Central Control of An Insect Segmental Reflex: II. Analysis of the Inhibitory Input from the Metathoracic Ganglion

A previous paper (Rowell, 1961) described a reflex grooming movement of the front leg of the locust Schistocerca gregaria and other grasshoppers, and showed that under normal conditions it is almost completely inhibited, but that its responsiveness approaches 100% when the prothoracic ganglion is isolated from the rest of the central nervous system.

The origin of the inhibitory input from the C.N.S. to the prothoracic ganglion was analysed in the first paper of the present series (Rowell, 1964), and it was shown that progressive lesions to the C.N.S. result in progressive disinhibition of the prothoracic reflex. The main sources of inhibition are the metathoracic, suboesophageal and mesothoracic ganglia, in that order. The inhibitory output of these ganglia appears to be proportional to their level of activity; thus it diminishes if a ganglion is de-afferented, and increases if it is stimulated electrically.

In the present paper an attempt is made to correlate the change in responsiveness of the prothoracic grooming reflex after a specific lesion to the C.N.S. with the change in input to the prothoracic ganglion, recording electrically. The lesion was chosen because of its simplicity and marked effect on reflex responsiveness. It consisted of cutting the meta/mesothoracic connectives in a preparation consisting of only the intact thoracic nervous system, with abdomen and head disconnected. This lesion raises the average responsiveness from 13% before to 84% after (Rowell, 1964).

The input to the prothoracic ganglion in these experiments is recorded from the meso/prothoracic connectives. Each of these contains nearly 4000 axons, of which only some 3% are greater than 5 μ in diameter (Rowell & Dorey, 1967), and only a very small sample of this total will be recorded with external electrodes. The inhibitory input to the grooming reflex may be confined to one or a few axons; in this case it is unlikely to be recorded, especially as there is no obvious reason why such axons should be large in size. If, on the other hand, the inhibition is a function of the total input to the ganglion, rather than of a specific neuron (see discussion in Rowell, 1965) then even a small sample of the total activity of the connective might be expected to show changes which correlate with the observed behavioural changes. On this hypo-thesis, one would expect the input to the ganglion to decrease as a result of the lesion, and the output from the ganglion perhaps to rise, if the disinhibition is non-specific not only in source but also in target system. In order to increase the size of the sample of axons recorded, recordings have been made from different points around the circumference of the connective. The selection of axons recorded with an external hook electrode depends on both their size and proximity to the recording site, and identifiably different units can be recorded at different points on the connective circumference.

In a normal recording made from an intact connective, there is no indication of the direction in which the propagated action potentials (P.A.P.S) are propagating, i.e. what fraction of the activity is being emitted by the prothoracic ganglion and what received by it. Cutting the connective resolves this difficulty, but is likely to result in highly abnormal working of both ganglia, which will be deprived of their normal input. To avoid this dilemma use has been made of a device which distinguishes the direction of propagation of P.A.P.S in an intact connective; with this the traffic in the connective has been analysed according to the frequency, direction of propagation and recorded amplitude of the P.A.P.S.

Mature adult desert locusts (Schistocerca gregaria Forskål) from a crowded laboratory culture were used. The legs were cut off and the animals were pinned ventral surface up. The cervical and abdominal/metathoracic connectives were cut, the meso/ prothoracic connectives were exposed for recording, and the extraneural sheath (Boulton & Rowell, 1968) was removed. The following experimental routine was followed with each animal.

Table 5 compares the UP/DOWN ratio for the five animals before and after the lesion. All five animals show a decrease, averaging 65 % and varying from 27 to 79%. The difference is significant (P = 0.03, Wilcoxon matched pairs test, Walsh test (Siegel, 1956)). This decrease agrees with the figures derived in A and B above; a decrease of 62% in the UP signal and an increase of 19% in the DOWN signal would change the ratio by 76%, which is close to the observed value.

Experiment 1. Activity in the left - hand meso/prothoracic connective was recorded for 1 min.

Experiment 2. The meta/mesothoracic connectives were cut and the animal was allowed 5–10 min. to recover. A further recording was then made from the left-hand meso/prothoracic connectives as in Expt. 1.

Experiment 2A. Immediately after Expt. 2 the electrodes were transferred to the right-hand meso/prothoracic connective and a further recording made. This experiment controls for accidental damage to the left-hand connective caused during Expt. 1.

The lesions and recordings are summarized in Fig. 1.

In all experiments two silver wire hook electrodes were used placed approximately 3 mm apart on the connective, each connected to a separate matched pre-amplifier. The pre-amplifier outputs were recorded on separate channels of an FM tape recorder; a third channel recorded a spoken commentary.

A 30 sec. length of tape was selected from the middle of each experimental record, and replayed into an analyser circuit, the design and working of which will be fully described elsewhere (Rowell, in the Press). The principle of this device is to use the arrival of a P.A.P. at one electrode to open, after a short delay, a gate in the recording channel connected to the second electrode ; the latter channel then transmits only P.A.P.S which are propagating from the first electrode to the second. The output from the gated electrode is fed to a 1 Mc/s electronic counter, and after mathematical corrections to compensate for random coincidence of open gates and reverse direction P.A.P.S, the input activity is resolved into three fractions ; (i) P.A.P.S propagating in one direction, (ii) P.A.P.S propagating in the reverse direction, and (iii) all potentials recorded at one electrode, but not recorded, or recorded simultaneously, at the other electrode. This last category, which is discarded, includes interference such as switching transients or muscle potentials.

The experiments were performed on each of five animals. As the results from all five showed the same trends, no further replicates were made.

Table 1 shows the percentage of the total recorded activity which was assigned by the analyser circuit to either the UP (mesothoracic to prothoracic ganglion) or the DOWN (prothoracic to mesothoracic ganglion) categories. The averages of Expts. 1, 2 and 2A are 65, 67 and 65% respectively, with extreme values of 51 and 79% for individual animals. (When only large-amplitude P.A.P.S are analysed, much higher percentages are obtained). The percentage of total activity accepted by the analyser is an index primarily of the comparability of recording conditions at the two electrodes (see Rowell, 1968) and these figures show that conditions were uniform throughout the experimental series.

Photographs of superimposed oscilloscope traces triggering on the P.A.P.S usually show 4–8 clearly separable large spikes and an indeterminate number of smaller ones. The highest frequency of P.A.P.S retrieved by the analyser circuit from the recording is less than 200/s under the experimental conditions. If a very rough estimate of average activity of an axon of 5–10/sec. is accepted, then probably not more than 20–40, and almost certainly less than 100 axons contribute most of the recorded activity. 40 axons is 1% of the total number in the connective, and this is about the proportion expected on dimensional arguments (see Introduction).

Table 2 shows the frequency of P.A.P.S of five different amplitude classes in the UP and DOWN fractions of the traffic in the connective for each of the five animals and in three experiments. The means for all five animals are shown as histograms in Fig. 2. These results are analysed below.

Table 3 shows the frequency of P.A.P.S propagating in this direction before and after the lesion. In all five individuals there is a decrease in activity averaging 62·2%, ranging from 44 to 79%. If the activity is meaned for all five animals, and grouped in the five amplitude categories, the difference between before and after the lesion is significant (2 ×5 χ2 > 100, P < 0·001).

Table 4 shows the frequency of P.A.P.S propagating in this direction before and after the lesion. The average of all five animals was an increase of 19 ˙1 %. In one individual there was no change ; in one other there was a small decrease ; in the remaining three there was an increase ranging from 15 to 67%. When activity is meaned for all five animals, and grouped in the five amplitude categories, the difference between before and after the lesion is significant (2 × 5χ 2 > 40, P < o.oo1).

A and B above compare absolute frequencies of P.A.P.S recorded on different occasions. As the amount of residual blood, etc., at the recording electrodes can make large differences to recordings, this comparison is likely to give no more than a rough indication of the effects of the lesion. A much better indication is given by comparing the ratio of frequency of UP P.A.P.S to frequency of DOWN P.A.P.S. This ratio is more or less independent of variation in recording conditions, which affect both fractions equally.

Table 5 compares the UP/DOWN ratio for the five animals before and after the lesion. All five animals show a decrease, averaging 65 % and varying from 27 to 79%. The difference is significant (P = 0·03, Wilcoxon matched pairs test, Walsh test (Siegel, 1956)). This decrease agrees with the figures derived in A and B above; a decrease of 62% in the UP signal and an increase of 19% in the DOWN signal would change the ratio by 76%, which is close to the observed value.

This control procedure is only partially adequate, as it is designed to provide a new, previously untouched connective. By definition there can be no recording made before the lesion with which to compare it, and comparison must be with the count made from the left-hand connective. If these two connectives differ in their overall level of activity, either inherently or because of differences in recording conditions, then the comparison is invalid. It is known that the two halves of the prothoracic ganglion differ markedly between individuals in their reflex responsiveness (Rowell, 1964), and simultaneous recordings from the left and right connectives often seem very different. Similarly, McKay (1968) has shown that paired neurons in the cervical connectives of tettigoniid grasshoppers often show marked asymmetries, differing from one individual to another. Comparison of the results of Expt. 2A and Expt. 1 can therefore be expected to show a much larger scatter and variation than the comparison of the results of Expt. 2 and Expt. 1, carried out on the same connective.

Table 6 makes the comparison of UP/DOWN ratio between Expts. 1 and 2A, and shows the expected large variation. The average performance is a decrease, confirming the results presented above, but the decrease averages only 17% and is not significant. This is largely accounted for by an aberrant figure of 121 % increase from one animal ; three of the five animals show decreases of over 50%.

Table 7 shows that the total activity in the left-hand connective, obtained by summing both UP and DOWN fractions, fell by an average of 22 % in the five animals. This effect of the lesion is small, relative to the very large changes brought about in the distribution of activity between the ganglia.

In all nervous systems mechanisms are necessary to ensure the temporary dominance of the most strongly evoked response system. The simplest way of achieving this is by mutual inhibition between the different functional systems, and this is performed at several levels (see discussion in Rowell, 1964, 1965). In these previous papers it was further suggested that the demonstrated inhibitory regulation of reflex responsiveness in the prothoracic ganglion by the rest of the nervous system was a function of much or all of the input received at the ganglion, rather than of a specific inhibitory axon or group of axons.

As the small size of most of the axons in the connective precludes direct recording from them, the ‘specific inhibitory axon’ hypothesis cannot be tested directly. The experiments reported here have produced results predicted by the ‘non-specific inhibitory input’ hypothesis, though they have not finally established it. When the disinhibiting lesion is made, there is a decrease in the afferent flow of action potentials to the prothoracic ganglion in the small sample of axons available to the recording electrode; as these are defined by their large size it is very unlikely that they include a specific inhibitory axon to a minor reflex system. In addition, there is a concommitant increase in the efferent flow of action potentials out of the prothoracic ganglion. This is predicted by the hypothesis; if the disinhibition of the grooming reflex is brought about by a reduction in a non-specific inhibitory input, then one expects the activity of the prothoracic ganglion to rise in other ways too. The specific inhibitory axon hypothesis, on the contrary, does not predict this increase in activity in the prothoracic ganglion, and leaves the observed increase unaccounted for.

The term ‘non-specific’ inhibition is not meant to imply a random connexion of the neuropile, as speculated by Horridge (1961), in either the emitting or receiving ganglion, but that the inhibition acts on the general responsiveness or arousal state of the ganglion rather than on the functional mechanism of one specific reflex system. Such an effect could, for example, be produced by a bifurcation of input in the ganglion, one fraction having connexions with appropriate specific systems and the other influencing an arousal system for the ganglion, which in turn would have synaptic connexions with the specific integrative areas. This is a simple analogy of the reticular activating system of the vertebrate brain, and some such system may well be present in all integrative areas above a certain level of complexity.

Generalizing, one obtains a picture of the insect nervous system consisting of functionally discrete units connected by mutually inhibitory links, possibly mediated by non-specific arousal systems operating at the ganglionic level. Such units include whole ganglia, separate halves of the same ganglion (Weiant, 1958; Rowell, 1964), or different reflex systems within one or more ganglia. A system of this sort would be stable under constant input conditions, but would change to a new stable state following a change of input. The change would be rapid, because the stimulated unit would increasingly inhibit the others, and decrease their inhibitory effects on itself. This positive feed back would result in a ‘bistable’ action, tending to eliminate intermediate or oscillatory states which would otherwise occur when two different behaviour patterns were evoked at about the same intensity.

In the mammal brain, integration of sensory input is affected not only by the reticular system, but also by corticofugal systems which act on sensory interneurones in a highly specific manner (Gordon & Jukes, 1962; Dewson, 1967). Horn (1965) has pointed out that while both these systems can influence responsiveness or attention, they probably work in different and complementary ways. The corticofugal system appears to accentuate patterns of activity in the sensory pathways, whereas the reticular system seems likely to mask or depress them. The present work suggests a non-specific system in the insect nervous system with a role analagous to the reticular system: this does not of course affect the possibility that the functional equivalent of a corticofugal system may also be involved. An essential part of the mechanisms controlling attention and reflex responsiveness must be specific inhibitory or excitatory relationships in the nervous system, many examples of which are known. The nonspecific inhibition suggested by the experimental results presented here would merely provide a substrate of elementary integration, preventing the animal from attempting to perform incompatible activities simultaneously, and this substrate could then be modified by more specific and selective mechanisms.

The influence of the brain on the responsiveness of the insect nervous system is obscure. In the prothoracic grooming reflex of the locust the brain exerts no constant influence, but while it remains connected to the posterior nervous system the responsiveness of the reflex is quite unpredictable (Rowell, 1964). Weiant (1958) found that cutting the neck connectives increased efferent activity in metathoracic nerves of the cockroach, but this disinhibitory effect could, as in the grooming reflex of the locust, be due to the removal of the suboesophageal ganglion, rather than of the brain. The brain appears to influence both the level of response and the response decrement to repetitive stimuli in certain sensory interneurones in the locust thorax (unpublished work), and changes in behaviour brought about by stimulating the brain can be interpreted (Rowell, 1963) as due to changes in the responsiveness of local reflexes so caused. The extensive earlier work on the results of decapitation (reviewed by Roeder, 1958) was performed in ignorance of the hormonal repercussions of injury or stress (Milburn, Weiant & Roeder, 1960; Highnam, 1961), and is difficult to interpret. It is probable that most of the output from the brain, other than descending sensory intemeurones similar to those described in the crayfish by Wiersma & Mill (1965), consists of command fibres and direct excitatory or inhibitory connexions to specific systems, rather than to non-specific ganglion arousal systems.

The work reported here has been sponsored in part by the United States Government under Contract 61052/67/C/0016, and by Makerere College Research Fund, Grant No. 242.