THE LOCUST JUMP II. NEURAL CIRCUITS OF THE MOTOR PROGRAMME. BY W. J. HEITLER* AND M. BURROWSf

Transcription

1 exp. Biol. (i977), 66, figurct Printed in Great Britain THE LOCUST JUMP II. NEURAL CIRCUITS OF THE MOTOR PROGRAMME BY W. J. HEITLER* AND M. BURROWSf Department of Zoology, University of Oxford, and Department of Zoology, University of Cambridge. {Received 4 August 1976) SUMMARY 1. Neural circuits which co-ordinate the motorneurones of the metathoracic tibiae of the locust in jumping and kicking have been investigated. 2. The fast extensor motorneurone is reflexly excited by the subgenual organ, by a network of cuticle strain receptors, and by Brunner's organ. The subgenual organ and the cuticle strain receptors are excited by tension in the extensor muscle and mediate a positive feedback which could help to produce the burst of fast extensor spikes which precedes a jump or kick. Brunner's organ is stimulated by pressure from the flexed tibia, and will be excited by the initial flexion and throughout the co-contraction phase of a kick. 3. A central excitatory, connexion from the fast extensor to the slow extensor ensures that extensor muscle tension is as great as possible early in the co-contraction phase of a kick. 4. A central excitatory connexion from the fast extensor to flexor motorneurones is confirmed. This ensures that flexor muscle tension is great enough to keep the tibia flexed when the extensor muscle tension starts to develop before a jump or kick. 5. Reflex excitation of flexor motorneurones occurs in response to an extensor muscle twitch when the tibia is flexed. This helps to maintain the frequency of flexor spikes after habituation of the central fast extensor to flexor connexion. 6. A receptor, the 'lump receptor', which is stimulated by flexor muscle tension only when the tibia isflexed,can inhibit theflexormotorneurones and may activate the trigger system which allows the extension of the tibia in a jump or kick. 7. Receptors in the suspensory ligaments of the joint inhibit the fast extensor when the tibia extends. INTRODUCTION There are three stages in the motor programme of the locust kick (Heitler & Burrows, 1976). First, the hind tibiae are firmly flexed about the femora. Then the flexor and extensor tibiae muscles co-contract while the tibiae remain flexed. Thirdly, Present address: Department of Zoology, University of California, Berkeley, California 94720, U.S.A. t Balfour Student, University of Cambridge. 13 EXB 66

2 222 W. J. HEITLER AND M. BURROWS FEMUR, n5bl Extensor muscle Retractor unguis muscle Chordotonul organ n5- Lateral nerve Brunner's organ n5b2 Campaniform sensilla Subgenual organ Fig. i. A diagrammatic representation of the metathoracic femur and proximal tibia. The main muscles, nerves and sense organs are shown. the flexor tibiae muscles relax so that the tibiae are extended rapidly. In this paper some of the neural connexions involving excitatory motorneurones which underly this motor programme will be described. Two circuits have already been implicated. First, there is a peripheral re-excitatory reflex which depolarises the fast extensor tibiae motorneurone (FETi) if that neurone spikes when the tibia is restrained in a flexed position, and which may help to produce the sequence of FETi spikes that precedes a jump or kick (Burrows & Horridge, 1974). Secondly, there is a central excitatory connexion from the FETi to the flexor tibiae excitor motorneurones which might produce the co-activation of the antagonist motorneurones before a jump or kick (Hoyle & Burrows, 1973). These circuits will be described in detail, and further reflexes, mediated by a variety of sense organs in the hind legs, will be examined to determine their possible roles in jumping and kicking. MATERIALS AND METHODS The experimental methods, and abbreviations used for the motorneurones are as described in the previous paper (Heitler & Burrows, 1976). Additional techniques were as follows. Paired stainless steel hook electrodes were used to record differentially from nerves and to stimulate them. Muscle and cuticular stresses were measured with an R.C.A mechano-electric transducer. Sense organs were stained by dissecting the hind leg so that the cut ends of the appropriate nerves could be placed in a small pool of cobaltous chloride. The preparation was kept in a moist chamber for 24 h and the cobalt was then precipitated as cobalt sulphide. Standard fixation, dehydration and clearing techniques were used. RESULTS A single large nerve trunk enters the femur from the coxa. It is formed by the fusion of nerve 3 b and the major branch (5 b) of nerve 5, both from the metathoracic ganglion. Within the femur the nerve splits into several branches which innervate the following structures (Fig. 1).

3 Neural circuits of the locust jump 223 I Free I Held Held "^ 1 Ipsi Contra Intact (0 (c)(i) V (ii) If ~\ N5 (<O(0 (iii) (iv) Mm N5bl Fig. 2. Phasic re-excitation of the FETi is mediated through nerve 5b 1 distal to the chordotonal organ, (a) The FETi is stimulated antidromically at 1 Hz by electrodes placed in the extensor muscle. ((a)(i)) The tibia is initially free to kick; ((a)(ii), (iii)) the tibia ia held at 90 0 and a large depolarizing wave then re-excites the ipsilateral FETi (upper trace), while a smaller and more rapidly habituating wave excites the contralateral FETi (lower trace). (6) The re-excitation is present with the leg intact (i), after cutting nerve sb2 (ii), and the lateral nerve (iii), but is abolished by cutting nerve 5bi distal to the chordotonal organ (iv). (c) Stimulation of nerve 5b 1 distal to the chordotonal organ at 1 Hz (open triangle in (i) marks the stimulation artifact). Spikes recorded in nerve 5 at its entry to the ganglion (lower trace) are recruited with increasing stimulus intensity (i iv), as is a compound EPSP in the FETi (upper trace), (d) Antidromic stimulation of the FETi at 1 Hz with the tibia held at oo did not initially cause re-excitation (i iii), although a spike recorded in nerve sbi distal to the branch that innervates the extensor muscle (lower trace) was consistently followed by an EPSP in the FETi (upper trace, open triangle in (i)). After some minutes the re-excitation suddenly reappeared (iv). Calibration: vertical, (a) upper 8 mv, lower 4 mv, (6) 8 mv, (c) 4 mv, (d) 10 mv; horizontal (a, b and d) 160 ms, (c) 40 ms.. The extensor tibiae muscle, the femoral chordotonal organ, and the subgenual organ in the tibia. N5b2. The flexor tibiae muscle, the retractor unguis muscle, campaniform sensilla and cuticular hair receptors in the tibia, and tarsal muscles and receptors in the distal part of the tibia. The lateral nerve. This is a branch of nerve 5b2 that innervates cuticular receptors in the proximal region of the femur, Brunner's organ, the ' lump' receptor and receptors in the suspensory ligaments of the femoral-tibial joint. The task is to assess the contribution of these various sensory structures to the production of the motor programme of a kick. 15-3

4 224 W. J. HEITLER AND M. BURROWS Neural circuits controlling extensor motor activity In a kick, both fast and slow tibial extensor motorneurones spike at high frequency starting about \ s before and continuing until the tibia extends. Both motorneurones are inhibited immediately after the kick. What are the neural circuits that produce this sequence of activity? Excitation mediated by the subgenual organ A spike in the fast extensor motorneurone (FETi), when the tibia is fixed less than half extended about the femur, evokes a wave of excitation in both itself and the slow extensor motorneurones (Fig. za, and Hoyle & Burrows, 1973). The contralateral FETi is depolarized also but this crossed effect is much smaller and decrements rapidly upon repetition (Fig. 2 a). The excitation of the ipsilateral FETi is mediated by a peripheral reflex activated by the isometric twitch of the extensor muscle (Burrows & Horridge, 1974). The excitatory post-synaptic potentials (EPSPs) which underly the excitation can be matched in both the ipsilateral slow and fast extensor motorneurones, so they probably originate from the same source (Burrows & Horridge, 1974). Which sense organs are involved in the reflex? The re-excitation of the FETi persists after cutting nerve 5b2 in the proximal region of the femur (Fig. 2(b) (ii)). This eliminates input from the campaniform sensilla which lie close to the subgenual organ in the tibia. The re-excitation also persists after cutting the lateral nerve (Fig. 2 (b) (iii)), which contains the axons of receptors in the suspensory ligaments of the joint (Coillot & Boistel, 1968). It is abolished, however, if nerve 5bi is cut distal to the femoral chordotonal organ (Fig. 2(6) (iv)). Cobalt sulphide staining of the nerve shows that it contains the axons of the subgenual organ and those of a diffuse subcuticular network of sensory neurones including cuticular hair receptors in the tibia (Fig. 3). To control against the possibility of having damaged the chordotonal organ itself whilst cutting nerve 5b 1, the proximal tibia of another locust was cauterized with a hot needle in the region of the subgenual organ. The chordotonal organ still gave spikes in response to tibial movement, but spikes from the subgenual organ were abolished and the re-excitation of the FETi was eliminated. We conclude that the subgenual organ is responsible for this reflex. Evidence below (see section ' Excitation mediated by cuticle strain') will suggest that the diffuse subcuticular network mediates a reflex with different properties. Stimulation of nerve 5b 1 distal to the chordotonal organ evokes post-synaptic potentials (PSPs) in the FETi, but the exact nature of these can vary. Sometimes a gradual increase in the intensity of stimulation increased the size of an EPSP, and this was correlated with the recruitment of units recorded in nerve 5 at its entry to the ganglion (Fig. 2 (c) (i-iv)). On other occasions a similar regime of stimulation produced a compound excitatory and inhibitory PSP in all-or-none manner in the FETi. Variability was also apparent in the re-excitation of the FETi itself. In most preparations it could be elicited routinely, but, in one, antidromic stimulation of the FETi with the tibia held at 90 0 failed to produce the re-excitation (Fig. 2 (d) (i iii)). This was not due to blocking at the hook electrodes recording from nerve 5b 1 (the only remaining intact nerve), since an EPSP consistently followed a sensory spike occurring

5 Journal of Experimental Biology, Vol. 66, No. Fig-3 n5bl 1 mm *? Distal Fig. 3. A cobalt sulphide stain of the peripheral terminals of axons in nerve 5b 1 distal to to the chordotonal organ. The structures which have been stained are the cell bodies of hair cells, a variety of diffuse subcuticular processes and the subgenual organ. The inset shows the location of these tissues in the proximal tibia. W. J. HEITLER AND M. BURROWS (Facing p. 224)

6

7 Neural circuits of the locust jump 225 N5bl 200 ms N5bl Fig. 4. Reflexes mediated through nerve sbi distal to the chordotonal organ. All other peripheral nerves had been cut. (a) Pressing on the proximal tibia causes spikes in nerve sbi (upper traces) and EPSPs in FETi (lower traces). (6) A similar experiment in a less excited locust does not reveal any phase-locking between the spikes (dotted) in nerve sbi and EPSPs in the FETi. (c) Loud noises produced by hand-claps cause spikes in nerve sbi (lower trace) and EPSPs in the FETi (upper trace). Calibration: vertical (a, c) 5 mv, (6) 3 mv. after the main burst (Fig. 2 (d) (i)). The stimulation was maintained at low frequency for some minutes and suddenly the re-excitation appeared (Fig. z{d) (iv)). So far as could be ascertained neither the stimulation regime nor the peripheral situation had changed, and there was no obvious difference in the activity recorded in the nerve. Pressing on the proximal tibia close to the subgenual organ, in an attempt to mimic the femoral distortion that occurs before a jump, causes excitation of the FETi (Fig. \a, b). There are other units in the subgenual organ which respond to sound and which also evoke a depolarization of the FETi (Fig. 4c). In an aroused locust whose tibia is fixed at 90 0 about the femur the re-excitation of FETi after a single antidromic stimulus often triggers a second spike, which then causes further re-excitation and a third spike, and so on, producing a long sequence of FETi spikes and a strong push by the tibia (Fig. 5 a). This behaviour is often seen when an intact locust performs a defensive kick, or is struggling, and the tibia meets with an obstruction which resists the extension. The amplitude of the re-excitation with the tibia fixed at 90 0 is considerably greater than that with the tibia held flexed (Fig. 5 b). This no doubt reflects the greater lever ratio of the extensor muscle when the tibia is at 90 0 relative to that when the tibia is fully flexed (Heitler, 1974). It means that when the tibia is in the flexed position a higher degree of arousal is needed before the re-excitation following an FETi spike will evoke a second spike. The required arousal will normally result, however, from

8 226 W. J. HEITLER AND M. BURROWS (*) 20 Hz Wk j Extended Fig. 5. Re-excitation of the FETi mediated by the subgenual organ, (a) The re-excitation following an antidromic spike in the FETi with the tibia fixed at 90 about the femur can trigger a train of spikes which is terminated only when the re-excitation becomes subthreshold (open triangle). (6) Three superimposed traces of antidromic stimulation of the FETi with the tibia restrained at 90 0 (upper), fully flexed (middle), and fully extended (lower), (c) With the tibia held fully flexed and unable to move the re-excitation following antidromic stimulation of the FETi is progressively increased following a single stimulus (lower trace), 10 Hz for 0-5 s (middle trace), 20 Hz for 0-5 s (upper trace). Calibration: vertical {a, b) 5 mv, (c) 12 mv; horizontal (a) 100 ms, (b, c) 500 ms. the sort of external stimuli which evoke jumps or kicks. Once the tension in the extensor muscle starts to build up in response to the first few FETi spikes, the re-excitation reflex will then be augmented (Fig. 5 c). The positive feedback from the subgenual organ therefore tends to produce a runaway excitation and an accelerating sequence of spikes both in the FETi and in the SETi. Excitation mediated by cuticle strain receptors If the tibia is forcibly extended to the most extreme position possible, large EPSPs occur in the fast and the slow extensor motorneurones (Burrows & Horridge, 1974). There are two points to establish. First, what are the receptors that signal this effect? Secondly, is there any stimulus which could activate the receptors during the motor programme of the jump? For example, could extreme extension of the tibia (which is not achieved until the jump is complete and so cannot be involved in producing the jump motor programme) be activating receptors other than those that signal extension itself. The EPSPs evoked by extreme extension of the tibia still occur in the FETi and can lead to a burst of spikes even if the tendons of the flexor and extensor muscles, and nerves other than the lateral nerve, are cut in mid-femur (Fig. 6 a, b). Single pulse stimulation of the lateral nerve when all other nerves are cut usually causes an inhibitory post-synaptic potential in the FETi (Fig. 6(c) (i)), probably because a resistance reflex is induced by stimulating axons of sensory neurones sensitive to extension movements (see section 'Reflex inhibition of the fa9t extensor' below). As the intensity of stimulation is increased, a compound EPSP is recruited which decrements with continued stimulation (Fig. 6 c). Repetitive high intensity stimulation of the lateral nerve can evoke a burst of spikes in the FETi similar to that caused by forced extension (Fig. yd). Low-intensity stimulation can also produce a burst of FETi spikes, but many more stimuli have to be delivered (Fig. yb). The simplest explanation for this is that the excitation is mediated by an interneurone or group of interneurones which themselves receive excitatory inputs from many small axons in

9 Neural circuits of the locust jump 227 (i) (iv) Fig. 6. Excitation mediated by cuticle strain. Forced tibial extension excites the FETi when all muscle tendons and all nerves other than the lateral nerve have been cut. (a) Tibial extension causes spikes in the lateral nerve (lower traces), and forced extension causes a sequence of spikes in the FETi (upper traces) which is not terminated even if the tibia is flexed (at cessation of lateral nerve spikes). (6) This sequence of FETi ipikes is not prolonged even if forced extension is maintained, (c) Stimulation of the lateral nerve at 1 Hz causes PSPs in the FETi. The initial stimulus was of low intensity (i); succeeding ones were of high intensity (ii iv). Calibration: vertical (a) 16 mv, (6) 8 mv, (c) 10 mv; horizontal (a) 400 m>, (6) aoo tns, (c) 160 mi. Spike Fig. 7. Effects of stimulation of the lateral nerve, (a) High-intensity stimulation of the lateral nerve at 30 Hz rapidly induces a burst of FETi spikes which continue after stimulation ceases, (b) Stimulation at low intensity also causes a burst of spikes, but with a longer latency. The short vertical lines in (a) and (6) are stimulation artifacts, (c) Maintained high intensity stimulation initially causes EPSPs and spikes in the FETi (upper trace), but also later causes IPSPs. The lower trace is a monitor of the stimulation, (d) Pressing on the distal femoral cuticle after amputation of the femoral tibial joint causes EPSPs and a burst of spikes in the FETi. Calibration: vertical 8 mv; horizontal (a, 6) 100 ms, (c) 400 ms, (<f) 200 ms.

10 228 W. J. HEITLER AND M. BURROWS the lateral nerve. Low intensity stimulation excites only a few of these, and there must be considerable temporal summation before the threshold of the interneurones is exceeded, while high-intensity stimulation excites more of the inputs so that the resulting spatial summation rapidly takes the interneurones above their threshold. Once the burst of FETi spikes has started it cannot be terminated simply by cessation of stimulation (whether mechanical or electrical), but the frequency of spikes gradually declines until the burst stops (Fig. 6 a, ya). If forced extension of the tibia is maintained, the burst of FETi spikes still stops after a similar period (Fig. 6 b). This suggests that the excitation, once initiated, runs its own course independent of subsequent stimulation. If high-intensity electrical stimulation is maintained the burst of FETi spikes can be prolonged, but after a few seconds the stimuli produce IPSPs as well as, or instead of, EPSPs (Fig. yc). Presumably the excitatory circuits have fatigued. A similar burst of spikes can be elicited in the FETi by distorting the distal femoral cuticle after removal of the entire femoral tibial joint (Fig. yd). Distortion of this region of the cuticle occurs during the co-contraction of flexor and extensor muscles that precedes a kick or jump (Bennet-Clark, 1975). It seems likely that forced extension of the tibia to its extreme causes a distortion of the femoral cuticle mimicking that which occurs before a kick or jump when the tibia is flexed and the flexor and extensor muscles are co-contracting. Therefore in a kick it will be the tension in the femoral muscles that distorts the femoral cuticle and excites the receptors. These experiments show that stimulation of axons in the lateral nerve can elicit the excitatory reflex, and that at least some of the receptors responsible must be proximal to the femoral-tibial joint. The only sensory structure, proximal to the femoral-tibial joint, that is revealed by cobalt staining of the lateral nerve is a subcuticular plexus similar to that proximal to the subgenual organ (Fig. 3). The reflex excitation of the extensor motorneurones mediated by the subcuticular plexus is distinguishable from that mediated by the subgenual organ on the following grounds. It is not abolished by cutting nerve 5b 1, and it persists after cessation of the stimulus. There is no reduction in the re-excitation in response to a single FETi spike when the lateral nerve is cut and nerve 5b 1 is still intact, indicating that the cuticle strain receptors of the lateral nerve do not mediate significant excitation in response to the low strains induced by an isometric muscle twitch from an FETi spike. The subgenual organ mediates a phasic re-excitation which operates even in response to a single isometric muscle twitch, while the cuticle receptors mediate a tonic re-excitation which only operates when the muscle tension, and consequently cuticle strain, has reached a higher level. Excitation mediated by Brunner's organ Brunner's organ is a small tubercule on the ventral surface of the femur about a quarter of the way from the proximal end for which no function has yet been shown (Jannone, 1940; Joly, 1951; Uvarov, 1966). There are typically three sensory hairs and two campaniform sensilla just proximal to the base of the tubercle (Fig. 8b, c). The tubercle is flattened on to the hairs when the tibia fits into a groove on the ventral femur upon full flexion. Mechanical stimulation of the organ with a blunt needle evokes a barrage of small amplitude EPSPs in the ipsilateral FETi which are often

11 Journal of Experimental Biology, Vol. 66, No. i Fig mv 400 ms 160//m (< ) External Internal Anterior Distal Proximal Posterior Fig. 8. Brunner's organ, (a) Mechanical stimulation (indicated by dots) of Brunner's organ after cutting all distal nerves causes EPSPs in the ipsilateral FETi (upper trace) and has no effect on the contralateral FETi (lower trace). Scanning electron micrographs of the ventral femoral cuticle show (b) the external and (c) internal structure of Brunner's organ. W. J. HEITLER AND M. BURROWS (Facing p. 228)

12

13 Ext. T Flex Neural circuits of the locust jump ms YrmTi 1 M I I I HI III II III Illlll II II IIIIII III H II II 1 Fig. 9. Inhibition mediated by the lateral nerve. The FETi (upper trace) is inhibited by axons in the lateral nerve (lower trace) upon imposed tibial extensions after all other nerves have been cut. Upward arrows indicate extension, downward flexion, (a, b) Imposed phasic extensions and flexions, (b) The inhibition has been accentuated by depolarizing the FETi with about 4 na of injected current, (c) The tibia is maintained at about 90 and the axons are tonically active. IPSPs can be correlated with sensory spikes when they occur in groups early in the trace (connecting lines), but not later (dashes). masked by other synaptic inputs. The EPSPs can be revealed by cutting the femoral nerves distal to the organ (Fig. 8a), thereby reducing the level of synaptic input from other sense organs. The EPSPs are never large enough to evoke a spike in the FETi on their own, but when summed with other excitatory inputs, such as those resulting from the arousing stimulus applied to induce a kick, and those described above, the input from Brunner's organ could contribute to the production of an FETi spike. The locust can jump and kick after excision of Brunner's organ, and so this contribution is not essential. Reflex inhibition of the fast extensor motorneurones The FETi is usually inhibited immediately after the extension of the tibia in a kick. An imposed tibial extension (but not forcing it to its extreme position) evokes an hyperpolarization in the FETi if the lateral nerve is intact but all other nerves and muscle tendons are cut (Fig. 9). The effect depends upon the integrity of the suspensory ligaments of the joint and it is the receptors in these ligaments (Coillot & Boistel, 1968, 1969) which are probably responsible. The hyperpolarizations in the FETi are proportional to the degree and speed of the imposed tibial extension and

14 W. J. HEITLER AND M. BURROWS 00' FETi Fig. 10. A central excitatory connexion from the FETi to the SETi. (a) Repetitive injection of short current pulses (not monitored) causes the FETi to spike at o-a Hz in an isolated ganglion. An EPSP and spike in the SETi follows each FETi spike. Six FETi sweeps have been superimposed (upper trace), while the corresponding six SETi sweeps are displayed separately (lower trace). (6) A current pulse (monitored in the lower trace with downwards depolarizing) of longer duration causes the FETi (upper trace) to spike at 10 Hz. The EPSPs in the SETi (middle trace) rapidly decrement, (c) Injection of current (lower trace) into the FETi which does not evoke spikes (upper trace, two sweeps superimposed) has no effect on the SETi (2nd and 3rd traces which have been separated), (d) Hyperpolarizing the SETi increases the amplitude of the EPSP. Three FETi sweeps at o-i Hz are superimposed (upper trace), with FETi spiking twice per sweep. Three SETi sweeps are displayed separately, with 25 na hyperpolarization (2nd trace), 20 na hyperpolarization (3rd trace), and 10 na hyperpolarization (4th trace). Only one sweep of the current monitor is shown to indicate the duration of the injected pulse (lower trace), (e) A short duration current pulse (lower trace) produces an FETi spike just after the pulse (upper trace), and a short latency EPSP in the SETi (middle trace). Four sweeps are superimposed in all traces. Calibration: vertical FETi (a-c) 25 mv, (d, «) 50 mv; SETi (a) 12-5 mv, (b-d) 5 mv, (e) 2-3 mv; current (b-d) 123 na, (e) 5 na; horizontal (a, c, d) 325 ms, (6) 650 ms, («) 30 ms. can be accentuated by passing depolarizing current into the FETi (Fig. gb). Individual IPSPs cannot usually be distinguished but if the tibia is held partially extended, groups of IPSPs can be associated with the groups of spikes in the lateral nerve (Fig. 9 c). A central excitatory connexion from the fast to the slow extensor motorneurone If the FETi is induced to spike in an isolated ganglion by injection of depolarizing current into its soma, an EPSP and sometimes a spike occurs in the slow extensor of the tibia (SETi) following each FETi spike (Fig. 10a). The EPSP occurs only when there is a spike in the FETi and decrements rapidly (Fig. 106); a depolarization below the spike threshold of the FETi has no effect on SETi (Fig. 10c). The size of the EPSP in the SETi can be increased by hyperpolarizing the SETi (Fig. lod) and occurs with a latency of about 2-5 ms from the peak of the FETi spike (Fig. ioe). The result is contrary to previous reports that a spike in FETi produces centrally mediated inhibition of the SETi (Hoyle & Burrows, 1973: Burrows & Horridge, 1974). If the FETi is stimulated antidromically, IPSPs do indeed often occur in the SETi

15 Neural circuits of the locust jump 231 ' > SETi (a) (ii) (P) \ c) (i) (d) (0 FETi AFFITi [XI Fig. n. Common IPSPs in FETi, SETi and AFFITi. ((a) (i), (ii)) Antidromic stimulation of the FETi with the tibia free to extend slightly and then hitting a barrier. Common IPSPs are visible before the re-excitation in the FETi (upper trace) and the SETi (lower trace). (6) The IPSPs fail in both motorneurones. ((c) (i), (ii)) Antidromic stimulation of both FETi and SETi with the tibia free to extend. (<f) Antidromic stimulation of SETi alone can cause the IPSPs. ((«) ( )» (")) Th e IPSPs are also common to a flexor motorneurone (lower trace) and curtail a centrally mediated EPSP. Calibration: vertical (a-d) 2 mv, («) i6mv; horizontal (a, b (c i) and e) 160 ms, ((c ii), d) 60 ms. but these IPSPs are reflex in origin. The IPSPs occur simultaneously in the SETi and the FETi before the re-excitation that results when the tibia kicks against an immovable bar (Fig. 11 a). Should the IPSPs fail in the FETi, then they are also absent in the SETi (Fig. 11 b). If an antidromic SETi spike as well as an antidromic FETi spike is evoked, the common IPSPs are more clearly seen as they are now superimposed upon a lower membrane potential during the falling phase of the SETi spike (Fig. 11 c). An antidromic spike in the SETi alone is sometimes sufficient to elicit the IPSPs in both neurones (Fig. 11 d) and this explains the inhibitory effect of the SETi upon the FETi occasionally observed by Hoyle & Burrows (1973). The same IPSPs are also common to some flexor motorneurones (Fig. 11 e). The reflex is maintained even if most of the femur is amputated, leaving just enough for antidromic stimulation of the FETi. It was probably this which caused the original mistaken identification of the reflex as a central connexion. In a kick the central, but rapidly decrementing excitatory connexion from the FETi to the SETi will only be effective in response to the first few FETi spikes. It is these early FETi spikes, however, that are crucial in initiating a variety of reflexes dependent upon tension in the extensor muscle. The tension developed by the first FETi spike is small relative to that later in the contraction so that the recruitment of the SETi and its contribution to the early tension may be important in ensuring that these reflexes are successfully initiated. The extensor tibiae muscle displays catch properties

16 W. J. HEITLER AND M. BURROWS 7\./V 100 ms Fig. 12. The central excitatory connexion from the FETi to the flexor motorneurones. (a, b) The FETi is induced to spike in an isolated ganglion by injecting depolarizing current. The FETi axon spikes are recorded in nerve 5 (dotted spikes, lower trace). A centrally mediated EPSP is recorded in the anterior fast flexor (AFFITi) (upper trace) following each FETi spike, and at least 4 different flexor axon spikes are recorded in nerve 5. The EPSP decrements with repeated FETi spikes until no flexor spikes are initiated (arrow in a), (c, d) The FETi is stimulated antidromically at 20 Hz (upper traces) in an intact locust, and evokes a centrally mediated EPSP in the AFFITi (lower traces) whose amplitude is correlated with the size of the FETi spike (arrows). Calibration: vertical (a, b) 10 mv, (c, d) upper 50 mv, lower 12 mv. (Wilson & Larimer, 1968), so that the interjection of a few extra SETi spikes might increase the amount of tension, and this level of tension could subsequently be maintained by a much lower frequency of spikes. Neural circuits controlling flexor motor activity Flexor motorneurones spike at high frequency for about \ s before the kick. This sequence is terminated by a large and rapid inhibition which we have called the trigger activity (Heitler & Burrows, 1976). What are the neural circuits which contribute to this sequence of action by the flexor motorneurones? A central excitatory connexion from fast extensor to flexor motorneurones A large compound EPSP occurs in the flexor motorneurones following each FETi spike even in a ganglion that has been isolated by section of its peripheral nerves. This confirms the report of Hoyle & Burrows (1973) that there is a central excitatory connexion from the FETi to the flexor motorneurones. Following each FETi spike a burst of spikes from at least four flexor motorneurones is recorded in the proximal end of the cut nerve 5 before it enters the femur (Fig. 12 a, b). With repeated FETi spikes the size of the EPSP gradually decreases in a flexor motorneurone and the total number offlexormotor spikes recorded in the nerve also diminishes (Fig. 12 a, b). Two questions arise from these observations: what is the nature of the central connexion, and what role does it play in the behaviour of the locust? The following observations indicate that the connexion is mediated by chemical synapses. Injection of current below spike threshold into the soma of the FETi has no

17 Neural circuits of the locust jump 233 effect on the membrane potential of the flexor motorneurones as recorded in their somata (Hoyle & Burrows, 1973); the interaction is only seen in response to an FETi spike. The amplitude of the EPSP can be reduced by injecting depolarizing current into a flexor motorneurone and increased by hyperpolarizing current (Hoyle & Burrows, 1973). Bathing the ganglion in a high Mg^1", zero Ca 2+, saline reversibly reduces the size of the EPSP. The EPSP is therefore probably chemically mediated. It has been suggested (Hoyle & Burrows, 1973) that the shortest latency EPSPs, which occur 3-6 ms after an antidromic FETi spike, might be mediated monosynaptically by a collateral branch of the FETi. The following observations argue against this. Small deflexions are superimposed on the EPSP in the flexor mediated by an FETi spike. Some of these are no doubt due to synaptic inputs unrelated to the FETi spike, but they predominate on the rising phase of the EPSP. The deflexions are not due to spikes in the flexor motorneurone itself, attenuated and smoothed in electrotonic conduction from the spike initiating site to the soma, since they occur even when recordings from nerve 5 show that no flexor motorneurones have spiked (arrow, Fig. 12 a). These deflexions are probably small synaptic potentials underlying a compound EPSP, resulting from activity of one or more interneurones in the path from the FETi to the flexor motorneurones. Furthermore, the size of the EPSP in the flexor varies with the size of the antidromic FETi spike (Fig. izc, d). This obviously raises interesting possibilities about the modulation of the interaction by some as yet undefined mechanism, but for the moment we present the result simply as evidence for the presence of at least one interneurone in the pathway mediating the FETi-flexor interaction. What role might the connexion play in the behaviour of the locust? In a kick, a single FETi spike can produce a large depolarization of several fast flexor motorneurones and an increase in the frequency of their spikes (Fig. 136). The high frequency of flexor spikes is maintained until they are inhibited by the trigger activity which thereby terminates the co-contraction. Any FETi spikes that occur shortly after the trigger inhibition fail to produce an EPSP in the flexor (Fig. 13 a). This might be due simply to fatigue of the interaction owing to the high spike frequency of the FETi during the co-contraction phase or to other as yet undefined mechanisms. The interaction does decrement (Fig. 12) but it has not yet been possible by intrasomatic depolarization to make the FETi spike at high enough frequency to mimic its activity prior to a kick. The central interaction contributes initially to the production of co-activation, but is it still effective in the later phases? The problem lies in the difficulty of separating the central from possible reflex components of flexor depolarization. This has been partially achieved in the following way. The FETi and a posterior group fast flexor (PFFITi) were penetrated on one side of the locust. The ipsilateral hind leg was then amputated at the coxa, and the locust induced to kick with its other hind leg. The stage of arousal of one locust was so high that, despite the absence of any ipsilateral peripheral feedback, a kick-like sequence of spikes occurred in the FETi (Fig. 13 A). The sequence of FETi spikes evoked EPSPs and spikes in the flexor. The frequency of flexor spikes increased following the first four FETi spikes, but later FETi spikes were not so effective. By the time the frequency of FETi spikes had reached a maximum (just before the contralateral kick),

18 234 W. J. HEITLER AND M. BURROWS IKick Fig. 13. Failure of the FETi-flexor connexion, (a) A kick in which the FETi (upper trace) spikes shortly after the movement (arrow, and monitored in lower trace). This spike (open triangle) does not cause an EPSP in the AFFITi (middle trace). (6) One hind leg has been amputated and the locust induced to kick with its remaining hind leg. The FETi on the amputated side (2nd trace) has a similar sequence of spikes to the FETi on the intact side (extensor myogram in 3rd trace), including three spikes (open triangles) after the kick (arrow and movement monitored in lower trace). A posterior group flexor on the amputated side (upper trace) is depolarised by the EPSPs resulting from the FETi spikes, but the EPSPs decrement as the spike frequency increases towards the kick, and recover after the kick at the lower frequency of FETi spikes. Calibration: vertical (a) 16 mv, (6) upper 16 mv, lower 32 mv. the flexor motorneurones had stopped spiking, and no depolarization is correlated with the FETi spikes. Three FETi spikes occurred at low frequency after the contralateral kick and there was a gradual recovery in amplitude of the EPSP in the flexor motorneurone (Fig. 136). This suggests that the FETi-flexor interaction might be ineffective in the later part of the co-contraction and that its function in kicks induced by tactile stimulation therefore might be to ensure that as soon as the FETi starts to spike there is also a high level of flexor muscle tension. Hence the tibia will be securely locked in the flexed position. This raises the question of what maintains the frequency of flexor splkes in the later stages of co-contraction. Reflex inputs to flexor motorneurones A twitch of the extensor muscle that results from an FETi spike evokes reflexive inputs to the flexor motorneurones that depend on the position of the tibia and on the tension in the femoral muscles. A single antidromic FETi spike and the associated twitch of the extensor muscle evoke an EPSP in the posterior slow flexor (PSFITi) whose initial height is similar when the tibia is held flexed or extended (the central component). There is, however, a long depolarizing tail to the EPSP in the flexed position (Fig. 14a) which is absent when the tibia is extended (Fig. 146). If the FETi is stimulated repetitively at 20 Hz different rates of decrement result, depending upon whether the tibia is flexed or extended, and are due probably to the summation of peripheral and central components. After 2 s of stimulation with the tibia flexed, the compound EPSP is still large enough to evoke a burst of nine PSFITi spikes (Fig. 14c), while after only ci s of stimulation with the tibia extended the EPSPs are so small that they fail to take the motorneurone above its spike threshold (Fig. i4<f). If the tibia is held flexed during the initial period of stimulation and then released

19 Neural circuits of the locust jump Flexed PSFITi /\T Extended 235 Fig. 14. Interaction of peripheral and central components in the excitation of a slow flexor motorneurone. The FETi is stimulated antidromically and the position of the tibia is varied, (a) Single pulse stimulation with the tibia flexed. The spikes recorded intracellularly in the posterior slow flexor (PSFITi) (upper trace) can be correlated with spikes in the flexor myogram (lower trace). (6) Single pulse stimulation with the tibia extended. Note the reflex inhibition following the central EPSP (arrow), (e) Repetitive stimulation at 20 Hz with the tibia flexed; (d) with the tibia extended. («) The initial stimulation is with the tibia held flexed, and then it is allowed to extend. The PSFITi is inhibited following the movement (arrow). while the stimulation is maintained, there is a repolarization of the membrane at the moment of extension (Fig. 14c). This is the opposite effect to that which a resistance reflex would produce. The same effects occur in fast flexor motomeurones (Fig. 15). The posterior fast flexor (PFFITi) does not spike so readily in response to the centrally mediated EPSP as does the posterior slow flexor, and the differences in the reflex components are best revealed by averaging. The prolonged excitatory tail then becomes clear when the tibia is flexed, but is absent when the tibia is extended (Fig. i$a- ). Repetitive stimulation reveals the difference in the rate of decrement of the depolarization in the two situations (Fig. i^d-f). There are IPSPs which curtail the EPSP in the flexor motomeurones whenever the tibia moves freely in response to an FETi spike, or is restrained in an extended position. Lack of the inhibition when the tibia isflexeddoes not explain the prolongation of the EPSP, which must therefore result from a peripheral excitatory input. It presumably stems from receptors in the leg which monitor the higher extensor tension that results when the tibia is flexed. In a kick, extensor tension is highest when the centrally mediated and decrementing excitation of the flexors is at its lowest. Cutting in turn, nerve 5bi to the chordotonal organ and subgenual organ, nerve 5b2 to the campaniform sensilla on the proximal tibia, and the lateral nerve innervating the joint receptors, produces a progressive reduction in the tail of the compound EPSP. Therefore a variety of sense organs probably contribute to the reflex, although variations in the central level of arousal of the locust make precise interpretation of such ablation experiments difficult.

20 236 W. J. HEITLER AND M. BURROWS (A) V Fl. (c). Ext. Fig. 15. Interaction of peripheral and central components in the excitation of a fast flexor motorneurone. The FETi (upper trace) is stimulated antidromically with single pulses at 0-2 Hz (a-c), and quadruplet pulses of 20 Hz at 0-2 Hz (d-f), while recording from the posterior fast flexor (PFFITi) (and trace). In (a) and (d) the waveforms are averaged over 64 stimulus sequences with the tibia held flexed and then extended. (6) and (e) are single sweeps with the tibia held flexed, (c) and (/) are single sweeps with the tibia held extended. In (6), (c), (e) and (J) a flexor myogram is on the lower trace. Calibration: vertical (a, d) 8 mv, (b, c, e, f) upper 25 rav, lower 16 mv; horizontal (a, d) 45 ms, (b, c) 80 ms, («,/) 400 ms. The 'lump' receptor Units in the lateral nerve which respond to extension of the tibia over the normal range of movement originate in two groups of receptors (the dorsal-anterior-lateral receptor and the dorsal-posterior-lateral receptor) situated in the suspensory ligaments of the joint (Coillot & Boistel, 1969). There is a third receptor group, with axons in the lateral nerve, the ventral-posterior-lateral receptor which can be stimulated by direct, experimentally applied pressure, but for which no natural stimulus has yet been described (Coillot & Boistel, 1968, 1969). The receptor occurs in the medial (posterior) groove between the femoral wall and the cuticular invaginarion in the distal ventral femur, the 'lump' (Fig. 16a: Heitler, 1974). Because of its position in the groove, the only time that this receptor, for brevity called the 'lump receptor', is subjected to pressure is when the tibia is flexed and there is tension in the flexor tendon. A pocket in theflexor tendon then pulls down over the lump (Heitler, 1974) and the posterior arm of the tendon bifurcation, which forms the pocket, rests directly on the receptor. When the tibia is fully flexed, spikes can be recorded in the lateral nerve in response to tension in the flexor tendon (Fig. 166, c), but if the tibia is held at 20 about the femur there is no response to tension in the flexor tendon (Fig. i6d). Mechanical stimulation of the lump receptor does not usually have any direct effect onflexor motomeurones. In only one of five preparations was any effect noticed and this was in a highly aroused locust in which the flexor excitor motomeurones were spiking spontaneously. Pulling on the flexor tendon with the tibia flexed caused a hyperpolarizing wave in the AFFITi which temporarily prevented spikes (Fig. 17a.) It seems reasonable to attribute the effect to the lump receptor, but it is possible that other sense organs, such as cuticle strain receptors are also involved. Electrical stimulation of the lateral nerve usually evokes EPSPs but sometimes IPSPs in flexor

21 Neural circuits of the locust jump 237 (a) Femur Extensor tendon Tibia (p) Lateral nerve Position Force (c) (d) Fig. 16. The lump receptor, (a) A cobalt sulphide stain of the peripheral terminals of the lateral nerve viewed from the anterior (i.e. external) side of the femur. The muscle tendons have been cut and deflected dorsally and ventrally, and the ventral femoral cuticle deflected so that it i» viewed from the dorsal rather than the lateral side. One of the joint receptors, the receptor dorsal-posterior-lateral (RDPL), can be seen, the other has been removed with the anterior femoral cuticle. The lump receptor has been bracketed and is shown enlarged in the inset. (6, c) Pulling on the flexor tendon (increasing force upwards in the lower traces) with the tibia flexed (flexion upwards in the middle traces) causes spikes in the lateral nerve (upper traces). {d) Pulling on the flexor tendon with the tibia held at about 20 does not cause spikes in the lateral nerve. The inset shows the experimental arrangement. Calibration: vertical 25 g; horizontal (a) 27s *nm (inset 300 /an), (b-d) 400 ms. motorneurones (Fig. 176, c). The EPSPs may be a resistance reflex evoked by stimulating the axons of the extension-sensitive joint receptors. Repetitive electrical stimulation causes a hyperpolarization of a flexor motomeurone, although individual stimuli are followed by a depolarization (Fig. 17 d). This suggests that the excitation consists of a non-labile circuit, while the inhibition is mediated by interneurones which only transmit when the locust is aroused. If the lateral nerve is cut in the femur of an otherwise intact locust it becomes extremely difficult, but not impossible, to elicit a jump or kick. The tibia is still 16 EXB 66

22 W. J. HEITLER AND M. BURROWS (6) (i) (6) (ii) (c) (i) (c) (ii) N5 Fig. 17. Response of the AFFlTi to units in the lateral nerve, (a) Pulling on theflexortendon with the tibia flexed (which activates the lump receptor) inhibits the AFFlTi (dots, upper trace). An extracellular recording from nerve 5 at its entry to the ganglion is shown in the lower trace. (6) Single pulse stimulation of the lateral nerve can cause EPSPs or (c) IPSPs in AFFlTi. (d) Repetitive stimulation at 20 Hz causes an overall hyperpolarization, but with depolarizations following each stimulus pulse. Calibration: vertical (a) 10 mv, (b, c) a.5 mv, (d) s mv; horizontal (a, d) 200 ms, (b) 80 ms, (c) 160 ms. Metathoracic ganglion Sense organs Mechanical system Flexor muscle Lump receptor j -Tension Brunner"s organ Joint receptor [-«- Position-»Movement- Joint -Load Unknown Cuticle strain Subgenual organ] -Tension Slow extensor x g Inhibitory synapse C"^Excitatory synapse Fig. 18. A summary of the neural circuits described in this paper. Dotted lines indicate the probable existence of interneurones.

23 Neural circuits of the locust jump 239 flexed as normal, and there can be co-contraction and distortion of the femoral-tibial joint, but this is not usually followed by tibial extension. The partially deafferented behaviour resembles an abortive kick and might be due to the same cause, i.e. the absence of the trigger activity in flexor motorneurones. It is tempting to suppose that the lump receptor is one of the sense organs that excite the interneurones responsible for the trigger inhibition, and so increase the probability of a successful kick. DISCUSSION This paper has described a variety of neural circuits which may be involved in producing the locust kick and jump (Fig. 18). Can the described pathways explain the sequence of movements that occurs in a jump or kick? The initial response to a tactile stimulus likely to evoke a jump is tibial flexion. As the tibia moves into the flexed position the fast extensor tibiae motorneurone may also spike. It is already excited by the tactile stimulation, while resistance reflexes in response to the flexion movement excite it further (Burrows & Horridge, 1974), and Brunner's organ excites it when full flexion is achieved. A fast extensor spike when the tibia is flexed initiates a series of events which result in flexor-extensor co-contraction. First, it excites the slow extensor tibiae by a short latency central circuit and thus ensures an increase in the tension of the extensor muscle. Secondly, it excites the flexor motorneurones by a central circuit. This ensures that tension in the flexor muscle remains high, and the tibia remains flexed. Thirdly, tension in the extensor muscle resulting from the fast and slow motor spikes evokes an excitatory reflex, mediated by unknown receptors that further depolarizes the flexor motorneurones. Fourth, the tension evokes an excitatory reflex mediated by the subgenual organ that excites the fast extensor motorneurone causing a second spike and a consequent increase in extensor muscle tension. The same sequence of events then recurs. As tension develops in the extensor muscle another re-excitatory reflex onto the fast extensor is initiated by strain receptors in the distal femoral cuticle. These circuits combine to produce a sequence of spikes in both flexor and extensor motorneurones. The central circuits decrement rapidly at these high frequencies of spikes so that in the later stages the co-contraction is thought to be maintained largely by the peripheral reflexes. As the tibia extends the fast extensor is inhibited, and this is in part due to extension-sensitive receptors in the suspensory ligaments of the joint. The co-contraction stage of the motor programme can be explained by the circuits described here, given a sufficient state of arousal in the locust or perhaps if the experiments were to be performed at a higher temperature. All experiments were performed at C which is well below the temperature at which locusts are most active. The crucial step following the initial flexion is the first extensor spike. This intiates all the circuits which we have identified as underlying the co-contraction. Experimentally inducing a fast extensor spike by electrical stimulation with the tibia flexed, however, will not usually produce a prolonged co-contraction in the quiescent locust although the circuits described here all operate under these conditions. This is because the reexcitation of the FETi mediated by the subgenual organ is usually subthreshold in the quiescent locust. Central arousal is thus essential for the initiation of co-contraction. After tibialflexionresulting from a natural arousal stimulus, a single fast extensor spike 16-2

24 240 W. J. HEITLER AND M. BURROWS will often initiate co-contraction. There may be other parallel circuits not so far identified which are also involved in producing the co-contraction, but they need not be implicated. The same cannot be said for the third stage, the trigger inhibition of the flexor motomeurones. The 'lump receptor' may be involved in this stage because it is only stimulated during the co-contraction, and it can inhibit flexor motomeurones. It is, however, not the only circuit responsible for the trigger activity, since ablation of the lump receptor reduces the likelihood of the occurrence of the trigger activity, but does not eliminate it. Another sense organ which has been implicated in this stage of the programme is the chordotonal organ (Bassler, 1968). Cutting the tendon of the chordotonal organ produces a high probability of a behaviour similar to an abortive kick, i.e. co-contraction which is not terminated by the trigger activity. Cutting the chordotonal organ tendon will, however, cause it to respond as if the tibia were permanently extended. This would result in contradictory sensory input to the central nervous system which could inhibit the trigger activity. This result does not necessarily imply that the chordotonal organ itself is involved in initiating the trigger activity. There is evidence to suggest that the trigger activity is not a direct result of a peripherally initiated reflex. High frequency antidromic stimulation of the fast extensor motomeurone with the tibia flexed produces considerable extensor muscle tension and via, the circuits described, considerable flexor muscle tension, but it has never been seen to produce the trigger activity. Kicks by both legs together can occur in which the synchronization of the trigger activity bears no relation to synchronization of the muscle tension on the two sides (Heitler & Burrows, 1976). These observations suggest that while peripheral reflexes may be important in bringing the trigger interneurones into a state of readiness, the occurrence of the trigger activity in a normal kick is dependent upon a central excitation of the trigger system. In so far as this central excitation constitutes the ultimate decision in a jump or kick, it is of considerable interest, but it can only be investigated further by identification and recording from the interneurones of the trigger system itself. This work has been supported by an S.R.C. studentship to W. J. Heitler, a NufBeld Foundation grant to M. Burrows, and in part by USPHS grant number 5 ROI NS to C. H. F. Rowell. REFERENCES BASSLER, U. (1968). Zur Steuerung des Springens bei der Wanderheuschreke Schistocerca gregaria. Kybernetik 4, 112. BENNET-CLARK, H. C. (197s). The energetics of locust jumping. J. exp. Biol. 63, BROWN, R. H. J. (1067). The mechanism of locust jumping. Nature, Land. 314, 939. BURROWS, M. & HORRIDGE, G. A. (1974). The organization of inputs to motorneurones of the locust metathoracic leg. Phil. Tram. R. Soc. Land. B 869, COILLOT, J. P. & BOISTEL, J. (1968). Localisation et description de recepteure a retirement au niveau de l'articulation tibic-femorale de la pane sauteuse du criquet Schistocerca gregaria. J. Insect Pkytiol. 14, COILLOT, J. P. & BOISTEL, J. (1969). tude de l'activite electrique propagee de recepteurs a retirement de la patte metathoracique du criquet, Schistocerca gregaria. J. Insect Phytiol. 15, HEITLHR, W. J. (1974). The locust jump: specialisations of the femoral-tibial joint. J. comp. Phj/siol. 89,

25 Neural circuits of the locust jump 241 HETLER, W. J. & BURROWS, M. (1977). The locust jump. I. The motor programme. J. exp. Biol. 66, HOYLE, G. & BURHOWS, M. (1973). Neural mechanisms underlying behavior in the locust SMstocerca gregaria. I. Physiology of identified motorneurons in the metathoracic ganglion. J. Neurobiol. 4, JANNONE, G. (1940). Studio morfologio, anatomico, e istologico del Dociostaurus maroceanus (Thunb) nelle sue fasi transiens, congregans, gregaria, e solitaria. Boll. Lab. Ent. Agr. Portia 4, JOLY, P. (1951). Role de l'organe de Brunner chez Locutta imgratoria. C.r. Sianc. Soc. Biol., Paris 145, UVAROV, B. (1966). Grasshoppers and Locusts: a Handbook of General Acridology. Cambridge University Press. WILSON, D. M. & LARIMER, J. L. (1968). The catch property of ordinary muscle. Proc. not. Acad. Set. U.S^l. 6I,