1. The isolation of a thoracic ganglion from the rest of the central nervous system results in a loss of differentiation of the motor output, although repetitive rhythms may appear during the later stages of isolation. Total isolation of the ganglion in vitro results in a further reduction of motor activity to low-frequency, steady-level discharges in a few fibres of some nerves only.

  2. Two or three months after implantation a steady low-frequency discharge can be recorded from many of the branches of the implant ganglion, and these may have functional contacts with adjacent muscles. There is little evidence of afferent connexions.

  3. Four to seven months after implantation the efferent connexions of the implanted ganglion often show a highly differentiated pattern of spontaneous electrical activity, and the ganglion will respond in a remarkably delayed and progressive manner to the stimulation of adjacent sense organs.

  4. The spontaneous rhythms of the long-term implant ganglion may be determined by a balance between central and peripheral input levels similar to those occurring during progressive isolation of the ganglion.

  5. The functional relationship between the host and the donor ganglion appears to consist largely of an inhibitory effect exerted by the host ganglion on the donor or implant ganglion. A justification for this in adaptive terms can be found.

The behaviour of ganglia or central nervous explants introduced into a situation where it is possible for them to make functional connexions with the muscles and sense organs of the host has been very little investigated. Bodenstein’ s work in 1957 on nerve regeneration in the cockroach demonstrated very clearly the possibilities offered by insect material for work of this kind. He showed that thoracic ganglia implanted into the coxae of a host insect survive for many weeks and appear to make some kind of functional connexions with adjacent muscles. Further studies in this laboratory indicated that thoracic ganglia make more effective thoracic implants than abdominal ganglia do, and that facilitating responses in the tibial depressor muscle follow stimulation of tibial spines in the absence of the host nervous system (Guthrie, 1966). It has also been suggested that the donor or implant ganglion only innervates previously denervated muscles (Jacklet & Cohen, 1967).

Several questions can be posed by experiments of this kind, but the most central of these are : whether a ganglion can develop patterned behaviour when isolated from the rest of the central nervous system, and whether this pattern can be varied in an adaptive manner.

Recent work in this laboratory has been aimed at investigating in more detail the autonomous activity of long-term implant ganglia, and these observations form the basis of this communication.

Larval mesothoracic ganglia were implanted into the coxal or subcoxal region of the mesothorax of 4th to 8th instar larvae. Ninety-three insects were cultured, but only 41 of these gave useful results. Many of the operations were performed using sterilization by ultraviolet light, and the instruments were washed in alcohol and flamed before use. After suitable growth periods of 2–7 months the implant ganglion was examined in situ, or it was excised together with many of its nervous connexions and examined in a saline bath. In vitro preparations were pinned out with glass threads in a wax-lined Perspex chamber filled with aerated saline. The composition of the saline followed the formula given by Becht, Hoyle & Usherwood (1960). Ganglia and attached muscles did not lose their activity when stored at 4° C. for 2 or 3 days in this saline.

External recordings were made from nerves and muscles using 30 μ silver wire electrodes, and unit contact records were obtained using electrolytically sharpened tungsten wires insulated with Perspex. The tip diameters of the microelectrodes were between 1 and 3 μ. Signals were amplified using standard a.c. ‐coupled amplifiers of high gain, and displayed on a multichannel oscilloscope or penwriter assembly. Rates of impulse discharge were monitored by means of a pulse counter and timer, incorporating a level selector and pulse shaper.

An attempt was made to utilize the excitatory or depressant effects of various insecticides and drugs (D.D.T., B.H.C., Strychnine, 5‐hydroxytryptamine, and nicotine), but only nicotine proved reliable. At concentrations of 1 × 10–4, augmentations of single-unit discharge rates of as little as 50% could be effected if the drug was introduced into the lateral neuropile through a glass pipette with a tip of about 20 μ diameter. Under these conditions there appeared to be no conduction block.

Patterns of activity in the normal mesothoracic ganglion

Considerable difficulty was encountered in describing the behaviour of an implanted ganglion due to the absence of information on the activity of the normal ganglion in vivo and in vitro, and therefore a brief study of this was made, and is presented at the beginning of this paper as an introduction to its main subject..

The aspects of ganglionic activity to which most attention was devoted were: (1) the reduction in the level of activity following the removal of a ganglion from the rest of the central nervous system; (2) the degree of ganglion isolation that appeared to be most closely associated with the appearance of regular bursts of activity in the motor axons; and (3) the distribution of active fibres between the different nerves of a freshly excised ganglion. The methods which corresponded with these aspects of the inquiry were: (1) External recordings from the anterior proximal region of the tibial depressor muscle fed to a high-gain amplifier, and then to a pulse counter-timer. (2) Simultaneous recordings from external leads on nerve 5 and an ipsilateral posterior connective, and from an extracellular microelectrode penetration of the lateral neuropile of the ganglion. (3) External recordings from the nerves of an excised ganglion in vitro. In the first two kinds of preparation the insects were pinned out so that the limbs projected freely into space and spontaneous movements were not inhibited; furthermore the ganglia of the central nervous system were cut away during the course of the experiment.

Central control of motor activity

The tibial depressor muscle (muscle 143) is a convenient point in the motor periphery to observe the contribution of different parts of the central nervous system as it has a fairly simple innervation, and in the pinned out ‘no stimulus’ preparation exhibits an output of high variability. Some aspects of the behaviour of this muscle have been briefly described elsewhere (Guthrie, 1967). The muscle appears to receive five ‘fast’ axons, two ‘slo20–30 sec. when there is now’ axons, and one inhibitory axon. In the proximal anterior region of the detectable discharge at the 80detectable discharge at thee muscle there is a clear distinction between very large ‘slow’ potentials associated with isometric tensions up to about 3 g., and small, gradually augmenting trains of muscle potentials which are seldom associated with tensions above 800 mg. Under ‘no-stimulus’ conditions there are periods of 20–30 sec. when there is no detectable discharge at the 80 μV. level, although very faint rhythms can be sometimes detected below this level—as shown in Fig. 2 D. These are interrupted by spontaneous bursts involving ‘fast’ and ‘slow’ elements, and 80% of these bursts occur synchronously in the tibial levator and depressor in the pinned-out preparation where there is no tarsal inhibition, thus producing co-contractions. The explanation of this paradox is worth mentioning, as synchronous bursting is a feature in the graft innervations of tibial levator and depressor. Maximal co-contraction of the muscles moving the tibia results in tibial depression for three reasons: first, the depressor muscle has a maximum isometric tension of about 9 g., as against 4 g. in the levator; secondly, the normal rest position of the tibia is 6o° below a line passing through the femoral axis, or is inclined 6o° towards the depressor muscle insertion; thirdly, the depressor muscle insertion is further from the point of articulation between femur and tibia than the levator insertion, in the proportion of 3 to 2. Thus the depressor muscle is effectively much more powerful than the levator, which merely acts as a rather weak brake. Complete levation of the tibia is accompanied by a burst of muscle potentials in the levator only, aided by a large elastic component at the joint resulting from the large rotation during maximal depression (120° from the rest position). Under these experimental conditions the system could be regarded as a dual synchronous output with inhibition available on the depressor pathway only.

By making consecutive 10 sec. counts of muscle potentials from the tibial depressor, and extracting the standard deviation, range and arithmetic mean from groups of 20 counts, it is possible to characterize the result of progressive isolation of the ganglion. A balance had to be struck initially between leaving the preparation too long before counting, so that degradation due to damage began to appear, or too short a time so that the effect of stimulation due to cutting nerve fibres was still evident while the effect of removing active input had not developed. Eventually, a 5 min. rest period was judged to give most consistency, but individual variation was always high, and Fig. 1 shows the results from the most representative of eight trials.

Fig. 1.

Changes in the motor activity of a ganglion following the progressive removal of the rest of the scentral nervous system. Each point is derived from twenty consecutive 10 sec. counts of muscle potentials recorded in the anterior proximal zone of the tibial depressor muscle after various treatments. ▴, range; ○, standard deviation; •, mean, a, Intact preparation; b, removal of the protocerebrum ; c, removal of the rest of the brain ; d, removal of the sub-oesophageal ganglion ; e, removal of the abdominal nerve cord ; f, removal of the prothoracic ganglion ; g, removal of the metathoracic ganglion ; h, removal of the contralateral segmental nerves; i, removal of all ipsilateral nerves with the exception of n. 5.

Fig. 1.

Changes in the motor activity of a ganglion following the progressive removal of the rest of the scentral nervous system. Each point is derived from twenty consecutive 10 sec. counts of muscle potentials recorded in the anterior proximal zone of the tibial depressor muscle after various treatments. ▴, range; ○, standard deviation; •, mean, a, Intact preparation; b, removal of the protocerebrum ; c, removal of the rest of the brain ; d, removal of the sub-oesophageal ganglion ; e, removal of the abdominal nerve cord ; f, removal of the prothoracic ganglion ; g, removal of the metathoracic ganglion ; h, removal of the contralateral segmental nerves; i, removal of all ipsilateral nerves with the exception of n. 5.

The considerable degree of variance implicit in the high range and high standard deviation of impulse samples from the intact animal may be regarded as indicating a high temporal selectivity of central pathways, and the degradation of this selectivity is more marked than the drop in mean numbers of impulses, following successive stages in the isolation of the mesothoracic ganglion. Cutting away the protocerebrum often produced an initial period with little activity, while a greater level resulted from removal of the rest of the brain. Removal of the contra-lateral nerves in the mesothoracic segment also heightened activity, as did bisection of the ganglion (not shown in Fig. 1), and this fits in well with the observations on transverse inhibition by other authors, one of the more recent being Rowell (1964). Regular bursting rather than unpattemed bursting or steady discharge often followed removal of the abdominal nerve cord and segment isolation rather than other lesions. The commonest frequency was 20–30 bursts per minute.

When input to the mesothoracic ganglion has been reduced to the afferents in the 5th nerve of one side, the loss of scatter in terms of standard deviation is about 75 %, and in range is 80%. Mean levels are much less affected, being only reduced by 50%, perhaps largely due to the fact that once inhibition of tonic motor elements is removed and phasic activity drops away, a steady discharge results and the mean and standard deviation curves must flatten out.

The origin of patterns of activity in the ganglion

The extent to which events occurring in the connective and the neuropilar regions of the ganglion adjacent to the root of the 5th nerve are connected with activity in this nerve is not clear, especially with regard to the appearance and regularity of motor bursts. In order to shed some light on this, simultaneous records from a connective, from the 5th nerve and from the lateral neuropilar region of the ganglion were made while the central nervous system was gradually cut away. Oscilloscope records taken from a preparation of this kind are shown in Fig. 2A‐C.

Fig. 2.

Aspects of normal ganglionic activity. A‐C. Recording points. Lateral neuropile, upper trace; posterior connective, second trace; ipsilateral 5th nerve, third trace. Note the single independent bursts in A, and in B the cessation of activity in the lateral neuropile during activity in the nerve and in the connective. In C the ganglion has been isolated and a regular rhythm appears in the nerve. D. Low-amplitude rhythm in the tibial depressor muscle. Two parts of the same record are illustrated. E. Simultaneous records from the ipsilateral posterior connective (upper trace), and from the left mesothoracic fifth nerve (lower trace) in a totally excised ganglion. The frequency of the time traces in all records is 15 eye./sec., with the exception of D which indicates 1 sec. intervals.

Fig. 2.

Aspects of normal ganglionic activity. A‐C. Recording points. Lateral neuropile, upper trace; posterior connective, second trace; ipsilateral 5th nerve, third trace. Note the single independent bursts in A, and in B the cessation of activity in the lateral neuropile during activity in the nerve and in the connective. In C the ganglion has been isolated and a regular rhythm appears in the nerve. D. Low-amplitude rhythm in the tibial depressor muscle. Two parts of the same record are illustrated. E. Simultaneous records from the ipsilateral posterior connective (upper trace), and from the left mesothoracic fifth nerve (lower trace) in a totally excised ganglion. The frequency of the time traces in all records is 15 eye./sec., with the exception of D which indicates 1 sec. intervals.

Bursts might occur at any one of the recording points without being reflected in a change of activity in the other two. Examples of bursts within the ganglion not clearly observable in the posterior connective or in nerve 5 can be seen at the beginning of Fig. 2 A and of Fig. 2B.

Activity at the recording point within the ganglion remained at a high level despite successive lesions, until the 5th nerve was cut, when it dropped to a very low level.

There was a striking correspondence between the shutting down of certain elements in the ganglion at the same time as a discharge developed in the connectives, and sometimes also in nerve 5. This can be seen in Fig. 2A towards the end, and in Fig. 2B.

Regular bursts occurred in the nerves when the segment was isolated, and more strikingly when ipsilateral nerve 3 was cut, as can be seen in Fig. 2C. In this record a large burst can be seen to occur within the ganglion that in no way disturbs the rhythm in nerve 5.

The excised ganglion in vitro

The aim of these observations was to see which, if any, of the nerve trunks contained spontaneously active motor elements that merely required the development of neuromuscular junctions to initiate muscular contractions.

A distinction was made between those fibres whose presence could be demonstrated by simply lifting a nerve trunk above the level of the saline and recording immediately, and those which were only revealed by prolonged drying out of the nerve. In some cases the characteristics of the activity obtained by the second method were rather similar to those provided by microelectrode penetrations and may have been due to intra-ganglionic rather than peripheral fibres. Where there were large differences in signal magnitudes from the wet nerve, a classification into large and small is introduced into Table 1. The level of distinction corresponds to about 80 μV. The results shown in Table 1 may be found to vary a little from one ganglion to another, but if male mesothoracic ganglia are used and allowed 15 min. or so to settle down after excision’ a considerable degree of uniformity will be observed.

The information given in Table 1 suggests that at the time of implantation large fibres with a sustained steady discharge are to be found in nerves 2 and 5, and in the posterior connectives. The lack of easily demonstrable elements in the anterior connectives, together with the fact that a major part of nerve 2 rejoins the anterior connective near the prothoracic ganglion, suggests that perhaps the active fibres in nerve 2 are inter-ganglionic parts of inter-neurones, which leaves nerve 5 as the only source of spontaneously active motor axons. The possibility of intemeurones in the implant ganglion establishing connexion with the muscles cannot, however, be discounted. In Fig. 2E the lack of correspondence between the discharge pattern of the larger elements in the connective and in nerve 5 can be seen.

Table 1.

Spontaneously active nerve fibres in the isolated mesothoracic ganglion

Spontaneously active nerve fibres in the isolated mesothoracic ganglion
Spontaneously active nerve fibres in the isolated mesothoracic ganglion

The development of activity in implanted ganglia

During the weeks following implantation, the tracheal supply of the ganglion is re-established, and large numbers of new axons grow out into the muscles or into the cut ends of pre-existing nerves. Ganglia implanted into the haemocoel near the root of the 5th nerve tend to develop very thick connective sheaths, and the nerve roots become so altered that it is almost impossible to identify them (Fig. 6 F). Histological reconstructions from sections stained in silver nitrate show that the major connexions made with the host ganglion and the motor periphery are through the connectives of the implant ganglion, but the fibres remain so fine that it is difficult to determine whether they are derived from motor neurones or inter-neurone cell bodies. Ganglia implanted into the coxal muscles also push out large numbers of new axons, but they conserve their original form much more closely. The total size of the ganglion increases much less, and its ability to innervate distant muscles such as the tarsal depressor (muscle 144) is more limited. Both types of implanted ganglion develop striking rhythms and profoundly modify the behaviour of the left mesothoracic leg. Their functional properties may be classed broadly under three headings: (1) the direct control of the muscles by the implant ganglion; (2) the afferent innervation of the implant ganglion; and (3) the relationship between the implant and the host ganglion.

The direct control of the muscles by the implant ganglion

The developing implant ganglion appears to innervate any muscle fibres that lie in the vicinity. Jacklet & Cohen (1967) came to the conclusion that only muscles which had been previously denervated by section of the host nerves were innervated by the graft ganglion. In coxal implantations where nerve 5 had been left intact both depressors and levators of the trochanter were innervated, but it must be admitted that no sections of this area were examined to confirm that nerve 5 had been undamaged. This innervation of antagonistic muscles held equally for the femoral muscles. Attempts were made to identify the nerve roots of the implant ganglion, and to see whether their muscle specificity was maintained, by examining serial sections of the graft ganglion, but the nerve roots were often much expanded. Growth from the connectives was often noticeably weak in coxal implants, but nerve 2 appeared to innervate a bundle of trochanteral depressor fibres in one implant, and this nerve contains interneurones. A large proportion of the fibres in any connective will derive from cell bodies in distant ganglia so that growth from the connectives of an excised ganglion will always appear rather weak as compared with segmental nerve growth.

The development of autonomous activity in the graft area passes through three phases. An initial phase when very little activity can be detected; this may last for 3 or 4 weeks, and is followed by the succeeding period when steady discharges can be picked up in the nerve strands, but these have little effect on the muscles. An example of a steady-level discharge in a young implant ganglion is shown in Fig. 3B. Two large elements appear to be active at an overall frequency of 10 impulses per second. It was often difficult to find nerve strands with so few active elements, although overall frequencies in other nerve strands seldom exceeded 50 impulses per second. In some instances the electrical activity of the implant ganglia did not develop further, and where nerve 5 had been cut at the time of the implantation host motor axons redeveloped and the donor ganglion became a conduit for them. The record illustrated in Fig. 3 A shows simultaneous recordings from either side of an implant ganglion at its proximal connexion to the root of nerve 5 and its distal connexion to the motor periphery. It is clear from this that the same active elements are present at both recording points, and in preparation of this kind removal of the host ganglion abolished electrical activity in the limb muscles.

Fig. 3.

A. Synchronous rhythms in proximal (upper trace), and distal (lower trace) connexions of an implant ganglion, believed to originate in host motor axons. B. Steady discharge in a 3‐month implanted ganglion. Teased out preparation. C. Responses in the distal (upper trace) and proximal (lower trace) connexions of an implanted ganglion following electrical stimulation of the ipsilateral posterior connective. Note the feedback discharge from the implant ganglion in lower trace. Horizontal scale 50 msec. D. Delayed response in the motor connexions of the implant ganglion (upper trace) not visible in the proximal connexion with the host ganglion (second trace) even at a higher amplification. Antidromic single shock applied to the tibial depressor muscle. All time traces 10 cyc./sec., except for B which is 20 cyc./sec.

Fig. 3.

A. Synchronous rhythms in proximal (upper trace), and distal (lower trace) connexions of an implant ganglion, believed to originate in host motor axons. B. Steady discharge in a 3‐month implanted ganglion. Teased out preparation. C. Responses in the distal (upper trace) and proximal (lower trace) connexions of an implanted ganglion following electrical stimulation of the ipsilateral posterior connective. Note the feedback discharge from the implant ganglion in lower trace. Horizontal scale 50 msec. D. Delayed response in the motor connexions of the implant ganglion (upper trace) not visible in the proximal connexion with the host ganglion (second trace) even at a higher amplification. Antidromic single shock applied to the tibial depressor muscle. All time traces 10 cyc./sec., except for B which is 20 cyc./sec.

Where the implant was allowed to grow for periods up to 6 or 7 months, strikingly rhythmic activity could be detected in the muscles of the femur and the coxa in the absence of the host ganglion. Examples of these rhythms can be seen in Figs. 4 A and 5 A although neither of these demonstrates the regularity of some of the rhythms. Variation in the time between bursts might be as little as 10 % in 20 bursts. In the coxal record (Fig. 4A) a 5 sec. or 6 sec. rhythm consisting of fast and slow muscle potentials produces a twitch-like contraction, while in record 5 A there are similar bursts, and marked slow potential rhythms between 1·0 and 0·5 per sec. in addition. Another type of discharge rather similar to the one just described sometimes occurs, in which only the small tonic potentials are observed, and a rather irregular rhythm of this type can be seen in Fig. 4B. Here changes in tension oscillate rhythmically over periods of 5 or 6 sec., but their relationship to the more frequent bursts of muscle potentials is not clear. Fig. 4C shows the effect on this preparation of gradually releasing a small quantity of nicotine on to the implant ganglion. The frequency of muscle potential bursts increases, and long bursts disappear. The regularity of the discharge is also increased. The frequency of bursting is about once a second and muscle tension varies little although some very small contractions can be observed. In the record from muscles in the femur shown in Fig. 5 A the bursts of small potentials in the depressor (upper trace) coincide with drops in tension, and therefore must be accounted inhibitory. If the larger bursts are examined carefully, however, it can be seen that two bursts of small potentials may follow a burst of large potentials, the first associated with a rise in tension and the second with a decrease ; this is most clear in the first of the major bursts in this figure. In this record the activity of the tibial levator muscle can be compared with events in its antagonist. Initially the pattern of activity is very similar, but after the preparation has been set up for a few minutes the frequency of bursting falls from about one every 2 sec., to one every 7 or 8 sec., and the pattern in the levator matches that of the depressor less precisely. This is the stage reached in the preparation from which Fig. 5 A was obtained. The faster rhythm characteristic of a high state of excitation can be seen at the end of Fig. 5 C.

These observations suggest that the graft ganglion innervates neighbouring muscles in an unselective manner and its motor outgrowth contains fast, slow and inhibitory axons.These are driven by at least two repetitive systems which tend to run down under no-stimulus conditions when the graft is separated from the host nervous system.

This development of an unpatterned discharge, and later, rhythmic activity and delayed responsiveness, finds a striking parallel in the observations of Crain (1966) on expiants from the vertebrate spinal cord and brain.

Fig. 4.

A–C. Spontaneous mechanical (upper trace), and electrical (second trace) activity in the trochanteral depressor muscle, innervated by an implanted ganglion. In C a small quantity of nicotine has been topically applied to the ganglion. D. The response of the posterior connective of the host ganglion (upper trace), the right trochanteral depressor muscle (second trace), and the left trochanteral depressor muscle innervated by an implanted ganglion (third trace) to mechanical stimulation of an abdominal cercus (solid rectangle on time marker). All time markers show i cyc./sec.

Fig. 4.

A–C. Spontaneous mechanical (upper trace), and electrical (second trace) activity in the trochanteral depressor muscle, innervated by an implanted ganglion. In C a small quantity of nicotine has been topically applied to the ganglion. D. The response of the posterior connective of the host ganglion (upper trace), the right trochanteral depressor muscle (second trace), and the left trochanteral depressor muscle innervated by an implanted ganglion (third trace) to mechanical stimulation of an abdominal cercus (solid rectangle on time marker). All time markers show i cyc./sec.

Fig. 5.

A. Mechanical and electrical recordings from a tibial depressor muscle innervated by an implanted ganglion (two upper traces). One second time marker (third trace). The fourth trace indicates the electrical activity in the tibial levator of the same side. Synchrony lessens with decline in burst frequency. B. The mechanical and electrical response of a trochanteral depressor muscle, innervated by an implanted ganglion, to pressure on the trochanteral hair plate (solid rectangle on time marker). C. The effect of electrical stimulation (upper trace) of the periphery of an implanted ganglion on the electrical activity of adjacent muscle fibres in the coxa (second trace). Stimulus frequency 20 cyc./sec. All time markers 1 cyc./sec.

Fig. 5.

A. Mechanical and electrical recordings from a tibial depressor muscle innervated by an implanted ganglion (two upper traces). One second time marker (third trace). The fourth trace indicates the electrical activity in the tibial levator of the same side. Synchrony lessens with decline in burst frequency. B. The mechanical and electrical response of a trochanteral depressor muscle, innervated by an implanted ganglion, to pressure on the trochanteral hair plate (solid rectangle on time marker). C. The effect of electrical stimulation (upper trace) of the periphery of an implanted ganglion on the electrical activity of adjacent muscle fibres in the coxa (second trace). Stimulus frequency 20 cyc./sec. All time markers 1 cyc./sec.

Recordings made from muscle very close to the implant ganglion with metal microelectrodes often allowed the activity of what were believed to be single muscle units to be observed. Fig. 6C shows a preparation of this kind. The fibre is rhythmically active every 10 or 12 sec., but the activity which follows the leading pulse varies from a train of 15 or 20 potentials to none at all; like a pre- and postsynaptic relationship. Incidentally, this particular element was very close to a connective of the implant ganglion. Within the ganglion itself it was possible to record from axons or axon branches within the area shown in Fig. 6E. Recordings from this region showed unit spikes with changing frequency and grouping but without any very sharp correlation with the activity of muscular units. In Fig. 6A, two large elements and a much smaller one can be distinguished. One of the larger units produces fairly regular spike trains at a frequency of 6 impulses per second, of about the same duration as the discharges in the adjacent depressor muscle illustrated in Fig. 6C. Smaller units may produce brief bursts of a more regular nature (2 per second), as shown in Fig. 6B.

Fig. 6.

A. Spontaneous activity in the lateral neuropilar region (shown in E) of an implanted ganglion. Irregular large-unit activity during regular motor activity. B. Spontaneous activity in the lateral neuropile (as in A). Regular small-unit bursts, during a lower frequency of motor bursts. C. Rhythmic activity in a muscle unit adjacent to the anterior connective of a ganglion implanted into the coxal depressor. Two components are visible. D. Spontaneous electrical activity at the two recording points shown in the m vitro preparation in F. Upper trace, proximal recording point; lower trace, distal recording point. E. Ventral view of the meso-thoracic ganglion to show the recording area used for microelectrode penetrations. Tracheal branches are shown. F. An excised preparation of host and donor ganglion in vitro (see D). The arrows indicate recording points.

Fig. 6.

A. Spontaneous activity in the lateral neuropilar region (shown in E) of an implanted ganglion. Irregular large-unit activity during regular motor activity. B. Spontaneous activity in the lateral neuropile (as in A). Regular small-unit bursts, during a lower frequency of motor bursts. C. Rhythmic activity in a muscle unit adjacent to the anterior connective of a ganglion implanted into the coxal depressor. Two components are visible. D. Spontaneous electrical activity at the two recording points shown in the m vitro preparation in F. Upper trace, proximal recording point; lower trace, distal recording point. E. Ventral view of the meso-thoracic ganglion to show the recording area used for microelectrode penetrations. Tracheal branches are shown. F. An excised preparation of host and donor ganglion in vitro (see D). The arrows indicate recording points.

Some comparison can be made with the repeater units isolated by Fielden (1963) from dragonfly connectives, but their activity appears to have been altogether more regular. An attempt was made to compare the amplitude spectrum of a recording point situated in the neuropile with that of one situated in adjacent muscles, as shown in Fig. 7. Each curve represents the impulses in successive 10 sec. samples counted at different amplitude levels, the lowest curve representing signals of greatest amplitude. As can be seen there is some degree of correspondence especially in the first part of each curve (counts 1-5). In addition, it is interesting to observe that the separation of the upper curves in A and B (large and small elements) from the lower curves (large elements alone) follows a similar pattern between counts 1 and 7. Thus between counts 1 and 2 all amplitude categories tend to decline in frequency, but a small unit peak occurs at count 3 when the frequency of large units is declining still further. Why the counts made at the highest gain in A are so high it is difficult to say. The counts shown in Fig. 7 indicate that parts of the neuropilar pattern of activity may match adjacent output, but suggest that the relationship is not a direct and simple one. This is confirmed by the neuropilar records shown in Figs. 6A and 6B.

Fig. 7.

Amplitude spectrum for: A—a recording point on a bundle of muscle fibres; and B—in the lateral neuropile of an adjacent implanted ganglion. Each curve shows successive 10 sec. counts of potentials at different amplitude levels, to demonstrate the correlation between changes in the activity of different pulse classes. The lower lines represent potentials of greater amplitude.

Fig. 7.

Amplitude spectrum for: A—a recording point on a bundle of muscle fibres; and B—in the lateral neuropile of an adjacent implanted ganglion. Each curve shows successive 10 sec. counts of potentials at different amplitude levels, to demonstrate the correlation between changes in the activity of different pulse classes. The lower lines represent potentials of greater amplitude.

These observations may appear inconclusive but they suggest that the parts of the neuronal pattern assembled into motor output do not necessarily reflect the same time course, and that activity in the neuropile is more complex and more intense than might be expected purely on the basis of motor events.

The afferent innervation of the implant ganglion

The sensory component in the graft ganglion varies considerably from one insect to another, but in the long-term implants there were usually some well-defined pathways from the sense organs, which could be demonstrated by the responses of their reflex connexions.

Stimulation of the graft ganglion by forced levation or depression of more distal limb segments designed to stimulate campaniform organs was seldom successful. The tactile spines of the femur and the tibia on the other hand often developed more effective connexions and remarkable augmenting series of muscular responses could be obtained for as many as ten successive movements of the spine. This has been described in an earlier communication (Guthrie, 1966). Similar results could sometimes be obtained by electrical stimulation of the crushed tarsus, but as with mechanical stimulation it was often very difficult to elicit a response after the first or second trial.

This ability of the graft ganglion to amplify a response could be observed in a rather different form following pressure on the trochanteral hair plate. Fig. 5B illustrates activity in a rhythmically active preparation of the coxal muscle. Pressure on the hair plate has no effect on the following burst of potentials nearly 3 sec. later, or on the rising phase of the succeeding one, 512 sec. after this, but its relaxation phase and that of four succeeding contractions is greatly reduced. This effect precedes any change in the frequency of the bursts by 5 sec. A change in frequency does occur at the end of the response, the interval between bursts being halved once, and then again. The electrical record gives very little information as to the origin of the contraction plateau. The slowness and the graded nature of the response, which reaches a maximum 14 sec. after the stimulus, is perhaps a reflexion of a lack of central inhibition, but must also involve some kind of short-term storage even if this is in terms of active pathways.

The relationship between the host and the implant ganglion

The nature of the connexions between the donor and the host ganglion was investigated by a number of different methods, the simplest of which entailed separating the implant from the host ganglion while recording from a muscle largely innervated by the implant. Another approach was to apply stimuli to the periphery and note their effectiveness in producing a response in connexions of the implant, as compared with the response on the contralateral, unoperated side.

In most preparations, mechanical stimulation of the cerci produced an immediate response in the contralateral tibial depressor innervated by host axons, but no synchronous effect in the ipsilateral tibial depressor innervated by the implanted ganglion. Figure 4D illustrates an experiment of this kind. Following stimulation (sohd rectangle on time trace) a burst appears in the ipsilateral posterior connective of the host ganglion (upper trace), and a well-marked burst of potentials in the unoperated muscle (second trace). The operated muscle is rather inactive, although there has been a burst 212 sec. before the stimulus; but for a second after the stimulus there is no response, then three single potentials occur, 0·5–1·osee, apart. Since a train of impulses like this does not appear in the spontaneous record this must be regarded as a simplified, delayed and prolonged response to cereal stimulation.

Electrical stimulation of the ipsilateral connective coupled with recording from the host-donor connexion and from the distal outflow of the implant provided further results. This is quite awkward to perform as the central nervous systems of the graft and host have to be dissected out free of the muscles and then worked on in vitro. Syn-chronous potentials can be observed in the two traces (Fig. 3C), but there is an elaborate secondary response on the donor side which does not appear in the implant periphery. An initial synchronous phase, followed by a non-synchronous phase appears in both channels. Perhaps the secondary burst in the host connexion is a feedback from the donor to the host ganglion. The difference between a primary and a secondary response is also illustrated by the effects of antidromic stimulation shown in Fig. 3D. Stimulation of the tibial depressor muscle with a single shock causes a direct response in both channels, and a reflex burst in the motor outflow of the implant ganglion only (upper trace).

That there may even be a spontaneously active pathway from the donor to the host ganglion is indicated by the recording in Fig. 6 D taken from an in vitro preparation like that shown in Fig. 6F. A low level discharge can be observed in all parts of the 5th nerve ramus, but a series of bursts occurs at the rate of one each second in the more anterior host-donor ganglion connexion. These bursts do not resemble normal activity in the 5th nerve.

If the numbers of muscle potentials occurring in the operated tibial depressor in successive 10 sec. periods are counted at different amplitude levels over several minutes, the activity can be seen to settle down eventually to a low level, as indicated in the first part of Fig. 8. Cutting away the host ganglion provokes phases of great activity, which fall again noticeably after 1 min., but still remain higher than in the intact animal. Rather less clear evidence of the inhibitory role of the host ganglion is furnished by impulse counts from a coxal implantation. Here the comparison is between the unoperated coxal depressor and that innervated by the implant (lower graphs in Fig. 9). Isolation of the segment produced great activity on the normal side, but only a slight increase in the frequency of potentials in operated muscles. This increase remains when the host ganglion is removed, while the numbers of potentials on the normal side drop rapidly to a very low-level. Fifteen minutes later, however, a few spontaneous bursts appear in the denervated muscles and the level of activity on the implant side has fallen off considerably. It must be pointed out that this implantation was not one of the most active, but it does serve to illustrate the autonomy of the implant ganglion.

Fig. 8.

Successive 10 sec. counts of muscle potentials at different voltage levels in a tibial depressor muscle innervated by an implanted ganglion. A intact, B after severance of the implant ganglion from the host ganglion.

Fig. 8.

Successive 10 sec. counts of muscle potentials at different voltage levels in a tibial depressor muscle innervated by an implanted ganglion. A intact, B after severance of the implant ganglion from the host ganglion.

Fig. 9.

Successive 10 sec. counts of muscle potentials in: A, a normal coxal muscle; and B, the ipsilateral coxal muscle innervated by an implanted ganglion, following various treatments, i, intact preparation; 2, isolation of the segment; 3, immediately after removal of the host ganglion, 4–5 min. later.

Fig. 9.

Successive 10 sec. counts of muscle potentials in: A, a normal coxal muscle; and B, the ipsilateral coxal muscle innervated by an implanted ganglion, following various treatments, i, intact preparation; 2, isolation of the segment; 3, immediately after removal of the host ganglion, 4–5 min. later.

Direct stimulation of the implant ganglion was attempted in order to modify the rhythmic output. The electrodes used for stimulation were placed at the periphery of the ganglion and a variety of stimulus frequencies and intensities employed. Single shocks even at high intensities had little effect but a large, although delayed, response often occurred following a half-second burst of current at 20 pulses per second. Figure 5C shows a response of this kind. Bursts had occurred previous to stimulation once every 5–7 sec., and the one immediately following stimulation is neither advanced nor delayed, but the succeeding three bursts occur in a much more closely spaced series. This is followed by a continuous barrage of muscle potentials which lasts for nearly half a minute, tailing off into a fast but normal rhythm. The delay before the burst frequency augments is 5 sec., and before the appearance of the barrage the delay is about 10 sec. The possibility of altering the rhythmic output by applying weaker electrical stimulation at more frequent intervals suggested itself. The original intention was to see whether shocks falling immediately before or immediately after the spontaneous burst from the implant could slow or accelerate the rhythm. No such effect could be produced and this type of experiment was abandoned. Looking at the results of these experiments again, there does seem to be a more general inhibitory effect where the stimulus intensity had been altered, and one example of a relationship of this kind is shown in Fig. 10. During the first period of stimulation an increase in the interval between bursts appeared before stimulation was begun, and the slow phase was hardly arrested by raising the stimulus intensity. The second phase of stimulation, however, does seem to coincide much more with a slowing of the rhythm which recovers subsequently, although anomalous details remain.

There seems little doubt that inhibitory pathways do exist from the host to the donor ganglion, but the strength of their influence varies a great deal.

Fig. 10.

The intervals between bursts of muscle potentials in a coxal muscle innervated by an implanted ganglion during intermittent electrical stimulation at 1 cycle per second, plotted against time. The lower profile is proportional to the magnitude of the applied voltage.

Fig. 10.

The intervals between bursts of muscle potentials in a coxal muscle innervated by an implanted ganglion during intermittent electrical stimulation at 1 cycle per second, plotted against time. The lower profile is proportional to the magnitude of the applied voltage.

In the normal animal repetitive limb cycling may occur at rates up to 1o eye./sec., although Wilson (1965) showed that the proprioceptive system was capable of following sinusoidal input up to 25 cyc./sec., the muscular and skeletal system limiting the rate of movement. Clearly these high rates of repetitive activity are coincident with exceptionally high levels of sensory activity both in the pinned-out preparation and the free-walking animal. Under conditions where the behaviour of the preparation is more spontaneous, lower rates of rhythmic activity may be expected. Isolating the thoracic ganglia, a segmental ganglion, or even the motor centre of one side does seem to have some effect in allowing the appearance of rhythmic activity, but this tends to die away unless the preparation is stimulated. Bursts within the frequency range produced by the implant ganglion, that is to say 7 to 30 bursts per minute, can be observed in the later stages of experiments involving cutting away the central nervous system ; but in the normal insect removal of other parts of the central nervous system deprives the local ganglion of both inhibitory and excitatory pathways simultaneously so that the level of excitability associated with the appearance of bursts only occurs transiently. The development of rhythmic bursts in the implant ganglion appears to coincide with the development of afferent connexions and it may be that the growth of these in the presence of slight or no central inhibition permits a higher degree of spontaneous rhythmicity than normal.

The responses of the implant ganglion often exhibit a remarkable latency coupled with a tendency to augment, as if there is some temporary storage of the stimulus effect, or very gradual transmission through a fine fibre network. Silver preparations do indicate the absence of large fibres within the ganglion. The augmentation may partly reflect a low level of inhibition, but is difficult to relate to the numerous examples of facilitation as in this instance it occurs after a single stimulus; perhaps the spontaneously active neurones may provide the additional active pathway themselves.

It was not possible to obtain any clear indication that the implant ganglion had any significant effect on the activity of the host central nervous system, although this was looked for on numerous occasions. Some recordings suggested that active fibres from the donor ganglion did enter the host ganglion, but the level of activity of the host is undoubtedly dominated by head and tail ganglia, and involves such large gradients of activity that a small effect of this kind is very difficult to observe by non-statistical methods.

Recording within the ganglion indicates that rhythms are more likely to be assembled from many units active at a variety of frequencies than from one or a few dominant cells discharging at the same frequency as the motor axons.

There seems no doubt that the host ganglion exercises a definite effect on the periphery even where the donor ganglia play the dominant role. Removal of the host ganglion removes its own motor axon contacts with the muscles, as well as its connexions with the implant ganglion, and it is difficult to separate the effects of these two operations; but it seems unlikely that there would be a large inhibitory component in the muscles from the host ganglion, when excitatory effects from this source were negligible. Therefore the great increase in activity following removal of the host ganglion visible in Fig. 8 can reasonably be referred to the removal of central inhibition. Excitatory pathways probably exist as well, but from the point of view of adaptiveness it can be assumed that an inhibitory effect is more likely to dominate.

Although the muscular twitches produced by the graft ganglion are weak (seldom more than 2 g.) they are basically non-adaptive. They may reduce the muscular atrophy which follows denervation, but since later recovery of control over the motor periphery by the host ganglion seems never to occur, this does not appear to entail any practical advantage. Some use of the limb as a strut in walking can be made by the action of the extrinsic coxal muscles, which remain innervated by the host ganglion, and this will be rendered more effective if the tibia and tarsus do not make uncoordinated movements. Are the elements concerned peripheral or central inhibitory cells? The course of events shown in Fig. 9 suggests the latter, as no increase in activity follows removal of the host ganglion as distinct from separation of the segment. Branches of inhibitory interneurones, say from the head ganglia, might be involved. Efficient and flexible control of the limb muscles in walking depends to a major extent on the activity of ganglia outside the mesothoracic ganglion, and even local reflexes depend on the sequential activity of different motor neurones connected in a prescribed manner to specific muscles. The motor neurones from the implant ganglion may be connected to the correct muscles, but the control of their sequence of function has become dominated by oscillatory activity incompletely reduced by the host central nervous system. The graft ganglion in some respects resembles a parasitic organism which successfully dominates the events within a local region of its host by vigorous growth and well defined electrical activity.

We would like to thank Mr L. Panko for his assistance in the early stages of this work, and the Science Research Council for their continuing financial support. A number of valuable criticisms and suggestions were made by colleagues, and by Professor G. M. Hughes, to whom we should like to record our indebtedness.

Becht
,
G.
,
Hoyle
,
G.
&
Usherwood
,
P. N. R.
(
1960
).
Neuromuscular transmission in the coxal muscles of the cockroach
.
J. Insect Physiol
.
4
,
191
.
Bodenstein
,
D.
(
1957
).
Studies on nerve regeneration in Periplaneta americana
.
J. exp. Zool
.
136
,
89
116
.
Crain
,
S.
(
1966
).
The development of organotypic bioelectric activities in central nervous tissues during maturation in culture
.
Int. rev. Neurobiol
.
9
,
1
43
.
Fielden
,
A.
(
1963
).
Properties of intemeurones in the abdominal nerve cord of a dragonfly nymph
.
J. exp. Biol
.
40
,
541
52
.
Guthrie
,
D. M.
(
1966
).
Physiological competition between host and implanted ganglia in an insect
.
Nature, Lond
.
210
,
5033
,
312
13
.
Guthrie
,
D. M.
(
1967
).
The regeneration of motor axons in an insect
.
J. Insect Physiol
.
13
,
1593
611
.
Jacklet
,
J.
&
Cohen
,
M.
(
1967
).
Synaptic connections between a transplanted insect ganglion and muscles of the host
.
Science, N.Y
.
156
,
1638
40
.
Rowell
,
C. H. F.
(
1964
).
Central control of an insect segmental reflex. I. Inhibition by different parts of the central nervous system
.
J. exp. Biol
.
41
,
559
72
.
Wilson
,
D. M.
(
1965
).
The nervous co-ordination of insect locomotion
.
In The Physiology of the Insect Central Nervous System
, pp.
125
40
. (
J. E.
Treherne
and
J. W. L.
Beament
, eds.).
London and New York
:
Academic Press
.