INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

During walking, the motor pattern driving the leg muscle system is the result of interactions between central neural networks, feedback signals from sense organs as well as coordinating pathways between the legs. For a variety of vertebrate and invertebrate walking systems, interactions between central and peripheral signals have been described in considerable detail on the operational level (reviews in Bässler and Büschges 1998; Clarac 1991; Grillner 1975, 1985; Pearson 1993; Prochazka 1996; Rossignol et al. 1988). Changes in load, information from position-sensitive sense organs of leg joints, and information on the actual step phase of adjacent legs strongly influence the motor pattern of each individual leg (e.g., Bässler 1977, 1993; Cruse 1985; Foth and Bässler 1985; Pearson and Iles 1973; Wendler 1964). Intrajoint proprioceptive information is used for sculpturing the motor output of the leg muscle system controlling this joint. These sensory signals can entrain motor activity produced by rhythm-generating networks (e.g., Anderson and Grillner 1983; Conway et al. 1987; Sillar et al. 1986), and they can induce phase- or state-dependent reflex reversals in motor control networks (e.g., Bässler 1976; DiCaprio and Clarac 1981; Pearson and Collins 1993; Skorupski and Sillar 1986; recent summary in Büschges and El Manira 1998). Furthermore generation of the functional walking motor pattern in the multijointed limb also relies on coordination between movements of adjacent leg joints. The knowledge, however, as yet mostly concerns the operational level in understanding interjoint coordination. Evidence exists for both central and peripheral influences on coordination between adjacent joints (e.g., Angel et al. 1996; Bässler 1993; Graham and Bässler 1981; Grillner and Zangger 1979; Robertson et al. 1985). Investigations of the neuronal mechanisms underlying interjoint coordination only recently have been initiated (El Manira et al. 1991). The question arises, particularly in walking systems in which individual joint oscillators have been shown (stick insect: Büschges et al. 1995; mudpuppy: Cheng et al. 1998), how activities between adjacent leg joints are coordinated and to what extent sensory signals play a role in interjoint coordination.

We set out to investigate what role sensory signals from one leg joint play in patterning motoneuronal activity of the adjacent leg joint in an insect walking system. We chose to investigate this question for sensory signals provided by the fCO in the stick insect middle leg muscle control system for the following reasons: the fCO can be stimulated very easily with various defined stimulus protocols that allow a separation between the influence of position and movement signals (Bässler 1983). In addition, the behavioral state of the animal can be controlled by monitoring the motor response that sensory signals from the fCO induce in intrajoint control networks of tibial motoneuron pools.

A previous investigation described interjoint reflex action between the fCO and trochanteral motoneurons in the stick insect middle leg in the inactive, i.e., resting animal (Hess and Büschges 1997). Elongation of the fCO, i.e., flexion of the FT joint, activates trochanteral levator motoneurons and inhibits trochanteral depressor motoneurons. Relaxation of the fCO, i.e., extension of the FT joint, induces activation of depressor motoneurons and in part coactivation of levator motoneurons. This interjoint reflex action contributes to posture control of the resting, i.e., standing animal. When the stick insect turns active (active behavioral state), the joint control networks switch from the posture-control mode to the movement-control mode acting during locomotor movements (Bässler 1976, 1988; summary in Bässler and Büschges 1998). When the locomotor system is active, proprioceptive signals from the FT joint are used for controlling tibial movements by reinforcement of movement over a broad range of movement velocities. Such reinforcement of movement becomes obvious from the exhibition of a state-dependent reflex reversal in the FT-joint control network toward joint flexion (Bässler 1988). The question arises what role interjoint influences of proprioceptive signals play in the active behavioral state. Initial evidence presented showed that in the active behavioral state elongation signals from the fCO still induce activation of levator motoneurons, however, in a burst-like fashion (Hess and Büschges 1997).

The present investigation addresses in more detail the role of position and movement signals from the fCO in sculpturing motoneuronal activity of the adjacent CT joint in the active behavioral state of the stick insect. We will show that sensory signals from the fCO can take part in patterning motor activity in trochanteral motoneurons by releasing transitions of activity between trochanteral motoneuron pools. The induction of these transitions did not depend on the actual stimulus parameters. The induction rather depended on the direction of movement signaled by fCO afferents. For the first time, we will provide evidence that proprioceptive signals from an insect leg joint affect central rhythm generating networks driving an adjacent leg joint.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experiments were performed on adult female stick insects of the species Carausius morosus raised in the animal facilities of the University of Kaiserslautern under daylight conditions and at room temperature (20-22°C).

The animals were mounted dorsal side up on foam platform with the forelegs and hindlegs being fixed aside the longitudinal axis of the body (for details, see Hess and Büschges 1997). The proximal leg segments of the left middle leg, i.e., coxa, trochanter, and femur (in the stick insect, trochanter and femur are fused to one segment) were fixed with dental cement (Protemp, ESPE) onto a foam rim pointing slightly upward at an angle of ~30°. The tibia was extending over the distal margin of the rim and the platform. The FT joint then was fixed with dental cement at an angle of ~150°.

The thorax of the animal was opened by a sagittal cut along the dorsal midline of the meso- and metathorax. The thoracic cavity was filled with stick insect saline (Bässler 1977; Weidler and Diecke 1969). The mesothoracic ganglion was placed on a wax coated platform and fixed with cactus spines. The activities of motoneurons were recorded extracellularly from the lateral nerves that carry the axons of motoneurons innervating the muscles of the three proximal leg joints with monopolar hook electrodes (Schmitz et al. 1988). The following motor nerves were recorded: C1, carrying the motoneurons of the levator trochanteris; C2, carrying the motoneurons of the depressor trochanteris. Extensor tibiae motoneuron activity was recorded for monitoring the behavioral state of the animal by either an extracellular recording from the extensor nerve F2 in the femur (for details, see Hess and Büschges 1997) or by a recording from nerve nl3, in which the extensor motoneurons leave the ganglion. In recordings from nerve nl3 the extensor motoneurons have the largest spikes (see also Büschges et al. 1995).

Animals were activated by touching their abdomen or head with a small paintbrush. During and after tactile stimulation, the experimental animal moved its unrestrained tarsi, its abdomen, and antennae. Furthermore fast motoneurons recorded in the leg nerves were activated in a burst-like fashion (see also Bässler 1983, 1988)

Preparation for stimulation of the fCO during pharmacologically induced rhythmic motor activity in the otherwise isolated mesothoracic ganglion

In these experiments, the influence of sensory signals arising from the fCO on centrally generated rhythmic motor activity in the trochanteral motoneuron pools was investigated. In this preparation, except for the fCO all other afferent input to the isolated mesothoracic ganglion was removed. Before the experiment, the trochanteral campaniform sensillae of the left middle leg were destroyed by drilling a sharpened tip of an insect needle through the cuticle at their location (Hofmann and Bässler 1982). Experimental animals were fixed to the foam platform (Hess and Büschges 1997). The mesothoracic ganglion was isolated from the thoracic nerve cord by cutting its anterior and posterior connectives. Partial deafferentation of the ganglion was achieved by cutting the following lateral nerves (labeled according to Marquardt 1940) close to the ganglion: nervus anterior (n_a); nervus posterior (n_p), nervus unparis [(n_up); this nerve leaves the ganglion on the dorsal surface] and the nervi laterali 2-5 (nl_2-5). After positioning an extracellular hook electrode on coxal nerves C1 and C2, action potential propagation in these nerves toward the muscles was terminated by squeezing these motor nerves with fine forceps distal to the recording site. The apodeme of the fCO was exposed according to the described procedures and fixed in the clamp of the stimulus device. Finally, the main leg nerve nervus cruris (n_cr) was cut at its entrance to the femur. Stick insect saline was applied by drops into the body cavity.

Stock solutions of 10² M pilocarpine (Sigma Chemicals) in saline were prepared in advance and were diluted with saline to their final concentrations prior to application. Application of the drugs was performed by removing the saline from the thorax of the experimental animal and replacing it by the drug solution under investigation with final concentrations in the range of 1 × 10⁴ M to 1 × 10³ M (Büschges et al. 1995).

Extracellular and intracellular recordings from motoneurons

Preparations of the mesothoracic ganglion for intracellular recordings were performed according to the established procedures (Büschges 1990). Intracellular recordings were performed in the ipsilateral neuropil region of the mesothoracic ganglion that is known to contain the arborizations of fCO afferents (Schmitz et al. 1991) as well as the arborizations of leg motoneurons (Storrer et al. 1986). Recordings were made using thin-walled glass microelectrodes (wpi), filled with a solution of 2 M KAc/0.05 M KCl (electrode resistance 15-20 M).

Mechanical stimulation of the fCO

For mechanical stimulation of fCO afferents, the femur of the left middle leg was prepared. The receptor apodeme was exposed and fixed in the clamp of a stimulus device; it then was cut distally and was moved over a range of positions corresponding to femur tibia-angles between 30 and 150°. Ramp-and-hold stimuli with different stimulus velocities and holding times were tested. Flexion movements of the tibia were mimicked by elongation of the fCO in most cases from a starting position of ~120° with an amplitude of 300 µm (corresponding to a joint movement of 60°) (Weiland and Koch 1987). Stimulus velocities were in the range of 20-1,200°/s. Movement velocities in this range have been reported to be generated by the stick insect leg muscle system (Bartling 1993; Bässler 1983). Acceleration values of the stimuli were not controlled independently.

Data storage and statistics

All data were stored on an eight-channel DAT-recorder (Biologic DTR-1802, sample rate 12 kHz) and analyzed off-line either on a personal computer or displayed on a printer for further evaluation (Yokogawa OR-2100). Peristimulus time (PST) histograms were generated by using the evaluation software Spike2 (Cambridge Electronics, Science Products, ver. 3.14).

Results are given either as single values for individual experiments or as mean values ± SD of the pooled data of the experiments. Maximal frequency of motoneuron activity was evaluated by counting the spikes elicited in the motor nerve classes (BINs of 200 ms width) by using a window discriminator. The resulting value was then transformed to spikes/s. Means were compared using a t-test (depending on the sample size) (Dixon and Massey 1969; Sachs 1997). Means were regarded as significantly different at P < 0.05. N = number of experiments; n = sample size.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sensory signals the fCO induce transitions of activity between motoneuron pools of the coxa-trochanteral joint

In the active behavioral state of the stick insect, the fast and semifast levator and fast depressor motoneuron of the CT joint were mostly active in clear antiphase to each other in a burst-like fashion (Fig. 1; for definition of the "active" behavioral state, see also INTRODUCTION and METHODS). Only slow motoneurons of both motoneuron pools sometimes were coactivated (see Fig. 1A). Sensory signals from the fCO influence the activity of trochanteral motoneuron pools in the active behavioral state (Hess and Büschges 1997, their Fig. 9). Elongation of the fCO resulted as a rule in a strong, i.e., burst-like activation of the levator motoneurons. The spike frequency of spontaneous active slow levator motoneurons increased rapidly, and strong activity in semifast and fast motoneurons could be elicited. Simultaneously, activity of slow and fast trochanteral depressor motoneurons was terminated (Fig. 1, A and B) (also Hess and Büschges 1997). In contrast, in the resting, inactive animal elongation of the fCO are known to induce only subtle excitation in levator motoneurons with slow and semifast levator motoneurons being activated suprathresholdly.

View larger version (37K): [in this window] [in a new window]   Fig. 1. Influence of sensory signals from the femoral chordotonal organ (fCO) on the activity of trochanteral motoneurons. A: comparison of the influence of sensory signals from the fCO (Pos) on the levator trochanteris (Lev) and the depressor trochanteris motoneurons (Dep) in the inactive and active animal (S, slow; sF, semifast; F, fast levator trochanteris motoneurons; SDTr, slow depressor; FDTr, fast depressor trochanteris motoneuron). Please note that sF and F levator and FDTr are mostly active in antiphase. Despite the overall variability of alternating motoneuronal activity, fCO elongation is associated mostly with a transition from depressor to levator activity (), whereas the reverse transition is initiated by relaxation signals from the fCO (). B: expanded presentation of one transition occurring with fCO elongation (i) and fCO relaxation (ii). C: average peristimulus time (PST) histograms of levator motoneuron activity for 1 preparation. Compared is the mean levator activity when a Dep/Lev-transition was elicited (n = 13) and when no Dep/Lev-transition is initiated (n = 24). D: in some experiments, trochanteral motor activity could be almost completely patterned by continual ramp-and-hold stimulation of the fCO.

In the active behavioral state, burst-like activation of trochanteral levator motoneurons, and simultaneous inactivation of depressor motoneurons (termed "transition"), could be elicited almost routinely by fCO elongation stimuli when semifast and fast levator motoneurons were not active before stimulus onset. This was true despite the well-known variability in rhythmic leg motoneuron activity of the active stick insect (Fig. 1A) (see also Bässler and Wegner 1983). In case of levator motoneurons being active at the time of fCO elongation stimulus onset, levator activity was enhanced. A quantitative evaluation of 25 experiments (ramp-and-hold stimuli applied to the fCO from 120 to 60°; stimulus velocity 120°/s) showed that on average ~67.8% of the elongation stimuli, in which the levator motoneurons were not active before stimulus onset (n = 977), induced a burst-like activation of levator motoneurons. Depressor motoneuron activity was terminated when present (Fig. 1). The remaining stimuli, in which semifast and fast levator motoneurons were not active before stimulus onset did not induce a specific motor response (32.2%). This is shown in Fig. 1C by PST histograms of levator activity in one preparation.

The reverse motor response was elicited by relaxation of the fCO when levator motoneurons were active during onset of a relaxation stimulus. In 71.1% of the presentations (N = 25, n = 863), elongation stimuli terminated levator motoneuron activity and activated depressor motoneuron activity (Fig. 1, A and D). The aforementioned influence of fCO signals on the activity of trochanteral motoneurons in the active behavioral state persisted after deafferentation of the proximal leg joints (except from the fCO; N = 5; not shown).

In 45% of the experiments, i.e., in 9 of 20, ramp-and-hold-stimulation of the fCO could completely pattern the activity in trochanteral motoneurons for longer sequences of the stimulus protocol. Elongation stimuli elicited levator motoneuron activity and terminated depressor motoneuron activity. Relaxation stimuli had the reverse influence (Fig. 1D). This phenomenon strongly resembled the "entrainment" of rhythm-generating network by proprioceptive signals that was shown previously in a variety of locomotor systems (e.g., Andersson and Grillner 1983; Grillner and Wallén 1977; Reye and Pearson 1988; Sillar et al. 1986).

The latency after which levator and depressor activity was affected in the active animal by elongation and relaxation of the fCO did exhibit considerable variability. Neither initiation of levator motoneuron activity by fCO elongation (Fig. 2) nor activation of depressor activity by fCO relaxation occurred at a fixed latency after the onset of the stimulus. Both transitions could take place sometime throughout the ramp movement of the fCO. As such, the latency for activation of levator motoneurons by fCO elongation was generally much longer and more variable than the latency of the response in the inactive behavioral state. For one preparation the mean latency of activation in levator motoneurons in the inactive state was 12.0 ± 7.9 ms (n = 9), whereas the mean latency was 436.0 ± 119.0 ms (n = 32) in the same animal when being active.

View larger version (20K): [in this window] [in a new window]   Fig. 2. Comparison of the latency for suprathreshold levator motoneuron activation in the inactive and active animal. Data from 19 experiments are given. Each point represents the latency of levator motoneuron activation during fCO elongation (n from 6 to 72). , occurrence of the 1st action potential in the inactive animal; , occurrence of the first action potential in the active animal.

As shown earlier (Fig. 1), movement signals from the fCO did induce specific transitions in activity in trochanteral motoneurons. A common feature of these transitions was that there was no overlap between the activity of semifast and fast motoneurons of levator and depressor motoneurons. Termination of activity in one motoneuron pool was complete before the antagonist turned active. The mean time for the switch in activity between depressor and levator motoneurons induced by fCO elongation was 32.8 ± 21.53 ms (N = 3, n = 122). The mean latency for the reverse switch between levator and depressor motoneuron activity was 25.1 ± 18.09 ms (N = 3, n = 111) and not significantly different.

The interjoint influence of sensory signals from the fCO was even more obvious when evaluating the general occurrence of transitions between activity in levator and depressor motoneuron pools in the active animal during ongoing ramp-and-hold stimulation of the fCO. For evaluation (Fig. 3A), similar ramp-and-hold cycles from each experiment were divided in eight BINs of constant length (500 ms), and the transitions in motoneuron activity were counted. In N = 20 animals, n = 820 transitions from depressor to levator motoneuron activity were evaluated during continual ramp-and-hold stimulation of the fCO. Transitions occurred mostly (48.2%) during fCO elongation stimuli, i.e., within the first BIN (Fig. 3A) compared with a low level of occurrence in the BINs of the remaining stimulus cycle (5-10.6%). In the same animals, occurrence of the reverse motor response, i.e., a transition from levator to depressor motoneuron activity, was evaluated as well (N = 20, n = 761). This transition showed maximal occurrence during relaxation of the fCO (45.2%; 5th BIN). When evaluating specifically the termination of those levator bursts that had been initiated by fCO elongation (N = 20; n = 395), i.e., the end of only those bursts that were counted in the first BIN of Fig. 3A, it became obvious that in the majority these bursts (44.1%, Fig. 3B) were terminated by the next relaxation stimulus at the fCO.

View larger version (19K): [in this window] [in a new window]   Fig. 3. Frequency plot of the occurrence of transitions in activity between trochanteral motoneurons in the active animal during continual stimulation of the fCO. Each ramp-and-hold cycle was divided in 8 classes (BINs) of constant length (500 ms). Insets: exemplification of the actual transition evaluated (taken from the same experiment). A: frequency of transitions from depressor to levator motoneuron activity (, N = 20, n = 820) and frequency of transition from levator to depressor motoneuron activity ( N = 20, n = 761). B: frequency histogram for transitions from levator to depressor activity. In this histogram, only those transitions are counted for which ongoing levator activity has been elicited by a previous elongation stimulus at the fCO (1st filled BIN in A).

Intracellular recordings from trochanteral motoneurons revealed that in the active state, there was a marked enhancement of synaptic inputs induced by sensory signals from the fCO compared with the inactive state (Fig. 4). Elongation of fCO led to an increased depolarization in levator motoneurons and an increased hyperpolarization in the slow depressor motoneuron (N = 3) as compared with the inactive state (Fig. 4, A and B). On some occasions we observed a slight depolarization of depressor motoneurons that preceded their stimulus-induced hyperpolarization (see Fig. 4B). Relaxation of the fCO led to an increased depolarization in depressor motoneurons compared with the inactive animal (Fig. 4B). Termination of levator motoneuronal activity during relaxation signals from the fCO was the result of the membrane potential returning to the level before fCO stimulation without the occurrence of a marked hyperpolarization. It is not clear as yet as to the repolarization in levator motoneurons results from an inhibitory influence or from discontinuation of an excitatory influence (N = 3; Fig. 4A).

View larger version (31K): [in this window] [in a new window]   Fig. 4. Intracellular recordings from levator and depressor trochanteris motoneurons in the active animal during transition of motoneuronal activity (2 examples, top) and in the inactive animal (1 example, 2nd trace from bottom). A: semifast levator motoneuron (Lev MN). B: slow depressor trochanteris motoneuron (SDTr). Please note the marked depolarization occurring in the active animal for the levator motoneuron with fCO elongation, and for SDTr with fCO relaxation. - - -, resting potential before stimulus onset. Action potentials are clipped.

Correlation between intrajoint (FT joint) and interjoint (CT joint) influences of sensory signals from the fCO in the active animal

From these observations, it is clear that in the active behavioral state signals from the fCO induce activity transitions the trochanteral motoneurons. The question arose as to whether the observed motoneuronal response to fCO stimulation was linked to the motoneuronal activity of the FT joint. FT motoneurons are affected strongly by sensory signals from the fCO in the inactive and active behavioral state. In the active state, movement signals from the fCO induce a specific intrajoint motoneuronal response in tibial motoneurons, i.e., a reflex reversal ("active reaction") (Bässler 1976, 1988). Initial data (Hess and Büschges 1997) did not suggest a coupling between the generation of the reflex reversal in the FT joint and the transition in motoneuronal activity induced in the motoneuron pools of the CT joint. This conclusion was supported further by additional experiments that focused on the occurrence of both types of motoneuronal responses in the active state (Fig. 5). Both responses can occur either together or independently from each other. It thus is clear that in the active behavioral state the influence of sensory signals from the fCO on trochanteral motoneurons is not causally linked to the generation of a reflex reversal in the FT-joint leg muscle control system.

View larger version (34K): [in this window] [in a new window]   Fig. 5. Relation between the generation of the reflex reversal on fCO elongation in tibial motoneurons ("active reaction") and the occurrence of the transition in activity between depressor and levator motoneurons of the coxa-trochanteral (CT) joint in the active animal. A: frequency of occurrence of transition in trochanteral motoneuron activity when an active reaction in tibial motoneurons is generated (N = 16, n = 353). B: frequency of occurrence of an active reaction in tibial motoneurons when an activity transition in trochanteral motoneurons is generated (N = 16, n = 344). Please note that both events can occur independently from each other with similar probability (for details, see text).

Role of position signals from the fCO

In the following, we investigated what role position and/or velocity signals from the fCO play for the induction of transitions in activity between trochanteral motoneuron pools. We focused mainly on the role of fCO elongation, but we also will report briefly the results from fCO relaxation.

First, we investigated whether the starting position of the fCO movement (i.e., the starting angle of the FT joint) affected the induction of transition in motoneuronal activity between levator and depressor motoneurons (N = 4; Fig. 6). For elongation stimuli, we tested fCO stimuli of constant amplitude (60°) from different starting positions (145, 120, and 95°) and compared the probability of induction of a transition in activity from depressor to levator motoneuron activity. Elongation stimuli at the fCO could elicit the transitions in motoneuron activity independent from the starting position (Fig. 6A). There was no difference in the probability for eliciting a transition in motoneuronal activity with respect to the starting position for stimuli from 95 and 120°. However, probability was decreased slightly for stimuli starting from 145° (Fig. 6B).

View larger version (24K): [in this window] [in a new window]   Fig. 6. A: influence of starting position of fCO movement on the release of a transition in trochanteral activity in the active animal. Please note that in the 3 sample recordings a transition between depressor and levator motoneuron activity occurs independent of the starting position at the fCO. B: plot of the frequency of occurrence of transition between depressor and levator motoneuron activity vs. 3 different starting positions (N = 4; n from 89 to 131). , values of individuals; , mean for the pooled samples. C: latency of onset of levator motoneuron activity after the onset of an fCO elongation stimulus (velocity: 120°/s). , values of individual animals tested (N = 4; n from 12 to 26); , mean values ± SD (see METHODS for evaluation procedures). Please note the latency of onset of levator activity is significantly (*) prolonged at a starting position of 145°. D: plot of the angular change in femur-tibia (FT) joint after stimulus onset at which the transition from depressor to levator activity during fCO elongation vs. starting position of the stimulus. Values of all tested velocities were pooled. , values of 4 individual animals (n from 38 to 50); , mean values ± SD for the sample. E: plot of the maximal frequency of levator motoneuron activity, recorded from the levator nerve C1 induced by fCO elongation vs. the starting position of fCO elongation. , values of 4 individual animals (n from 42 to 69); , means ± SD of the sample. Please note the maximal levator activity is significantly (*) reduced with a starting position of 145°. For better comparison to the situation in the inactive animal mean values for levator motoneuron activation in 2 inactive animals are given as well (; n from 11 to 21).

Second, we investigated whether the latency of onset of levator motoneuron activity after the start of fCO elongation depended on the starting position of the stimulus. We tested this for stimuli with a common velocity of 120°/s (Fig. 6C) from three different starting positions. There was no significant difference in the latency of levator motoneuron activation for the starting positions 95 and 120°. Only in case of the most extended fCO starting position (145°), was the latency significantly longer (P < 0.01).

Third, we investigated the relation between the switch in activities of the trochanteral motoneurons and the angular change in the FT joint from stimulus onset (Fig. 6D). We stimulated the fCO with ramp-and-hold stimuli of constant amplitude but from three differing starting positions. The change in fCO position at the time of transition from depressor to levator activity during the ramp stimulus was evaluated in relation to stimulus onset, and the equivalent joint angle was calculated and plotted versus the joint angle of the starting position. Figure 6D shows that the transition between motoneuronal activity occurred irrespective of the starting position. This result furthermore shows that there is obviously no fixed position at which the transition in activity between both motoneuron pools occurs (see also Fig. 2).

Finally, we tested whether there was a correlation between maximal motoneuron activity during bursts of activity induced and the starting position of the stimulus (Fig. 6E). For the extracellularly recorded levator motoneurons, we evaluated the maximum compound discharge rate in nerve C1 induced by fCO elongation. Maximum levator motoneuron activity during the burst initiated by fCO elongation was similar for stimuli starting from 95 and 120° but significantly lower at 145° (Fig. 6E). These results were different from the situation in the inactive state in which levator activity induced by fCO elongation does show a marked dependence on fCO position (Fig. 6E, ).

Similar results were obtained for the transition between levator and depressor motoneuron activity induced by fCO relaxation (not shown).

Role of velocity signals from the fCO

To investigate what influence velocity signals from the fCO may have on the induction of a transition in activity between trochanteral motoneurons (Fig. 7A), we first tested the probability of initiation of a transition for six different elongation stimulus velocities, i.e., 20, 60, 120, 200, 600, and 1,200°/s (Fig. 7B). There was no significant difference in the probability for the induction of a transition between the different stimulus velocities.

View larger version (31K): [in this window] [in a new window]   Fig. 7. Influence of stimulus velocity on the generation of transition in activity between depressor and levator motoneurons. A: sample recordings from 1 experiment showing that transition in activity can be elicited with different stimulus velocities for fCO elongation. B: plot of the probability of occurrence of transition in activity from depressor to levator motoneurons vs. fCO elongation stimulus velocity. , values of individual animals (N from 3 to 8, n from 56 to 216); , mean of the pooled data. C: plot of the maximum activation of levator motoneurons as recorded from the levator nerve C1 induced by fCO elongation vs. stimulus velocity. , values of individual animals (N from 2 to 5; n from 4 to 55); , mean ± SD of the sample. For comparison to the situation in the inactive animal, mean values of maximal levator motoneuron activation on the same stimuli in inactive animals are given ( triangles; n from 6 to 12).

Second, we investigated whether the maximal motoneuronal activity induced by fCO stimuli in trochanteral motoneuron pools was correlated with stimulus velocity. We evaluated the maximum frequency in levator nerve C1 that was elicited during bursts of activity in trochanteral motoneuron pools by ramp-and-hold stimuli at the fCO (Fig. 7A). We did not detect any correlation between fCO stimulus velocities and maximum motoneuronal activity (Fig. 7C). This situation strongly differs from the inactive state of the animal, in which marked velocity dependency of levator activity has been described (Fig. 7C, ) (see also Hess and Büschges 1997)

Similar results were obtained for the transition between levator and depressor motoneuron activity induced by fCO relaxation (not shown).

In summary, movement signals from the fCO can induce specific transitions in activity between the trochanteral levator and depressor motoneurons in the active stick insect. The direction of movement appears to be the most relevant feature for the transition induced. Position signals were found to have no qualitative influence on the probability of generation of the transitions but appear to affect the actual activity of trochanteral motoneurons within the transition. Elongation signals from the fCO were found to induce a transition from depressor to levator activity and relaxation signals induce the reverse transition. In the active state, in contrast to the inactive state, we could not find a consistent quantitative dependence of the performance of these transitions on fCO stimulus parameters (position and velocity). In the active state, movement signals from the fCO can induce activity changes in trochanteral motoneuron pools in a qualitative way by inducing or releasing transitions of activity between both motoneuron pools of the CT joint.

Influence of sensory signals from the fCO on rhythmic activity in trochanteral motoneurons of the otherwise isolated and deafferented mesothoracic ganglion activated by muscarinic agonists

From the preceding results, the conclusion arises that signals from the fCO may exhibit their influence on activity of trochanteral motoneurons by affecting central rhythm-generating networks driving the motoneuron pools (Bässler 1993; Büschges et al. 1995). To test this hypothesis, a preparation was needed that had to meet two criteria: first, in contrast to the active animal, such a preparation should exhibit regular rhythmic activity in leg motoneuronal pools. Second, it should allow the stimulation of fCO afferents in an otherwise completely isolated and deafferented situation to exclude any reafferent effects on motor activity arising from other sense organs of the middle leg. This can be achieved in a completely isolated and deafferented thoracic ganglion in which central neuronal networks driving leg motoneuron pools can be activated pharmacologically by the muscarinic agonist pilocarpine (Büschges et al. 1995).

Regular rhythmic activity was induced in trochanteral motoneurons by topical application of pilocarpine (Fig. 8). During fictive rhythmic activity, sensory signals from the fCO strongly affected trochanteral motoneuron activity in a way similar to the one found in the active behavioral state of the animal. Elongation of the fCO always reproducibly initiated levator motoneuron activity and terminated depressor motoneuron activity if delivered during the depressor phase (Fig. 8A). In case fCO elongation was delivered during levator activity, it was enhanced and the burst duration in levator motoneurons could be prolonged (Fig. 8B). Relaxation signals from the fCO did initiate depressor activity and terminated levator motoneuron activity, if present (Fig. 8). Relaxation and elongation signals from the fCO did induce a complete reset of the rhythm generated (Fig. 8, A and B; compare arrowheads), as shown in four experiments.

View larger version (38K): [in this window] [in a new window]   Fig. 8. Reset of the pilocarpine induced rhythmic activity in trochanteral motoneurons by stimulation of the fCO. A: fCO elongation (Pos) terminated ongoing depressor activity and induced levator activity (Lev). Relaxation of the fCO did reset the rhythm by restarting it with the depressor phase of the cycle (Dep). Arrowheads denote the expected time of occurrence of the onset of depressor activity in the undisturbed situation. B: fCO elongation during the levator phase of the cycle prolonged the levator activity. Subsequent relaxation of the fCO did reset the rhythm by restarting rhythmic activity with the depressor phase of the cycle.

Furthermore, continual stimulation of the fCO with ramp-and-hold stimuli could entrain rhythmic burst activity in trochanteral motoneurons (N = 15, Fig. 9, A and B). Levator activity was elicited by fCO elongation and depressor activity was elicited by fCO relaxation. fCO stimulation could entrain rhythmic activity above and below the inherent rate of the preparation (Fig. 9, Ai and B) provided that stimulation frequency was close to the inherent frequency of the preparation. However, it partially failed when the discrepancy was too large (Fig. 9Aii). This finding closely resembled the situation in a fraction of the active animals (see Fig. 1D) when trochanteral motoneuron activity sometimes was coupled to continual fCO stimulation. We did not yet investigate systematically the frequency range of fCO stimulation that could entrain the rhythm in trochanteral motoneurons. The membrane potential of trochanteral motoneurons was modulated by fCO stimuli in the same way as in the active behavioral state. This is shown for the slow depressor motoneuron in Fig. 9B (see Fig. 4B for comparison).

View larger version (56K): [in this window] [in a new window]   Fig. 9. Entrainment of the pilocarpine induced rhythmic activity in trochanteral motoneurons by signals from the fCO. Ai: on continual stimulation of the fCO, each fCO elongation elicited levator motoneuron (Lev) activation and each relaxation stimulus elicited depressor motoneuron (Dep) activation. Cycle period of rhythmic activity was 4.7 ± 0.5 s before stimulation and 4.6 ± 0.9 s after stimulation. Ramp-and-hold stimulus period was 4 s. ii: please note that rhythmic activity in trochanteral motoneurons was not entrained with stimulus protocol exceeding the inherent frequency of the preparation twofold (example given on right; see text). Cycle period of rhythmic activity was 4.2 ± 0.1 s before and 4.2 ± 0.1 s after fCO stimulation. Ramp-and-hold stimulus period was 2 s. B: intracellular recording from SDTr during entrainment of rhythmic activity. , time of occurrence and expected time of occurrence of a depressor motoneuron burst before and during stimulation of the fCO, respectively.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We have shown that movement signals from the FT joint do take part in patterning motoneuronal activity of the CT joint in the active behavioral state with the locomotor system being active. Elongation of the fCO, mimicking joint flexion, induced a transition from depressor to levator activity and relaxation of the fCO, mimicking joint extension, induced the opposite transition. Compared with the interjoint reflex response of the inactive state, the influence of sensory signals from the fCO in the active state represents a drastic increase and thus a qualitative change. This influence of sensory signals from the fCO was not related to motor activity in the FT-joint motoneuron pools, i.e., the generation of the reflex reversal. The occurrence of the activity transitions in the trochanteral motoneurons furthermore appeared to depend on the direction of the fCO movement, i.e., either extension or flexion of the FT joint only. The generation of the transitions appeared to be basically independent of specific stimulus parameters at the fCO, i.e., starting position or stimulus velocity. In some animals, trochanteral motoneuron activity did couple tightly to ongoing fCO stimulation, resembling entrainment of rhythmic motor activity by sensory stimuli. In fictive, i.e., pilocarpine-activated, rhythmic preparations of the otherwise deafferented and isolated mesothoracic ganglion fCO stimulation did reset and entrain the ongoing rhythmic activity in trochanteral motoneurons. In summary, our results strongly suggest that when the locomotor system is active, movement signals from the fCO do have access to central rhythm generating networks that drive trochanteral motoneurons.

Role of sensory signals from the fCO in interjoint coordination between FT and CT joint in the active behavioral state

First evidence for influences of sensory signals from the FT joint on the activity of trochanteral motoneurons was derived from experiments in which sensory input from the fCO was reversed surgically by fixing its apodeme to the flexor muscle. In these animals (i.e., with a so called "crossed fCO receptor apodeme"), extension of the FT joint during walking movements, now inducing elongation of the fCO, initiated leg levation during walking and severely perturbed the locomotor output of the leg ("saluting") (Bässler 1993; Graham and Bässler 1981). In the present investigation, we have shown that elongation signals from the fCO initiate leg levation by eliciting levator motoneuron activity and terminating depressor motoneuron activity.

The present findings may give rise to the assumption that mainly sensory signals from the fCO determine trochanteral motoneuron activity. However, this view appears too simplistic in the light of the following arguments: 1) effectiveness of fCO stimulation, as measured by the probability of occurrence for the transitions, was not as high in the active behavioral state as in fictive rhythmic preparations (70% in the active animal compared with 100% in fictive rhythmic preparations), indicating that there is not a completely fixed relationship between sensory signals from the fCO and transitions in activity in trochanteral motoneurons and indicating that patterning of activity in trochanteral motoneurons also is influenced by other sensory and central influences. 2) Activity of trochanteral motoneurons in the active state is as well influenced by signals from at least two other sense organs, load-sensing sense organs, i.e., from the campaniform sensillae (Bässler 1993), and signals from intrajoint proprioceptors, e.g., the trochanteral hairplates and strand receptors (Bräunig and Hustert 1983; Schmitz 1985). These sensory systems contribute to the patterning of trochanteral motoneuron activity in the active animal (Bässler 1993; Schmitz and Schöwerling 1992).

The influence of sensory signals from the fCO on CT-joint motoneuron activity may contribute to the coordination of movements between both leg joints when the locomotor system is active. This becomes obvious from the interjoint coordination during searching and walking movements in the single-leg preparation (Bässler et al. 1991; Haas et al. 1996) and from the role that exhibition of the reflex reversal in the FT-intrajoint control networks plays in the generation of the locomotor cycle: 1) during walking movements in the single leg preparation, front legs and middle legs flex the FT joint during stance and extend the FT joint during swing. Concurrently during late stance and early swing, the CT joint is levated and the CT joint is depressed in late swing after extension of the FT joint. The interjoint coordination is similar for both leg joints during searching movements (Bässler 1993). 2) There was no fixed coupling between the generation of the reflex reversal ("active reaction") (Bässler 1988) in the FT joint and the generation of a transition in activity between trochanteral motoneurons. Bässler (1988) has shown that the initiation of extensor activity in the second part (part II) of the active reaction after the reflex reversal (part I) (Bässler 1988) reflects a portion of the locomotor cycle, i.e., the transition from stance to swing. Transition between trochanteral depressor to levator activity was independent of the active reaction of the extensor motoneurons. It mostly occurred rather late during fCO elongation, i.e., FT-joint flexion, and the levator activity initiated could outlast the stimulus (Figs. 1-3), suggesting that sensory signals from the fCO may contribute to levator activation at the end of stance and the begin of swing. Thus our data indicate that movement signals from the fCO can contribute to establishing proper interjoint coordination between FT joint and CT joint.

However, with respect to an immediate functional interpretation of the gathered results toward an understanding of walking pattern generation in the intact walking stick insect middle leg, it should be kept in mind that marked appropriate changes in the angle of the FT joint during stance only occur in the front leg, whereas they are much more subtle for the middle leg (Bässler 1988). This restriction, however, does not affect our basic finding, that sensory signals from the fCO take part in patterning of motor activity in the CT joint.

Neuronal control networks governing active leg movements

Our results may appear interesting in the light of the present understanding of movement generation in multijointed limbs. This is because data from "fictive rhythmic" preparations of locomotor systems emphasized the major role of "centrally preprogrammed" coupling between adjacent leg joints by the action of a common central pattern generator for locomotion (e.g., crayfish: Crachri and Clarac 1990; locust: Ryckebusch and Laurent 1993; cat: summary in Grillner 1981). Flexibility of the locomotor output generated and its proprioceptive regulation, however, resulted in the formulation of more complex ideas on the construction of pattern generators for locomotion, i.e., the "unit-burst generator concept" (summary in Grillner 1981) and the "module concept" (summary in Bässler 1993).

Our findings clearly indicate a pivotal role of sensory signals in the establishment of interjoint coordination during locomotion. In the stick insect walking system, individual central rhythm-generating premotor networks exist for the different leg joints of each leg (Bässler and Wegner 1983; Büschges et al. 1995). Recently evidence was presented for central as well as peripheral signals contributing to interjoint coordination during the execution of active leg movements (Bässler 1993; Büschges et al. 1995). Up to now, however, it still remained unclear whether there was a direct influence of sensory signals of one leg joint on the rhythm-generating networks of the adjacent leg joint. From our experiments, it has become clear that indeed sensory signals from proprioceptors of one leg joint do have access to rhythm generating networks of adjacent leg joints.

It is not clear yet, how this influence is mediated. This applies specifically for sensory-motor pathways that underlie interjoint posture control action in the inactive state (Hess and Büschges 1997). Direct as well as polysynaptic pathways have been described that transfer information from the fCO onto trochanteral motoneurons. Preliminary data suggest their contribution. For example, interneurons contributing to interjoint reflex action in the inactive animal by disinhibition of levator motoneurons, i.e., Li1 and Li2, do show drastically enhanced inhibitory synaptic inputs during fCO-induced levator activation in the active behavioral state (unpublished observation).

In the crayfish, evidence was presented for the phasic control of the efficacy of direct pathways from sensory signals from one leg joint, i.e., the coxa-basopodit joint, to motoneurons of the adjacent thoraco-coxal joint to contribute to interjoint reflex action on the adjacent during the exhibition of rhythmic motor activity (El Manira et al. 1991). This control of synaptic efficacy was provided by phasic presynaptic inhibition of sensory afferents by signals from central rhythm generating networks. In the stick insect, it is not clear yet which mechanisms mediate the influence of sensory signals from the fCO onto trochanteral motoneuron pools in the active animal. Mechanisms similar to the ones described in crayfish may contribute to the increased influence of sensory signals from the fCO in the active behavioral state, in comparison with the inactive animal. Future investigations will have to unravel the neuronal mechanisms that mediate the influence of sensory signals from the fCO on rhythm generation in the trochanteral motoneurons.