abdominal ganglion in both animals did not excite the common inhibitory neurone, indicating that the neurone does not directly innervate abdominal muscles. Stimulation of the ipsilateral abdominal nerve cord posterior to the first abdominal ganglion resulted in activation of the common inhibitor, but the variable latency and inability to follow high-frequency stimulation suggested the existence of synaptic connections in the activation pathway. The branch of the common inhibitory neurone sent to the suboesophageal ganglion or brain of the animal does not send out further branches in the nerve trunks leaving the meso-and prothoracic ganglia, as stimulation of any of these trunks does not result in activation of this neurone. Although only the common inhibitory neurone to musculature of the metathoracic segment has been described in this paper, a similar neurone exists in the mesothoracic ganglion of the cockroach (unpublished observations).

Common inhibitory neurones have not previously been described in insects but exist in many species of Crustacea (Wiersma, 1961). In comparing the two groups of animals it is of interest that branches of common inhibitory neurones in Crustacea similar to the type described in this paper (i.e. innervating functionally different muscle groups), have only been found to innervate limb musculature. This is apparently also true for the common inhibitory neurones of insects. The question is raised of whether this type of common inhibitory neurone has been specially developed for the control and co-ordination of limb musculature.

In crustacea the function of the common inhibitory neurones has always been an enigma, but the commonly held view is that they become active when the animal is moulting. This prevents the occurrence of strong muscular contractions and hence protects the very soft cuticle (Hoyle, 1957; Wiersma, 1961). There are a number of reasons why this suggestion is inappropriate for the function of the common inhibitory neurones in insects. First, the common inhibitor has little inhibitory effect when muscles are contracting under the influence of activity in the fast fibres (Usherwood & Grundfest, 1965) and in some cases the activity of the common inhibitory axon can facilitate the mechanical response (Hoyle, 1966b). Secondly, many of the most powerful muscles in the legs of insects are not innervated by the common inhibitory neurone. Finally, in the cockroach where the activity of the common inhibitor has been monitored during, and just after, the final moult, the activity of the common inhibitory neurone is, if anything, less than in the normal animal (unpublished observations).

In insects there is a variation of the potassium concentration in the haemolymph depending on the diet of the animal (Pichon & Boistel, 1966). As some of the tonic muscle fibres have a very low threshold for contraction, and may in fact always be partially contracted, the fluctuations in external potassium would alter their resting contractile state; an increase in potassium concentration producing a greater tonic contraction. Small fluctuation in external potassium would not affect the twitch fibres as the threshold for contraction in these fibres is quite high (Usherwood, 1967). In the posterior coxal levator muscles 182C and 182D of the cockroach every muscle fibre innervated by the small excitatory axon 4, which gives rise to slow graded contractions, also receives the common inhibitory axon 3. Also many of the fibres in the extensor tibiae muscle of the locust receiving the slow excitatory axon are innervated by the common inhibitory neurone (Usherwood & Grundfest, 1965). Thus another possible function of the common inhibitory neurone in insects may be to counteract the fluctuations in the external potassium concentrations; the activity increasing with an increase in external potassium. This would result in the tonic muscle fibres being maintained at the same degree of resting contraction. This suggestion is far more likely than the first but there remain a few difficulties. In the locust, the anterior coxal adductor muscle receives one branch of the common inhibitory neurone but is not innervated by a slow axon, and also a number of fibres in the extensor tibiae muscle having slow fibre innervation are not innervated by the common inhibitor. In the cockroach many muscles in the mesothoracic coxa have slow fibre innervation but do not appear to be innervated by inhibitory axons (Usherwood, 1962). Finally, the common inhibitory neurone would have to be selectively sensitive to variations in external potassium concentration. Before this suggestion for the function of the common inhibitory neurone can be discarded, the effect of variation of external potassium concentration on the relative discharge patterns of the common inhibitory and slow excitatory neurones must be studied.

The most plausible function for the common inhibitory neurones is suggested by the results of the experiments upon the discharge pattern of these neurones relative to those of the excitatory neurones to various muscles. Hoyle (1966b) reported that the inhibitory conditioning axon to the anterior coxal adductor (this axon being one branch of the common inhibitor) tended to fire just before the excitatory axon to this muscle. Runion & Usherwood (1968) reported that when the animal is walking regularly the inhibitory axon to the extensor tibiae muscle (this axon being another branch of the same common inhibitor) fires just before the beginning of activity to the flexor tibiae. During walking both the anterior coxal adductor and flexor tibiae muscles function at the same time to lift and promote the tarsus respectively. Thus in the locust, activity in the common inhibitory neurone precedes the stepping phase of walking. This activity can sometimes facilitate the mechanical response to the excitatory axon in the anterior coxal adductor and may cause a faster relaxation of the extensor tibiae muscle. Thus the over-all effect of the activity in the common inhibitor may be to cause a more rapid and stronger levation and promotion of the leg when the animal is walking.

In the cockroach a somewhat similar discharge pattern of the common inhibitory neurone is seen. Here the maximum firing rate of the common inhibitor occurs at the beginning of a burst of activity of excitatory motoneurones to the coxal levators. There is a strong reciprocal relationship between the activity of the excitatory axons larger than 4 to the levators and the activity of a large excitatory axon to the depressor muscle 177E (Fig. 16). About 50% of the fibres of muscle 177E are innervated by a branch of the common inhibitory neurone. The common inhibitor could therefore function to facilitate the relaxation of the depressor muscle 177E, this allowing a more rapid leg levation. Activity in this neurone would only slightly affect the rate of contraction of the posterior levator muscle as this contraction is due mainly to activity in the large excitatory axons. An increased rate of activity of the common inhibitor is maintained throughout the excitatory burst to the levators and hence during this activity it would function as an inhibitor to the depressor muscles. At the end of this burst the activity of the common inhibitor is reduced. Thus it does not function symmetrically so as to facilitate relaxation of the posterior coxal levator muscles. Throughout the period when the excitatory axons to the depressor muscles are active the discharge rate of the common inhibitor is low. Why then does this neurone innervate the posterior coxal levator muscles at all? One possibility is that the initial rate of rise of tension in the posterior coxal levator muscle is facilitated by the initial increase in firing rate of the common inhibitor. Since many muscle fibres receiving the large excitatory axons are not innervated by axon 3 (the common inhibitor), and since activity in this axon always gives rise to hyperpolarizing junctional potentials, it is difficult to understand by what mechanism this would occur.

Apart from the data from observations of the discharge patterns of the various motoneurones, another finding that must be considered when trying to understand the function of the common inhibitor is that axons 3 and 4 always innervate the same fibres of muscle 182C and 182D, activity in axon 3 having a very potent effect on decreasing tension produced by axon 4. The mechanical responses produced by axon 4 indicate that it is involved in the maintenance of posture when the animal is standing rather than playing any important role during walking. Therefore another possible function for the common inhibitor is in the regulation of posture, the tonic degree of contraction of the posterior coxal levator muscles depending on the ratio of the discharge rates of axons 4 and 3. Similar relationships between tension and the ratio of the discharge rates of slow excitatory axons to that of the common inhibitory neurone may exist in other muscles innervated by the common inhibitor. All that would be required for this neurone to function as a leg position regulator is that it has unequal effects in antagonistic muscles. Small changes in leg position may therefore be brought about by changes in activity in a single inhibitory neurone rather than changes in activity in many slow motor axons.

Thus in the cockroach, the common inhibitory neurone may have two different functions. First, to act as a leg position regulator when the animal is standing, and secondly, to facilitate relaxation and inhibit excitatory responses in the depressor muscles during leg levation when the animal is walking.

We wish to thank the Wellcome Trust for a grant to one of us (K. G.P.), and Drs R. B. Stein from this laboratory and P. L. Miller of the University Department of Zoology for their helpful suggestions and criticisms of this paper.