INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

During locomotion, sensory signals modify and control the ongoing motor program of each appendage. These signals thereby time and shape the motor output of each phase of the locomotor cycle to actual requirements. One of the mechanisms contributing to the generation of functional motor output in walking is the reinforcement of movement by sensory signals, which reflects the reversal of postural reflexes at rest. The neural basis of reflex reversal in the walking systems of vertebrates and invertebrates has been investigated in considerable detail (see reviews in Bässler and Büschges 1998; Büschges and El Manira 1998; Clarac et al. 2000; Pearson 1993). In the stick insect and locust, for example, flexion signals from the femur-tibia (FT) joint reinforce flexor activity (Bässler 1976, 1988) when flexion signals from the FT joint terminate extensor activity and initiate flexor activity in restrained preparations when the walking system is "active," i.e., in the locomotor state. In general, is it now clear that reflex reversal is generated by phase- and state-dependent modulation of sensorimotor pathways that control both posture and movement of the limb (Driesang and Büschges 1996; Le Ray and Cattaert 1999). In insects, it is still unclear whether the motor activity during reflex reversal is caused by the contribution of central rhythm-generating networks capable of generating alternating activity in antagonistic motoneuron pools of a leg joint, as was shown in crayfish (El Manira et al. 1991). The only hint toward such a mechanism derives from simulation studies (Bässler and Koch 1989).

In the present investigation, we use a preparation of the locust mesothoracic segment that expresses long-lasting rhythmic activity in leg motoneuron pools. Experiments with this preparation provide the first evidence that, in the locust walking system, motor output during reflex reversal is generated via central rhythm-generating networks.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experiments were carried out on adult locusts, Locusta migratoria, obtained from a crowded colony maintained at our institute. Experiments were performed at room temperature (20-22°C). After both pairs of their wings, forelegs, and hind legs were removed, the animals were fixed ventral side up on a foam platform; the femurs of both middle legs were fixed, with dental cement, perpendicular to the animal. Minutin pins were used to fix the FT joint of each middle leg to the platform at a 90° angle. The meso- and metathoracic ganglia were exposed by removing the ventral cuticle of the thorax as well as the fatty tissue and air sacks from the mesothoracic cavity. The sensory innervation of the leg was usually left intact. In experiments testing the influence of sequential deafferentation (Fig. 1G), the campaniform sensillae on the trochanter and femur were removed prior to the experiment by using an insect pin to destroy the sensory field on the cuticle. For intracellular recordings from leg motoneurons, the mesothoracic ganglion was stabilized on a ganglion holder. Activity of the muscles controlling the FT joint was monitored by inserting low-resistance electromyograph (EMG) wires (50 µm copper wire). In these recordings, muscle potentials from both antagonists could be distinguished based on their activity in phase with flexion or extension movements of the tibia. This was achieved by using tactile stimuli (a small paintbrush to the thorax or leg) to activate the animal. In some experiments, fast extensor motoneuron activity was monitored by an extracellular hook electrode on lateral nerve N3b (Campbell 1961). Activity of the motoneurons innervating the extensor tibiae muscle was also recorded intracellularly from their somata in the mesothoracic ganglion. This was done with an SEC-10 l amplifier (NPI Electronics, Tamm, Germany) and glass microelectrodes (Science Products) filled with 2 M KAc/0.05 M KCl (15-20 M). Leg motoneurons were identified by a 1:1 correlation of their action potentials in the intracellular recording and the EMG recording from the tibial muscles (Wolf 1992). To stimulate the femoral chordotonal organ (fCO), the femur was opened dorsally and a clamp was used to attach the apodeme of the fCO to an electromechanical stimulator. Ramp-and-hold stimuli with an amplitude of 560 µm were applied to the fCO. This mimicked a 60° movement of the tibia from its center position (Field and Pflüger 1989). Intracellular recordings, EMG recordings, injected current, and the position signal of the fCO stimulator were stored on a 16-channel tape recorder (Racal V-Store 16). For evaluation, Spike 2 software (Cambridge Electronic Design, CED) was used to analyze the data offline on a Pentium II personal computer. Analog/Digital conversion was performed with a CED 1401plus interface. Statistical analysis was performed with PlotIt for Windows and with Microsoft Excel.

View larger version (39K): [in this window] [in a new window]   Fig. 1. Rhythmic motor patterns generated in motoneurons and muscles supplying the femur-tibia (FT) joint in the mesothoracic segment after the anterior and posterior connectives were cut. A: electromyographic (EMG) recording of flexor and extensor tibiae activity during rhythmicity showing alternating bursts of activity in both motoneuron pools. B: simultaneous EMG recording from tibial muscles and intracellular recording from the slow extensor tibiae motoneuron (SETi) in the isolated mesothoracic segment. SETi is strongly hyperpolarized during flexor bursts. C: simultaneous EMG recording from tibial muscles and intracellular recording from a slow flexor motoneuron in the isolated mesothoracic segment. Note that the flexor motoneuron is strongly hyperpolarized during extensor bursts. D: distribution of burst patterns of extensor and flexor motoneuron pools generated by the isolated mesothoracic segment (E, extensor burst; F, flexor burst). Each bar represents the summed probability of single bursts (left bar), burst doublets (middle bar), and burst triplets (right bar). E: plot of extensor burst duration (open circles, solid regression line) and flexor burst duration (closed circles, dashed regression line) vs. period of occurrence of bursts (regression coefficient 0.001 for extensor and 0.0005 for flexor; n = 58). F: plot of flexor burst duration vs. extensor burst duration (regression coefficient 0.0228; n = 58). G: plot of the average relative time (SD) of recurrence of fast extensor tibiae motoneuron (FETi) bursts as a function of state of deafferentation ("deaff. step") for two preparations: 1, isolated mesothoracic segment with N2 and N6 cut; 2, deafferentation of the contralateral side; 3, N1 cut; 4, N3 cut; 5, N4 cut; 6, N5 cut at the level of the tarsus; 7, N5 cut at the level of the tibia.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

After the anterior and posterior connectives of the locust mesothoracic ganglion were cut, episodes of patterned activity were generated in the motoneuron pools of flexor and extensor tibiae in 57% of the preparations [N (number of experiments) = 96; Fig. 1, A-C]. In a majority of cases [61.5%; N = 13; n (sample size) = 587], these episodes consisted of doublets of one flexor and one extensor burst, in either order (Fig. 1D). Intracellular recordings from extensor and flexor motoneurons revealed that strong modulations in membrane potential underlie the observed bursts in motor activity. When bursts in tibial motoneurons were generated in doublets or triplets, there was no overlap in the motor activity of the antagonists between their bursts (Fig. 1, B and C). In the periods between burst generation, slow motoneurons of both pools could be tonically active, as shown for a slow extensor tibiae motoneuron (SETi) in Fig. 1B. The durations of bursts in extensor and flexor motoneurons showed considerable variability. For example, extensor burst duration could vary from 0.8 ± 0.42 to 3.7 ± 2.88 s (N = 7; n 5-59) in different preparations. The same was true for flexor burst durations, which ranged from 1.8 ± 0.80 to 3.6 ± 1.94 s (N = 7; n 6-59). In general, there was no correlation between burst duration and the period between bursts (Fig. 1E), as tested in seven animals. In addition, there was no correlation between the burst durations of the antagonists (Fig. 1F). The burst period was variable, ranging, for example, from 2.0 to 23.2 s (n = 42) in a given preparation and averaging 6.5 ± 3.27 to 23.8 ± 13.13 s in seven different preparations. Similar alternating activity patterns were generated by motoneurons supplying the coxal, trochanteral, and tarsal muscles (Knop and Büschges, unpublished observations).

The generation of burst patterns in the motoneuron pools of the FT joint was found to persist following deafferentation of the mesothoracic ganglion in 37% (N = 27) of these preparations (Figs. 1G and 3A), providing evidence for the central origin of these motor patterns. However, deafferentation of the mesothoracic ganglion increased the interburst interval (Fig. 1G), indicating that sensory signals influence the activity of the central neural networks underlying the generation of burst activity in the motoneurons. Sensory signals applied during bursts of motoneuron activity from the fCO strongly affected motor activity (Fig. 2). Elongation stimuli at the fCO applied later than ~300 ms after the beginning of fast extensor tibiae motoneuron (FETi) burst activity prematurely terminated extensor activity and initiated a burst of flexor motoneuron activity (Fig. 2, Ai, Aiii, B, and C) that always outlasted fCO elongation (Fig. 2, A and B). When fCO elongation was applied earlier than 300 ms after the start of the extensor burst, we observed either a brief inhibition of extensor activity or no response (Fig. 2, B and C). These results show that flexion signals from the FT joint induced flexor activity that would assist joint flexion in the intact leg. The motor output generated on fCO elongation therefore represents a reflex reversal when compared with the negative feedback response of tibial motoneurons in the resting animal (Field and Burrows 1982). Extension signals from the FT joint, which are mimicked by relaxation stimuli at the fCO applied during ongoing extensor bursts, did not induce a switch in activity between both motoneuron pools. They produced no change in motor activity or only a subtle increase in ongoing extensor activity (not shown). However, when relaxation (joint extension) signals were applied during a flexor burst, premature cessation of flexor activity and initiation of extensor activity was induced (Fig. 2, Aii and Aiii). During phases of tonic motor activity, stimulation of the fCO had little if any influence on tibial motoneuron activity and only slightly excited the motoneurons of both pools (not shown).

View larger version (39K): [in this window] [in a new window]   Fig. 2. Influence of movement signals from the femoral chordotonal organ (fCO) on activity patterns of tibial motoneurons. Ai: influence of fCO elongation on tibial motoneuron activity when delivered during extensor activity. An intracellular recording from SETi, together with an EMG recording of tibial muscle activity, is shown. Aii: influence of fCO relaxation signals during flexor activity. Aiii: burst pattern without stimulation. Arrow in fCO trace: direction of simulated FT joint movement. B: influence of fCO elongation signals on extensor burst duration as a function of delay between the start of the burst and the start of the elongation stimulus: with a delay of 70 ms (top traces), a short inhibition of extensor activity (open arrowhead) was observed; with a delay of 400 ms (middle traces), extensor activity was terminated prematurely and flexor activity was initiated. Slanted arrows: switch from extensor to flexor activity. Arrows in fCO traces: simulated FT joint movement. C: influence of fCO elongation on the activity of extensor motoneurons plotted as a function of delay between burst onset and stimulus onset [top graph: probability of no change in extensor motoneuron activity; middle graph: probability of short inhibition in extensor activity (see B, top); bottom graph: probability for a switch from extensor to flexor activity (see Ai and B, bottom)].

Flexion and extension signals from the fCO had similar effects on tibial motoneuron activity when the mesothoracic ganglion was otherwise deafferented. In this situation, flexion signals terminated extensor activity and induced flexor activity (Fig. 3A), with extensor burst duration correlated significantly with the time of fCO stimulus (Fig. 3B). In more than 70% of the stimuli, the termination of the burst in FETi occurred within 200 ms after the start of fCO elongation.

View larger version (21K): [in this window] [in a new window]   Fig. 3. Influence of fCO stimuli on the activity of tibial motoneurons when the mesothoracic ganglion is deafferented except for the fCO. A: extracellular recording of flexor motoneuron activity (top trace) and the FETi (nerve N3b, middle trace) together with fCO position (bottom trace). fCO elongation terminates FETi activity and initiates a flexor burst that outlasts the stimulus. The 4 FETi bursts were generated subsequently. Arrow in the stimulus trace: direction of the simulated joint movement. B: plot of burst duration in FETi as a function of the delay between extensor burst onset and stimulus onset.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The isolated mesothoracic segment and the isolated and deafferented mesothoracic ganglion of the locust are capable of generating episodes of patterned activity in tibial motoneuron pools. These episodes consist of alternating activity in the antagonistic motoneuron pools of the FT joint with no overlap. Basic characteristics of the generated motor output bear some similarities with a previous study of the pharmacologically treated metathoracic segment (Ryckebusch and Laurent 1993). Exceeding this previous study, our experiments reveal for the first time that the locust leg muscle control system in the isolated mesothoracic ganglion can be in a rhythmic state, driving leg motoneuron pools in episodes of alternating bursts. Furthermore, phasic proprioceptive signals influence the generated motor activity. Both flexion and extension signals from the FT joint are able to induce transitions between activity in motoneuron pools that resemble a reflex reversal previously observed in the locomotor state of insect walking systems (Bässler 1988, 1992). Flexion signals from the fCO induce a switch from extensor activity to flexor activity so that the generated motor activity would assist the movement signaled by fCO stimulation.

Our results indicate that, in the isolated mesothoracic ganglion of Locusta, proprioceptive signals can influence central rhythm-generating network (CRG) activity and thus contribute to reflex reversal. Similar activation of flexor motoneurons in nonresistance patterns has also been found in responses to simple joint movements (Siegler 1981) or mechanical stimulation of the hind leg fCO (Burrows et al. 1988). In addition, chordotonal organ inputs produced nonresistance responses during searching movements and after the connectives between the supra- and subesophageal ganglion were cut (Zill 1985). In the crayfish coxo-basopodite joint, reflex reversal is mediated via an interneuronal pathway that is modulated by the activity of a CRG during fictive locomotion in a phase-dependent manner (Le Ray and Cattaert 1997). It is particularly interesting that our results are consistent with earlier findings on the leg muscle control systems of crustaceans and vertebrates. In these systems, evidence also was presented that the generation of reflex reversal, and thus reinforcement of movement, results from the contribution of central pattern-generating networks (e.g., El Manira et al. 1991; Pearson and Collins 1993).