ABSTRACT
Visual stimuli presented to a locust during performance of a defensive kick fail to elicit a response from the descending contralateral movement detector (DCMD) visual interneurone. Similar control stimuli presented before the initial flexion and at least 1 s after the movement elicit a normal DCMD response (10–15 spikes).
External stimuli used to elicit defensive kicking can sometimes cause suppression of the DCMD without the occurrence of kicking, but only when there is an increase in central nervous activity, or associated behavioural response.
Proprioceptive stimulation similar to that occurring in a kick does not cause suppression of the DCMD unless there is an increase in central nervous activity or associated behavioural response.
Extensive dorsal dissection abolishes the suppression of the DCMD.
INTRODUCTION
The locust defensive kick involves the very rapid extension of each metathoracic tibia about the femoral-tibial joint. This extension is the termination of a three stage process (Heitler & Burrows, 1977a). First is the initial flexion, or ‘cocking’ (Pearson & Robertson, 1981), which brings the tibia into the pre-movement flexed position. Second is the co-contraction, in which the extensor muscle slowly contracts, storing energy in the distortion of cuticle springs (Bennet-Clarke, 1975), while the smaller flexor muscle contracts simultaneously (working through a large mechanical advantage) to maintain the tibia in the flexed position (Heitler, 1974, 1977). Third is the trigger phase, in which the flexor muscle relaxes, allowing the tibia to extend.
When a locust jumps the tibiae are forcefully extended as for the kick, and the motor programme controlling the tibiae has the same three essential features (Pflüger & Burrows, 1978). The differences between the jump and the kick lie in the direction in which the tibiae are aimed, and the posture of the animal, not in the movements of the tibiae.
The trigger phase is essential for the successful completion of the jump or kick; if the flexor muscle does not relax, the tibia does not extend and the jump or kick is aborted. Intracellular recordings from flexor motoneurones have shown that the trigger activity during a kick is caused by simultaneous strong inhibition of flexor excitor motoneurones and excitation of flexor inhibitor motoneurones (Heitler & Burrows, 1977a). Such recordings cannot be made during a jump, but myograms indicate that flexor activity ceases just before the movement, suggesting a similar neural mechanism.
The ‘M’ interneurone makes powerful spike-dependent inhibitory synapses onto flexor excitor motoneurones (Pearson, Heitler & Steeves, 1980). It produces a burst of spikes just before tibial extension in a defensive kick (Steeves & Pearson, 1982), and it is thought to be at least in part responsible for the trigger activity. The interneurone has been termed the ‘M’ neurone, because it is excited by multimodal sensory input. It receives particularly strong excitatory input from an identified visual interneurone, the descending contralateral movement detector (DCMD). A characteristic of M is that it has a high spike threshold, so that in the relaxed animal excitatory sensory input rarely, if ever, induces spiking. During the co-contraction phase of a kick M is depolarized by proprioceptive input from mechanoreceptors monitoring joint distortion. It has been suggested that a sudden sensory stimulus delivered in the later stages of co-contraction, which would be subthreshold in the relaxed animal, may be suprathreshold if summed with this depolarization (Pearson et al. 1980). In this manner the DCMD could act to trigger a kick or jump if, and only if, its spikes were timed to occur during co-contraction. This appears to make behavioural sense in that some low-level stimulus might induce co-contraction leading to an abortive kick or jump, while sudden DCMD input (signalling a possible increase in danger) occurring during co-contraction would convert the abortive behaviour into a real kick or jump by inducing trigger activity.
It was to test this hypothesis that the initial experiments described below were performed. The hypothesis is shown to be unlikely, at least in its simplest form, for restrained defensive kicks. This is not because the DCMD fails to elicit trigger activity, but because DCMD activity itself cannot be elicited during the appropriate time window. The response of the DCMD is almost completely suppressed during co-contraction.
MATERIALS AND METHODS
Adult locusts Schistocerca americana (Forskal, 1775) were reared in crowded colonies. Myograms of electrical activity in the metathoracic extensor muscle were made using implanted copper wires. Recordings of DCMD activity were made from the ventral nerve cord connectives between the pro- and mesothoracic ganglia using either implanted myogram-type electrodes, or conventional hook electrodes. The latter technique involved a slightly more extensive dissection, but yielded a better signal-to-noise ratio. No difference in results using the two techniques was observed. Visual stimuli were presented to one eye of the locust (the right eye unless otherwise stated) using a black disk of 20 mm diameter mounted on the moving arm of a Pen Unit transducer (Southern Instruments Ltd.), positioned 200mm from the animal. The disk moved back and forth through an arc of 50 mm driven by a half-sinusoid waveform generator. The contralateral eye was occluded by a Plasticene blinker, are a audio monitor of neural activity used in setting up the recordings was switched off during the actual experiment. This was to remove known sources of DCMD inhibitory modulation (Rowell & O’Shea, 1980). Visual stimuli could be presented at a set time after a myogram spike using a microcomputer-based spike discriminator and delay timer. Mechanical stimulation was provided by a small paint brush mounted on an electromechanical transducer (Ling Dynamic Systems Ltd.) driven by a gated sinusoidal oscillator.
RESULTS
Suppression of the DCMD during co-contraction
The aim of the initial experiments was to determine whether a burst of DCMD spikes occurring during the co-contraction prior to a kick could elicit trigger activity earlier than would otherwise have occurred. To test this a restrained locust was induced to kick repeatedly while measuring the duration of co-contraction with implanted myogram electrodes. In alternate kicks a visual stimulus such as would produce a burst of DCMD spikes in the quiescent animal was presented 400 ms after the first extensor muscle impulse. In a sequence of 16 kicks produced by one locust no trend in change of duration was observed, and no significant difference was found between the duration of co-contraction without a visual stimulus (501 ms; S.D., 44; N, 8)-and the duration with a visual stimulus (483ms; S.D., 38; N, 8). Similar experiments varying the timing of the visual stimulus in 50 ms steps between 250 and 600 ms after the start of co-contraction failed to demonstrate an effect on the duration of co-contraction.
The failure of these results to support the original hypothesis made it necessary to determine the response of the DCMD to the visual stimuli in more detail. Locusts were therefore induced to kick while recordings were made from the DCMD, with visual presentations made before, during and after each kick. The visual stimuli within an experimental triplet were spaced at least 20 s apart, and triplets themselves were spaced at 2–3 min intervals. The DCMD responded to the 1st and 3rd stimuli with 5–15 spikes, depending on the preparation, but the response to the 2nd stimulus (concurrent with the kick) was strongly suppressed (Table 1). This suppression of the DCMD during kicking lasted for a period from the initial flexion, through on contraction, until at least 1 s after the performance of the kick (Fig. 1). In most cases the DCMD response was abolished completely, but when present, the number and frequency of spikes was considerably reduced compared to the response to control stimuli (Fig. 2). This was true not only of co-contractions which eventually led to a kick, but also in those cases where the kick was abortive. Furthermore, DCMD responsiveness was often (although not always) reduced following the pre-kick flexion, even if no co-contraction phase followed (Fig. 3). Suppression was induced by motor activity of either one of the metathoracic legs alone. In short, the response of the DCMD was suppressed during motor activity associated with all phases of defensive kicking.
The suppression of the DCMD response to the 2nd stimulus of the triplet might perhaps have been due to habituation caused by the 1st stimulus, with sensitization by the kick leading to recovery with the 3rd stimulus. However, triplets in which no kick occurred showed no consistent decrement in response, and certainly not complete suppression of response to the 2nd stimulus. Indeed, in one preparation, 10 visual presentations at 20 s intervals showed no decrement (1st response, 6 spike 10th response, 7 spikes; mean, 6·6; S.D., 0·9). The absence of perceptible habituation in these experiments is not surprising since the locust is frequently aroused and active, and under these conditions the DCMD maintains an irregular but high level of response with no regular increment or decrement (Rowell, 1971b).
Is the DCMD suppressed by stimuli applied to induce kicking?
Defensive kicks were induced by mechanical stimulation of the locust, usually by tickling. It is possible that the suppression of the DCMD response resulted from some form of cross-modal inhibition, rather than from the performance of the actual behaviour. Indeed, a loud noise or a strong mechanical stimulus applied just before the visual stimulus can suppress the response of DCMD in an otherwise quiescent animal (Fig. 4; Rowell & O’Shea, 1980). However, when such suppression occurs it is always accompanied by an increase in central nervous activity as monitored from the ventral nerve cord (vnc) compared to the activity of the vnc when the DCMD is not suppressed by the applied stimulus. Thus it appears that the suppression of the DCMD is associated with some form of central arousal, rather than the external stimuli themselves.
Three lines of evidence support this conclusion with regard to defensive kicks. First, occasional kicks are performed in the absence of any apparent external stimuli. On such occasions DCMD is suppressed just as in a kick induced by mechanical stimulation (e.g. Fig. IB). Second, the mechanical stimuli applied to induce kicking can be very gentle; frequently no more than a single light touch with a brush. If these stimuli produce behaviour associated with kicking (flexion or co-contraction) just before the visual stimulus, DCMD is suppressed; but if a behavioural response occurs after the stimulus, or does not occur at all, DCMD responds normally (Fig. 3). Because of the dangers of experimenter bias in applying mechanical stimuli by hand, attempt was made to automate the system using a hair brush mounted on a vibrator.
Ithough the automated mechanical stimulus only rarely induced a full kick, it frequently induced an abortive kick, or at least a pre-kick flexion (Fig. 5). In each case when the mechanical stimulus induced a behavioural response, the DCMD was suppressed, while the same mechanical stimulus in the absence of consequent motor output had no effect on the response of DCMD.
Do proprioceptors gate the DCMD?
The experiments described above indicate that it is not exteroceptive stimulation that causes suppression of the DCMD, but rather the behavioural response to that stimulation. However, there remains the possibility that the proprioceptors excited by the behavioural response may be the cause of suppression. In particular, mechanoreceptors monitoring joint distortion during co-contraction, which are known to be important in the neural control of defensive kicks (Bassler, 1968; Heitler & Burrows, 1977b), are candidates for such a role.
To test this possibility an attempt was made to mimic the mechanoreceptive effect of co-contraction by stimulating the extensor muscle electrically through implanted myogram leads. Initial experiments were performed with the tibia held at an angle of 90° by a metal rod, because this is the angle at which the extensor muscle has the best mechanical advantage. The results appear to confirm the role of proprioceptors in gating the DCMD. The DCMD response was completely suppressed for the duration of extensor muscle stimulation (after about the second stimulus pulse), and reduced for up to 1 s after stimulation ceased (Fig. 6A–C). However, it was noticed that the stimulation also caused a considerable increase in vnc activity, and hence presumably an increase in central arousal. This arousal, rather than the proprioceptor stimulation, might have been responsible for DCMD suppression. Furthermore, the experimental situation was not an accurate mimic of co-contraction, since it involved fairly massive stimulation of the tibial spines which were thrusting into the rod restraining the tibia. A more realistic mimic was achieved with the tibia fully flexed. The poor mechanical advantage of the extensor muscle in this situation reduced the torque, and hence the exteroceptive stimulation of the tibial spines, although the extensor muscle tension and the consequent joint distortion remained high. Frequently in this situation no restraint at all was required, since the excitation of the fast extensor tibia motoneurone could cause sufficient central excitation of the flexor tibia motoneurones (Hoyle & Burrows, 1973) to ensure the tibia remained flexed. Under these conditions there was little increase in vnc activity, and little if any suppression of the DCMD response (Fig. 6D). Thus the suppression of the DCMD response during defensive kicking is unlikely to be a result of stimulation of the proprioceptors. It is also unlikely to be a result of simple efference copy, since the antidromic stimulation of the motoneurones is similar in the two positions.
Abolition of suppression of DCMD response
Under conditions which were found in the above results to lead to suppression of the DCMD response, Steeves & Pearson (1982) have made recordings of DCMD activity. In their experiments, the recordings were made after dorsal dissection. However, this postural difference does not appear to be the cause of the different results, since ventrally dissected locusts, restrained in a dorsal-upwards posture, show exactly the same suppression of DCMD during defensive kicks, arousing sensory Stimuli and electrical stimulation with the tibia restrained at 90° as occurs in the inverted position (Fig. 7). Instead, the difference in results appears to be due to the more extensive nature of the dorsal dissection. To test this the experiments were repeated using a dorsal dissection similar to that of Steeves & Pearson ( 1982). DCMD responsiveness remained high both during stimulation of the extensor muscle with the tibia held at 90°, and during defensive kicking (Fig. 8), suggesting that it is indeed the massive dissection which abolishes the suppression of DCMD. This might result from a central depression caused by adaptation of sensory systems, or anoxia caused by disruption of the trachea.
The neural circuitry controlling the locust defensive kick is located in the metathoracic ganglion. Not surprisingly, disconnection of that ganglion from the brain by severing both connectives between the mesothoracic and metathoracic ganglia prevented the suppression of DCMD during defensive kicks. Severing either one of the connectives alone did not prevent the suppression, indicating that information inducing the suppression ascends through both connectives.
DISCUSSION
The DCMD interneurone has long been known for its variable responsiveness and lability. The process of antifacilitation, involving decrement in response to repeated stimulation, and the counteracting sensitization have been extensively documented (Horn & Rowell, 1968; Rowell & Horn, 1968; Rowell, 1971a, b). However, there remains a considerable degree of ‘unexplained’ variability; so much so that DCMD has been quoted as the neural exemplification of the Harvard Law of Animal Behaviour – that ‘under precisely controlled experimental conditions the animal does what it damn well pleases’ (Rowell, 1976). It has been noted that DCMD can occasionally cease to transmit for considerable periods of time, and then suddenly return to full responsiveness. This has led to the suggestions that the locust can facultatively’turn off’the DCMD (Rowell & O’Shea, 1980). In this paper I report one set of conditions in which the animal does just that.
The results show that the response of the DCMD to a visual stimulus is suppressed during the motor activity of the defensive kick. The suppression is not the result of cross-modal inhibition from sensory stimulation causing the behaviour, since the same sensory stimulation fails to suppress the DCMD when it fails to elicit the behaviour. It is not a result of the sensory stimulation resulting from the behaviour, since electrical stimulation which (approximately) mimics the behaviour does not suppress the DCMD. The DCMD shows no significant habituation to control visual stimuli presented under the same experimental conditions. Thus it appears that the suppression of this interneurone is a result of central activity, and should be considered an integral part of the behaviour itself. This means that the DCMD cannot function to trigger the M neurone during a defensive kick.
Locus of suppression
The locus of the inhibitory gating of the DCMD has not been investigated in the experiments. However, a likely candidate is the electrical synapse between the lobular giant movement detector (LGMD) and the DCMD (O’Shea & Rowell, 1975). The DCMD receives a barrage of small chemical postsynaptic potentials (PSPs) at this site, as well as the large electrical PSP from the LGMD. These chemical PSPs have never been seen to initiate spikes in DCMD, and it has been suggested that their function is to inhibit the electrical synapse by short-circuiting the postsynaptic membrane (Rowell & O’Shea, 1980). Inhibitory modulation at this site has been shown to result from visual stimulation of the contralateral eye. Furthermore, tactile stimulation of the type used in this paper to induce kicking also causes an increase in frequency of the chemical PSPs, while electrical stimulation of the connective can cause transmission failure. The relative rarity with which such transmission failure has been reported previously may be due to the employment of dorsal dissection, since the present results show that this extensive dissection also abolishes the suppression of the DCMD during co-contraction.
Suppression during jumping?
The experiments of this report were all conducted on restrained locusts performing defensive kicks. Several attempts were made to record from the DCMD using implanted myogram-type electrodes in the free-walking animal. Although such recordings were successful so long as the locust was quiescent, as soon as it started to walk or perform any behaviour associated with jumping (pre-jump crouch, co-contraction etc.) the background activity recorded from the vnc and adjacent muscles reached such a level that it was difficult or impossible to tell whether DCMD was responding or not. This, coupled with the refusal of the animal to orientate consistently to the standard visual stimulus, made it impossible to determine whether DCMD is suppressed in the escape jump in the same way that it is suppressed during defensive kicking.
Function of suppression?
An unrestrained locust presented with a visual stimulus may respond by jumping, but a restrained locust rarely if ever responds by kicking (although the 1st stage of the behaviour, the initial flexion, may occur). This makes behavioural sense, in that a jump may be an appropriate response to a visually-perceived long-range threat, but a kick is appropriate only to a short-range threat indicated by more immediate stimuli. The function of the suppression might thus be to prevent a visual stimulus triggering a kick while the threat is still at long range. This has the not-unreasonable implication that the locust ‘knows’ whether a particular initial flexion is destined to lead to a jump or a kick, and switches DCMD on or off accordingly. It leaves intact the original hypothesis of the role of DCMD input to the M neurone as far as the jump is concerned.
Arguing against this possible function (which is specifically related to the differences between kicking and jumping), is the fact that stimuli such as vigorous handling, loud noises etc., which lead to an increase in central arousal and ‘struggling’ behaviour also cause suppression of DCMD (this paper; Rowell & O’Shea, 1980). When struggling is caused by handling, the distinction between a defensive kick and jump may simply reflect how tightly the handler has grasped the locust! DCMD is also suppressed during saccadic head movements (Zaretsky & Rowell, 1979), and partially suppressed during antennal cleaning (Rowell, 1971c). A feature common to all these activities (especially escape jumping) is that they might lead to massive ‘self-stimulation’ of visual interneurones. Central suppression of the DCMD may help to counteract this effect.
It is also important to consider the bioenergetic implications of jumping and kicking. The small size of the locust dictates the use of energy storage techniques in making powerful movements (Bennet-Clarke, 1976). This in turn necessitates a delay in response, the co-contraction phase. The delay must be disadvantageous to the locust, but this is presumably outweighed by the advantage resulting from the ability to make powerful movements. Thus it is extremely important that the tibia should not go off at ‘half-cock’, i.e. tibial extension should not be produced before extensor muscle tension has developed to the maximum. As the M neurone becomes depolarized by the proprioceptive input from the leg during co-contraction there is a danger that further sensory input from a different modality might induce premature trigger activity. The DCMD has been shown to inhibit pre-synaptically its own output synapses, and it is suggested that this is to ensure that DCMD input to the M neurone does not produce spurious trigger activity in the quiescent animal (Pearson & Goodman, 1981). The suppression of the DCMD during restrained defensive kicking and (if it occurs) escape jumping may serve a similar function in the more critical period of co-contraction.
Both of the above suggestions imply that the DCMD is suppressed during both kicking and jumping. If this is the case it raises the question as to whether the DCMD can ever trigger defensive kicking or escape jumping as was originally suggested. Two possible mechanisms exist. First, the visual stimuli likely to challenge the unrestrained locust in its natural environment may be much more potent than those available under experimental conditions. Thus the suppression described above may act as a filter to remove weak visual stimuli, but still allow powerful stimuli indicating an emergency situation to trigger the behaviour. Such filtering properties may be necessary to allow the DCMD to perform more than one function. As well as exciting the M neurone, the DCMD also excites the C interneurone which produces the initial flexion or cocking response (Pearson & Robertson, 1981). Since the cocking response is an essential prelude to a jump or kick, it must have a lower threshold then the later stages of the behaviour. Thus the DCMD may operate at maximum sensitivity to produce the cocking response, and then be subject to a filtering process to reduce sensitivity during co-contraction. The second possibility is that since the locust can evidently control the responsiveness of DCMD, there might be a ‘window’ of sensitivity when extensor muscle tension has reached maximum. The weak DCMD response occasionally recorded during co-contraction (e.g. Fig. 2B) may occur in such a window. However, this is perhaps unlikely, since DCMD responsiveness is certainly suppressed for a period following the kick as well as in the preceding co-contraction.
ACKNOWLEDGEMENTS
This work was supported by grants from the Royal Society and the Science and Engineering Research Council.