| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Department of Zoology, University of Cambridge, Cambridge, United Kingdom; and 2Abteilung für Biologische Kybernetik und Theoretische Biologie, Fakultät für Biologie, Universität Bielefeld, Bielefeld, Germany
Submitted 16 August 2007; accepted in final form 17 November 2007
 |
ABSTRACT |
|---|
|
|
|
INTRODUCTION |
|---|
|
Locusts respond to tactile stimulation of discrete locations on the surface of a forewing with appropriately aimed movements of the ipsilateral and sometimes contralateral hind legs that cross the stimulus site (Dürr and Matheson 2003
; Matheson 1997, 1998; Page and Matheson 2004). Behavioral analyses indicate that the movements are generated by a stimulus-dependent shift in a single underlying motor pattern (Dürr and Matheson 2003). The tarsus (foot) does not touch the wing surface in most scratches, so the movement is unimpeded and, in contrast to stepping, is not constrained by mechanical coupling to the other legs or a substratum. Grooming movements aimed at stimulus sites on the abdomen or middle leg involve characteristically different rhythmic patterns of activity in at least four leg muscles; the exact phase relationships between them depending on the location of the stimulus (Berkowitz and Laurent 1996). Where stimulus sites used by Berkowitz and Laurent (1996) lay beneath the folded wings, the grooming response was very similar to that made in response to stimulation of the overlying wing, suggesting that adjacent points on the surfaces of the abdomen and the wings are mapped to similar limb targets (Dürr and Matheson 2003). For all stimulus locations, antagonistic muscles such as tibial flexors and extensors are active in alternating bursts (Berkowitz and Laurent 1996).
To address our first objective, the issue of posture dependency, we quantified the timing of tibial extensor and flexor motor neuron activity and determined its effects on kinematics of movements aimed at different targets on the wing. We recorded from four to six of the nine excitatory flexor motor neurons (Sasaki and Burrows 1998) and simultaneously used video motion capture to measure all the contributing limb joint angles. We have previously hypothesized that high joint stiffness in the locust hind leg could provide a means for counteracting the effects of limb loading (Matheson and Dürr 2003). One mechanism that could generate high stiffness is an increase in the level of antagonistic muscle co-contraction. Typically, co-contraction is quantified as the amount of co-activation of antagonistic motor neurons, but the slow activation dynamics of insect muscle can permit strong co-contraction even without motor neuronal co-activation (Zakotnik 2006; Zakotnik et al. 2006). We are therefore careful to distinguish between co-activation (of motor neurons) and co-contraction (of muscles) throughout this paper because the latter can occur without the former.
In addition to examining the posture dependency of motor activity, we examined the effects and posture dependency of passive limb stiffness of the femur-tibia joint (Burrows and Horridge 1974).
Our second objective was to contrast the role of the fast extensor tibiae motor neuron (FETi) in aimed scratching movements with that of the only other excitatory extensor tibiae motor neuron, the slow extensor tibiae (SETi). The hind legs of a locust are specialized for kicking and jumping by virtue of their large size, biomechanics, and neuronal control. FETi is a key element of this system (see Burrows 1996 for a thorough review). It fires at high frequency during the preparation for a kick or jump when it provides a strong and direct central synaptic excitation to all of the flexor tibiae motor neurons—thus promoting their co-activation and considerable co-contraction of the antagonistic tibial muscles (Burrows et al. 1989; Hoyle and Burrows 1973). This central synaptic output from FETi onto the flexor motor neurons is unique to FETi of the metathoracic ganglion and is thought to reflect a specialization for kicking and jumping (Burrows 1996). FETi of the locust hind leg is not active in walking (Burns and Usherwood 1979), although it can be recruited during the forceful movements of righting when the leg is in contact with the ground (Faisal and Matheson 2001). Berkowitz and Laurent (1996) suggested that FETi might be active in larger sudden grooming movements aimed at the ear, and Burrows (1995) illustrated that FETi could produce one or two spikes during slow tibial extensions that preceded a subsequent co-contraction and kick. Little is known, however, about the differential recruitment of the slow (SETi) and fast (FETi) motor neurons. We show that FETi contributes to normal scratching movements, and we quantify the relative contributions of FETi and SETi to joint angular velocity and acceleration.
To carry out our analyses, we have made use of a well-described model system in which the excitatory innervation of the locust hind leg tibial extensor muscle is simple and well characterized, comprising just one fast and one slow motor neuron that each innervate different bundles of fibers within the whole muscle (Hoyle 1955a,b). A single common inhibitor motor neuron and a neuromodulatory dorsal unpaired median (DUM) neuron provide the only other input (Burrows 1973; Hoyle 1955a,b, 1978; Usherwood and Grundfest 1965). The tibial flexor muscle is innervated by only nine excitatory motor neurons (Burrows and Horridge 1974; Burrows and Hoyle 1973; Phillips 1980), three each of which have fast, slow or intermediate properties (Phillips 1980; Sasaki and Burrows 1998). The motor neurons innervate distinct but partially overlapping groups of muscle fibers within the muscle (Sasaki and Burrows 1998). Such patterns of multiple innervation are the norm in insects. Two common inhibitory motor neurons complete the innervation of the flexor tibiae muscle (Hale and Burrows 1985; Wolf 1990).
We have demonstrated that for aimed limb movements in an invertebrate, motor neuron groups are recruited in a posture-dependent manner, that there is differential recruitment of different motor neurons of the same muscle, that passive forces can drive movements, and that joint angle velocity is most strongly dependent on motor burst duration and spike number.
|
|
METHODS |
|---|
|
Electrophysiology and identification of tibial motor neurons
Tibial muscle activity was recorded using myogram electrodes made from pairs of 20-µm-diam polyester insulated copper wires that were passed from the recording sites in the metathoracic leg to the pronotum, fixed at intervals by small drops of beeswax (Fig. 1A). The tips of the wires were stripped of their insulation to improve the recorded signal. In each animal, three recording sites were selected on the basis of the known innervation of muscle bundles and the sites of bundle insertion, which are identified by characteristic patterns of cuticular coloration. The left mesothoracic leg was amputated at the base of the femur to prevent the locust gripping the myogram wires, and the stump was sealed with a wax plug (Lactona Surgident, Philadelphia, PA).
|
). It comprises 24 muscle bundles, including 2 accessory flexor bundles. All bundles are divisible into anterior and posterior parts and insert onto a common apodeme.
Nine excitatory motor neurons innervate the flexor tibiae muscle, although each innervates only a restricted subset of muscle fibers. For example, the most proximal and distal fibers are innervated by nonoverlapping subsets of motor neurons (Sasaki and Burrows 1998). Some fibers in the proximal bundles receive innervation from as many as seven of the nine excitatory motor neurons (Sasaki and Burrows 1998). In addition to the nine excitatory motor neurons, the tibial flexor muscle is also innervated by the common inhibitors motor neurons CI2 and CI3 (Hale and Burrows 1985; Wolf 1990).
The metathoracic tibial extensor muscle is 88% greater in mass than the tibial flexor muscle (Bennet-Clark 1975). It is composed of pennate fibers that insert on a common apodeme and is innervated by only four motor neurons: the FETi and SETi motor neurons; a common inhibitor (CI1) (Burrows 1973; Hoyle 1955a,b; Usherwood and Grundfest 1965); and a dorsal unpaired median neuron (DUM neuron to the extensor tibiae: DUMETi (Hoyle 1978). Individual muscle fibers are innervated in 1 of 12 neuron combinations and have been classified as slow, intermediate, or fast on the basis of their ultrastructure and thus contractile properties (Hoyle 1978). Fibers innervated by FETi but not SETi are invariably fast type, whereas those innervated by SETi and not FETi are invariably slow type. Intermediate fibers are innervated by both FETi and SETi (Hoyle 1978). The proportion of fast fibers decreases from proximal to distal along the femur, whereas the proportion of slow fibers increases.
Myogram electrodes were inserted into a distal anterior flexor tibiae muscle bundle a11, a proximal flexor bundle (p1 to a2) (nomenclature from Sasaki and Burrows 1998) and into the most distal extensor muscle bundle (Hoyle 1978) (Fig. 1B). The known innervation of these bundles indicates that the recorded activity should include both a fast and a slow flexor motor neuron in a11 (motor neurons 8 and 2) (Sasaki and Burrows 1998); fast, intermediate, and slow flexor motor neurons in p1–a2 (motor neurons 1, 3, 4–7, and 9); and both FETi and SETi in the distal extensor muscle bundle (Hoyle 1978). Our recordings of flexor bundle a11 sometimes included a second small amplitude motor neuron; presumably slow motor neuron 3 (Sasaki and Burrows 1998), which innervates the adjacent accessory flexor bundle and bundle a9. In electromyographic (EMG) recordings made from the extensor and flexor tibiae muscles of a locust leg, the largest-amplitude potentials are generally considered to be those of the fast motor neurons (e.g., Duch and Pflüger 1995; Sasaki and Burrows 1998), which is consistent with their patterns of activity and reflex responses in our study. Motor neurons were identified using standard criteria including spike amplitudes, resistance reflex responses to imposed movements of the tibia, and correlation of activity with spontaneous movements (Burrows 1995, 1996; Duch and Pflüger 1995; Field and Burrows 1982). The distal location of the extensor muscle recording site (see Fig. 1B) meant that the amplitude of FETi spikes was sometimes not much larger than those of SETi, as in Fig. 4A, because of the small muscle bulk at this location and the few fibers innervated by FETi. Despite this, FETi was always clearly recognizable because its spikes were also detected simultaneously on both the flexor channels (see e.g., Fig. 4A). We recorded both SETi and FETi in every animal.
|
EMGs were displayed off-line using custom-written software (Matlab 6.5; The MathWorks). Spikes were detected using a combination of adjustable thresholds and manual identification as described in the preceding text. The maximum error in our measurement of the time of spike occurrence is half the spike width, which is negligible relative to the time course of muscle activation and movement kinematics. Different small-amplitude (SA) flexor motor neuron spikes on a single channel were treated as a single pool. We analyzed separately the largest-amplitude (LA) spike in each flexor channel, which our criteria (outlined in the preceding text) identify as those of fast motor neurons. The proximal flexor muscle bundles, however, should be innervated by two fast motor neurons (see preceding text), so the second fast neuron would have been pooled with the slow and intermediate neurons in our analysis. Occasional cross-talk or artifacts were discounted manually by careful inspection of all recordings. SETi, FETi, and flexor motor neuron spikes were then analyzed separately for each recording site.
Bursts of activity were identified manually to take into account the wide range of spiking frequencies and patterns of activity in different scratches. A burst was defined as a discrete group of spikes of similar or smoothly changing frequency separated by an interval of more than two times the interspike interval from the subsequent group of spikes. In practice, bursts were clearly discernable and alternated between the flexor and extensor recordings. Individual spikes that occurred between bursts were ignored in statistical analyses of burst parameters.
Kinematics and stimulation protocol
Locusts were tethered above a foam ball with a mass of 7.2 g using a fine wire noose placed around the pronotum. This allowed free movement of all the legs and allowed the animal to adjust its posture (Matheson 1997). The eyes and ocelli were blacked out using water-based black acrylic paint (Cryla, Daler-Rowney). A 3-mm-diam horizontal rod provided a foot rest on which the left metathoracic tarsus was placed prior to each stimulus. To set a defined start posture, this rod was positioned 17 mm posterior to the metathoracic coxal joint, ventral to the abdomen. Retroreflective circular markers of 0.5- or 1-mm diameter were cut from reflective tape (Scotchlite, 3M: St. Paul, MN) and attached using clear nail lacquer to specific locations on the hindleg and thorax (Fig. 1C). The two anterior tibial spurs of the metathoracic leg were removed to give a flat surface for the attachment of the distal tibial marker.
Grooming behavior was elicited by applying light tactile stimulation using a fine paintbrush to either an anterior or a posterior location on a forewing. The mean stimulus locations are shown in Fig. 2. Anterior stimuli fell in regions 1 and 2 as defined by Dürr and Matheson (2003), whereas posterior stimuli fell in regions 4 and 5. Posterior and anterior stimuli were applied randomly until 
20 scratches were recorded for each stimulus site.
|
Animals were videotaped from the left side using a color video camera (JVC TK-C1460E) with a Tamron 28-mm lens, operated at a shutter speed 1/500 s. An optical fiber cold light source (Schott KL 1500-T) was positioned behind and immediately above the camera so that the light beam was aligned parallel to the center of the camera lens. This maximized the brightness of the reflective markers. The video images were time-stamped using a video timer (For-A VTG-33), recorded on an sVHS video recorder (JVC HR-S7500EK), displayed on a monitor (Sony PVM-1450MD) and played back for capture using a personal computer video interface card (miroVIDEO DC30 plus, Pinnacle Systems). Images were compressed using the Microsoft DV Video codec. Interlaced PAL video frames were captured at a size of 720 x 540 pixels using MiroVIDEO Capture which yielded a spatial resolution of 1 mm/pixel. The AVI files were then deinterlaced (to give 50 frame/s) and subsequently edited using VirtualDub (freeware; http://www.virtualdub.org/index), to produce video clips of individual scratches in which the retroreflective markers could be tracked. A custom-written program was used to access the AVI files and to track the changing coordinates of the retroreflective markers in three dimensions (Videotrack 3D) (Zakotnik et al. 2004). To synchronize the videotaped movements with the recorded motor activity, a light-emitting diode (LED) placed in the field of view was manually triggered to flash at the start and the end of each scratch. The LED triggering pulse was recorded together with the electrophysiological signal to give a temporal resolution of ±10 ms. Scratches were analyzed from the fifth frame prior to tarsal movement until the frame in which the tarsus touched the ground, the brush or stopped for 200 ms.
Data analysis
Hindleg joint angles (Fig. 1D) were computed from the tracked markers and then smoothed using a constrained quintic smoothing spline limited by a 2° window half-size determined by the video tracking resolution (Zakotnik and Dürr 2005). For all joints, the first and second derivatives of joint angle gave joint angular velocities and accelerations respectively. EMGs and smoothed joint angle profiles were displayed off-line using custom-written software (Matlab 6.5; The MathWorks). Statistical analyses were carried out using Statistics Package for the Social Sciences (SPSS). Unless stated otherwise, means are computed across animals from per-animal means, and we report the corresponding SE. For nonnormal distributions we give median values and interquartile ranges. Results are considered to be significant if P < 0.05.
Two components of each scratch were distinguished as described in Dürr and Matheson (2003). The first was a short (200 ms) initial component (outward phase) in which the trajectory of the tarsus was relatively straight, whereas the second was a cyclic component (grooming phase) in which the tarsus moved in repeated loops near the target. The initial component was quantified by calculating the initial movement vector of the scratching movements. This was defined as the average direction of tarsus motion relative to the start position in the first 200 ms, measured in degrees (for derivation, see Dürr and Matheson 2003). The cyclic component was quantified using movement distributions that described the likelihood with which the distal 5 mm of the tibia moved across each point in the leg's work space (for derivation, see Dürr and Matheson 2003).
To analyze the posture-dependency of motor neuron activity, we interpolated our measurements of leg joint angles to 5 kHz using a quintic spline so that they matched the sampling frequency of the EMG data. At each time point we re-sampled the joint angle and checked for the presence of a motor spike, to yield probability distributions for spike firing (Figs. 6 and 7). We calculated separate distributions for time points at which there was no spike and for time points at which each particular motor neuron of interest spiked. We compared the centers of the distributions (no spike vs. spike) to give a measure of the bias (posture dependency) of firing of each motor neuron.
|
|
|
|
RESULTS |
|---|
|
To measure tibial muscle activity in scratching locusts required implantation of myogram wires into the left hind femur and amputation of the left middle leg (see METHODS). Neither of these manipulations had a significant effect on the overall form of scratches as assessed by probability distributions representing the trajectories of all recorded responses. Probability distributions for movements of the distal tibia (Fig. 2, Aii and Bii) were not significantly different to those of intact locusts used in a previous study (Dürr and Matheson 2003) (data re-plotted in Fig. 2, Ai and Bi) for either anterior (Fig. 2A, i and ii, P > 0.05) or for posterior scratches (Fig. 2B, i and ii, P > 0.05).
Scratching movements were generated by coordinated extensions and flexions of all three principal leg joints, the thoraco-coxal, coxo-trochanteral, and femoro-tibial (Fig. 3). Movements of the tibio-tarsal joint were not monitored. All scratching movements began with the hind leg tarsus standing on a platform that was located approximately half way along the wing, but ventral to the abdomen (Figs. 1C and 3, Ai, Bi, and Ci). Movements from this posture that were aimed at posterior stimuli generally showed overall depression of the thoraco-coxal and coxo-trochanteral joints (e.g., Fig. 3, Aii and Cii, lower 2 traces), whereas movements aimed at anterior stimuli showed overall levation at the same two joints (e.g., Fig. 3Bii, lower 2 traces). We illustrate two posterior scratches to indicate that similar tarsal trajectories could be followed at rather different velocities (with the scratch in C being slower than that in A) and that similar trajectories could be achieved with different combinations of basal leg joint angles. This is because the coxa is relatively short, and the thoraco-coxal and coxo-trochanteral joints act in the same plane, meaning that these two joints essentially act as one with a redundant degree of freedom. In the examples shown in Fig. 3, A and C, the sum of these two joint angles is approximately the same, yet their relative contributions to posture of the limb differ, particularly in the initial part of the response: in Fig. 3A, the initial thoraco-coxal angle is small and the coxo-trochanteral angle is large, whereas in Fig. 3C, the reverse is true. Toward the ends of both trials, the limb postures converge, demonstrating that different start postures can give rise to similar, target-dependent final postures. In scratches aimed at anterior targets (e.g., Fig. 3B), the required anterior rotation (levation) of the femur was achieved by levation of both these basal joints.
|
|
Posture dependency of temporal structure in motor activity
Extensor muscle recordings contained activity of the slow extensor tibiae motor neuron (SETi in Fig. 4A) and the fast extensor tibiae motor neuron (FETi and 
in Fig. 4A). FETi spikes, although sometimes not much larger in amplitude than SETi spikes (see METHODS), could be identified unambiguously by characteristic cross-talk on both flexor channels (Fig. 4A, *). SETi and SA flexor motor neurons did not cause cross-talk to other channels, but the largest (putative fast) flexor motor neurons sometimes did so. Where simultaneous activity was present on the two flexor channels (e.g., Fig. 4, A and C, 
) close inspection invariably revealed that it resulted from independent activity of different flexor motor neurons in the two recordings. This is consistent with the known patterns of innervation (see METHODS). When a flexor recording contained activity of more than one SA (putative slow or intermediate) motor neuron, these spikes were pooled together as "small-amplitude flexor tibiae motor neurons" for quantitative analyses (e.g., SA in Fig. 4). We did not pool data from the two flexor recording sites. When LA flexor tibiae motor neurons (LA in Fig. 4) were active, they generally fired one or two spikes during ongoing SA flexor bursts (Fig. 4, A–C).
All tibial motor neurons could produce either individual spikes or bursts of spikes during a scratch. For example, in the scratch illustrated in Fig. 4A, SETi produced a burst of spikes (1 of which is indicated, 
) to drive each tibial extension (top trace) but single spikes during some flexions (e.g., 3rd flexions in Fig. 4, A and C).
Extensor and flexor tibiae motor neuron activity alternated to drive extension and flexion of the femoro-tibial joint (Fig. 4, A–C). There was considerable variability in the temporal structure of EMG activity (Table 1), which reflected variability in the movements (e.g., compare top traces in Fig. 4, A–C). For example, different extension/flexion cycles of a single scratching movement could be carried out at different rates (compare cycles 2 and 3 in Fig. 4A), driven by different patterns of both extensor of flexor tibiae motor spikes (also see Differential recruitment of flexor tibiae motor neurons). These summary measures of EMG activity had large variability (Table 1). Proximal SA flexor motor neurons had longer bursts of activity with lower spike rates during anterior scratches than during posterior ones (Table 1). SA flexor tibiae motor neurons innervating the proximal and distal muscle bundles were active simultaneously in 44% of scratches with their bursts overlapping by an average of 73 ± 102 ms for anterior scratches and 18 ± 21 ms for posterior scratches. This is a mean percentage overlap of 21% of the duration for which the proximal SA flexor motor neurons were active and 28% of the duration for which the distal SA flexor motor neurons were active per scratch (anterior and posterior scratches pooled). For example, in the first flexion of Fig. 4C the proximal SA flexor burst overlapped with that of the distal SA flexor by 86%, but in the second flexion there was only 63% overlap. A corollary of this is that many bursts of SA flexor tibiae activity in one region of the flexor muscle were not accompanied by SA motor activity in the other region (e.g., 3rd cycle in Fig. 4C). This is described in Differential recruitment of flexor tibiae motor neurons. The patterns of activity in both proximal and distal flexors differed for anterior and posterior scratches. For example, during anterior scratches there were more bursts of proximal SA flexor activity, higher spike rates in the distal SA flexors, longer bursts of activity in both groups of SA flexors, which followed each other more closely than in posterior scratches. Similarly, bursts of SETi activity were shorter and further apart, with lower mean firing frequencies than during posterior scratches. Proximal and distal LA flexors, and FETi were all more active during anterior than posterior scratches (Table 1).
|
There were posture-dependent differences in the timing of movement relative to the timing of motor activity. For example, in two trials with posterior targets (Fig. 4, A and C), the start of tibial flexion often preceded the onset of flexor motor activity, whereas in a trial with an anterior target (Fig. 4B), tibial flexion always followed flexor motor activity. In all cases where tibial flexion movements preceded flexor motor activity, the femoro-tibial joint was initially extended beyond 90°. Flexion was faster in cycles with particularly large extensions (e.g., cycle 3 in Fig. 4A) than in cycles where the tibia remained less extended (e.g., cycle 3 in Fig. 4C). The onset of flexor motor activity preceded flexion in Fig. 4B, where maximum tibial extension always remained <80°. The timing of movement thus depended not only on the timing of motor neuron activity but also on the limb posture.
Passive extension from flexed postures
The apparently passive component of flexion movements (i.e., that preceding any recorded flexor motor activity) could have been driven by one or more flexor motor neurons that were not recorded. Extension movements, however, could also occur prior to, or in the absence of, extensor motor activity. Because we always recorded both of the excitatory motor neurons innervating the extensor muscle, we can rule out nonrecorded motor activity as an explanation for these movements. Figure 5 shows a response to an anterior stimulus with repeated extension of the femoro-tibial joint that occurred in the absence of any SETi or FETi activity. All of the extension movements began from flexed angles <50°. A series of rhythmic tibial movements were driven predominantly by SA flexor tibiae motor neuron activity alone. The first small (2.5°) extension and the second extension of 32° (1, 2 in Fig. 5, top trace) occurred in the pauses between bursts of flexor activity (lowest 2 traces) but without any excitatory extensor motor activity. The third extension (3) was driven at least in part by two spikes in SETi (2nd trace). The fourth extension, of 3°, (4) occurred in the absence of extensor activity, whereas the fifth and sixth were driven in part by one and two spikes in SETi, respectively. Large amplitude tibial extensions that occurred without extensor spikes always began with the joint strongly flexed (1 and 2). When the tibia started from relatively extended angles, only small amplitude extensions occurred in the absence of motor activity (4). This would be expected if extensions were driven in part passively by spring-like properties of the joint, as the passive joint torque should decrease as the tibia approaches the resting angle of 70°. The inset to Fig. 5 demonstrates that both SETi and FETi were recorded in this animal.
|
SA flexor tibiae motor neurons innervating the proximal and distal muscle bundles were activated independently. Figure 4 illustrates the extent of this motor flexibility. In Fig. 4A, the first tibial flexion (1st 
) was driven by a burst of activity in one distal SA flexor motor neuron (SA and in 4th trace). The second flexion movement was faster, and was driven by coincident activity in the same distal SA flexor and additional activity of a proximal SA flexor (). The third flexion was driven initially by a burst of activity in the proximal SA flexor (SA and in 3rd trace) with just a single coincident spike of the distal SA flexor motor neuron. The later part of this same ongoing flexion movement was driven by coincident activity of SA and LA flexor neurons innervating both muscle bundles (SA, LA in 3rd and 4th traces). In a similar way, the first two flexions of the scratch illustrated in Fig. 4C were driven by coincident bursts of activity in proximal and distal SA flexors (e.g., SF and in 3rd trace), but the third, slower, flexion was driven solely by activity of a distal SA flexor motor neuron (SF and in 4th trace).
Although we did not analyze separately the different amplitude flexor motor neurons innervating one region of the muscle, it was nevertheless possible to do so in some cases. Within a single scratch, each flexion movement could be driven by a different combination of SA motor neuron activity. For example, the tibia was held flexed at the start of Fig. 4B by a burst of activity of one distal SA flexor [SA(1) and in 4th trace] and two spikes of a second SA motor neuron [SA(2), 
in 4th trace]. The first active flexion ( labeled 1) was driven by activity of SA(1) (), SA(2), and one spike of a third SA motor neuron [SA(3), ]. The second active flexion was driven by activity in all of these motor neurons and one spike in a LA motor neuron (LA in 4th trace).
Recruitment of LA flexor tibiae motor neurons
Proximal LA flexor motor neurons fired one or two spikes in 10.2 ± 4.1% of anterior scratches and 5.0 ± 2.4% of posterior scratches, whereas distal LA flexor motor neurons fired between 1 and 16 spikes in 40.7 ± 2.3 and 26.4 ± 9.5% of the same scratches. In the scratch illustrated in Fig. 4A for example, a proximal LA flexor fired once during the third flexion movement (LA in 3rd trace) and the distal LA flexor fired once during each of the third and fourth flexion movements (LA in 4th trace). In the scratch illustrated in Fig. 4B, the proximal LA flexor fired one spike during the rapid large amplitude second flexion movement (LA in 3rd trace). This spike coincided with a burst of distal SA and LA flexor activity, but there was no concurrent activity in the proximal SA flexor motor neurons (3rd trace).
Bursts of activity of distal SA flexor motor neurons were likely to be of a significantly higher spike frequency if they were coincident with activity of an LA flexor motor neuron than if they were not [anterior scratches 146 (96, 185) Hz versus 116 (64, 150) Hz: Mann-Whitney U, Z39,242 = –3.46, P < 0.001; posterior scratches 135 (100, 198) Hz vs. 82 (59, 128) Hz: Mann-Whitney U, Z15,97 = –2.99, P < 0.001]. Values are medians and (25,75) percentiles. It was not possible to test this relationship for proximal flexor motor neurons because too few bursts were coincident with the spikes of LA flexors.
Recruitment of the fast extensor tibiae motor neuron (FETi)
FETi was active in 26 ± 13.0% of anterior scratches and 17 ± 2.2% of posterior scratches and could fire between one and nine spikes, usually during bursts of SETi activity. For example during the six extension movements in the scratch illustrated in Fig. 4A, FETi was silent during two extensions, fired one spike during three extensions and two spikes during one remaining extension ( in 2nd trace). In this example, FETi did not contribute to the first or last extension movements, which were therefore driven solely by SETi. The firing frequency of SETi bursts that coincided with an FETi spike was significantly higher than for those that did not, for both anterior and posterior scratches [anterior 100 (66, 121) Hz vs. 58 (39, 106) Hz, Mann-Whitney U, Z26,162 = –3.32, P < 0.001; posterior 153 (100, 200) Hz vs. 99 (61, 133) Hz, Mann-Whitney U, Z11,158 = –2.55, P < 0.05 respectively]. In other words, FETi was more likely to fire when SETi was firing more strongly.
There are three key observations about the patterns of FETi firing during scratching that distinguish them from the well-described patterns observed during kicking and jumping behaviors (Burrows et al. 1989; Hoyle and Burrows 1973). First, in scratching FETi often fired just one or two spikes (e.g., Fig. 4A) and did not fire prolonged high-frequency bursts. Second, FETi fired at times when the antagonist flexor motor neurons were silent, so that there was no recorded co-activation (Fig. 4A). In contrast, during preparation for kicking and jumping movements, FETi fires spikes at the same time as the flexor tibiae motor neurons, at least in part because it provides a strong and direct central synaptic excitation to all of them. Third, FETi fired at joint angles markedly greater than fully flexed ( in Fig. 3Aii, see next section), whereas it fires primarily at flexed positions in kicks and jumps.
Posture dependency of motor neuron recruitment
Having analyzed posture-dependent differences in the timing of motor neuron activity and differential recruitment of individual motor neuron subsets, we next determined the extent to which the recruitment of motor neurons from the different pools was also posture dependent. To do this, we compared the overall distribution of leg postures (Figs. 6 and 7, left) with the distribution of postures measured when each motor neuron of interest was active (Figs. 6 and 7, right, see METHODS). We analyzed anterior (top pair in each panel of 4) and posterior (bottom pair in each panel of 4) scratches separately. The center of density of each overall distribution is indicated by a white cross in the left column that is replicated in the corresponding panel of the right column to ease comparisons. The center of density of the subset of postures corresponding to only those frames in which the motor neuron of interest was firing is indicated by a white circle in the right column. A white arrow highlights the difference between the two centers in joint angle space.
SETi and SA flexor motor neurons
For anterior scratches, the overall distribution of leg postures was centered on 58°, 75°, i.e., a femoro-tibial joint angle of 58° and a coxo-trochanteral joint angle of 75° (Fig. 6Ai). For anterior scratches, there was no evidence for posture dependency in the activity of SETi or either SA flexor motor pool (Fig. 6, Aii–Cii). The distributions of angles at which SETi or the SA flexor motor neurons fired all resembled closely the corresponding overall distributions, with no angular bias exceeding 7° in either femoro-tibial or coxo-trochanteral joint angle (Fig. 6, Aii–Cii). The working range of joint angles was 
100° for both the femoro-tibial and coxo-trochanteral joints. We consider a bias of 10% of this range (i.e., 10°) to indicate a posture dependency of motor neuronal firing.
For posterior scratches, the overall distribution of leg postures was centered on 73°, 47° (Fig. 6Aii). For posterior scratches, there were larger, but still only small, biases in both joint angles for SETi and all SA motor neurons (Fig. 6, Aiv–Civ). SETi was more likely to fire when the femoro-tibial joint was relatively flexed. The center of the joint angle distribution was at 61°, 54°, representing a bias of –12°, +7°; Fig. 6Aiv). SA flexor motor neurons were more likely to fire when the femoro-tibial joint was relatively extended and the coxo-trochanteral joint was relatively depressed (proximal flexors: bias +11°, –12°, Fig. 6Biv; distal flexors, bias +6°, –15°, Fig. 6Civ).
Posterior scratches used more extended and depressed limb postures than did anterior scratches (e.g., compare locations of crosses in Fig. 6A, i and iii). A comparison of the posture dependence of motor neuron recruitment for these two situations reveals different trends for extensor and flexor motor neurons. The posture dependence of SETi recruitment was weak and similar for anterior and posterior scratches despite the postural differences underlying movements to each target. In contrast, SA flexor motor neuron recruitment in posterior scratches tended toward even more extended and depressed postures than expected from the overall extension and depression in postures used to reach the posterior target; ie., crosses (representing the centers of the distributions of all postures) in Fig. 6, Biv and Civ shifted down and right relative to positions in Bii and Cii, respectively, whereas circles (representing the centers of distribution for SA motor activity) shifted even further in the same direction.
FETi and LA flexor motor neurons
For anterior scratches, the overall distribution of leg postures was centered on 58°, 75° (Fig. 7Ai) and for posterior scratches 73°, 45° (Fig. 7Aii). For anterior scratches, FETi was most likely to fire when the femoro-tibial joint was relatively (but not fully) flexed and the coxo-trochanteral joint was relatively levated (bias –34°, 25°, arrow in Fig. 7Aii). Some of these anterior scratches superficially resembled kicks because tibial extension was rapid, but there was no initial full flexion of the tibia and no co-activation of FETi and flexor motor neurons. For posterior scratches the likelihood of FETi firing depended only on femoro-tibial angle (bias –12°, +2°, Fig. 7Aiii). FETi generally fired near the start of extension movements as illustrated by the examples of Figs. 3Aii and 4A) but could also fire later during an extension. The majority of FETi spikes therefore occurred while the tibia was partially extended (range of angles, 10.5–101°, Fig. 8), which differs markedly from the pattern of firing seen during kicks.
|
Effect of motor neuron activity on limb kinematics
Bursts of activity in SETi or SA flexor motor neurons caused significant changes in the femoro-tibial joint angle, velocity, and acceleration as measured immediately before and after the burst (Table 2). There was a posture dependency in the effect of SETi bursts on angular velocity and acceleration, whereas this was not so for SA flexor motor neurons (Table 2). The magnitude of changes in tibial movement was controlled by the number of SA motor neuron spikes in a burst, the burst duration and, to a lesser extent, the firing frequency during the burst (MANOVA, Table 3). There was posture dependency in the effects on tibial movement of all three measured burst parameters for both SETi and proximal SA flexors. All three parameters reflect increased extensor activity in anterior scratches and increased flexor activity in posterior scratches. Post hoc testing of the results in Table 3 revealed several key relationships that were largely responsible for the overall significance of the MANOVA. An increase in the total number of spikes in a burst was significantly correlated with an increase in the subsequent angular change of the femoro-tibial joint during both anterior and posterior scratches, for SETi and all SA tibial motor neurons (Fig. 9, Ai–Ci). The same was true for the velocity of angular change (values of Spearman's Rho for anterior scratches: SETi, rs = 0.39; proximal flexor motor neurons, rs = –0.30; distal flexor motor neurons, rs = –0.25: posterior scratches: rs = 0.31, r55 = –0.61, rs = –0.21 respectively. P < 0.001 in all cases. Data not shown). Longer bursts of firing in SETi, proximal SA flexors or distal SA flexors all drove significantly larger angular excursions, in both anterior and posterior scratches (Fig. 9, Aii–Cii). Bursts with higher SETi firing frequency drove smaller tibial extensions (anterior, rs = –0.30, P < 0.001; posterior, rs = –0.29, P < 0.001; data not shown). This somewhat counterintuitive relationship occurred because high firing frequencies only occurred in short bursts.
|
|
|
The change in velocity following an FETi spike was significantly greater than that which could be driven by bursts of SETi alone, during both anterior and posterior scratches [type I ANOVA: anterior, Z2,194 = 68.13, P < 0.001; posterior, F(2,170) = 7.38, P < 0.01].
|
|
DISCUSSION |
|---|
|
–8). Furthermore, the simultaneous recording of three distinct groups of excitatory motor neurons from two antagonistic muscles allows us to highlight differential recruitment of motor neurons innervating distinct regions of the same muscle (Fig. 4), to quantify the amount of neuronal co-activation, and to isolate the kinematic effects of motor neurons of different physiology (Figs. 9 and 10). Differential recruitment and co-activation of motor neurons
This study is the first to record simultaneously the activity of two distinct groups of flexor tibiae motor neurons (proximal and distal) in a moving locust. Flexor tibiae motor neurons are not coupled by central synapses, and no two receive identical patterns of synaptic input, but they can have a synergistic action during movements of the tibia caused by many common inputs from interneurons (Hoyle and Burrows 1973). During scratching movements, proximal and distal SA flexors often acted synergistically as reflected by their overlapping bursts of activity. On the other hand, distal LA flexors were recruited more strongly and more often than were proximal LA flexors. This suggests a regional division of labor within the flexor tibiae muscle during scratching movements, reflected both in the patterns of innervation of the motor neurons and their differential recruitment. A similar division is proposed for the stick insect extensor tibiae muscle, where the distal region is specialized for co-contraction with the tibial flexor muscle in the generation of catalepsy (Bässler and Stein 1996; Bässler et al. 1996).
The differences in motor neuron recruitment observed for anterior and posterior scratches (e.g., Table 1) suggest that it is likely that the tibial flexor muscle is relatively more important than the extensor in driving the movement pattern underlying anterior scratches, whereas the tibial extensor plays a larger role in posterior scratches. This difference corresponds to the relatively more extended tibial angles required to reach posterior targets.
Co-activation of the tibial flexor and extensor motor neurons occurs in upside down walking of locusts (Duch and Pflüger 1995), and during walking and running in the cockroach (Ewing and Manning 1966; Krauthamer and Fourtner 1978; Larsen et al. 1995; Watson and Ritzmann 1998a), but is not evident in grooming movements of locusts aimed at the ear or abdomen (Berkowitz and Laurent 1996). In contrast, the trochanteral levator and depressor motor neurons show considerable co-activation during grooming of the ear, which is located on the anterior abdomen (Berkowitz and Laurent 1996). We show that there is little co-activation of the recorded antagonistic MN pools during scratches aimed at the wing. Despite the low levels of coincident activity in antagonistic tibial motor neurons, there is nevertheless likely to be considerable muscular co-contraction because the time constants of single muscle twitch contractions are long compared with the intervals between antagonistic motor neuron bursts (Zakotnik et al. 2006). The time constant for the locust extensor tibiae muscle responding to SETi stimulation is 75 ms (Zakotnik 2006) and that for the same muscle responding to combined SETi/FETi/common inhibitor stimulation in a stick insect (Carausius morosus) middle leg is 25 ms (Guschlbauer et al. 2007; their Fig. 7). The time constant for a stick insect (Cuniculina impigra) flexor tibiae muscle responding to spontaneous activity of an LA flexor motor neuron is 40 ms (Bässler 1984; his Fig. 3). In all these cases, deactivation time constants are longer than activation time constants, and together these slow activation dynamics lead to considerable residual tension and, as a consequence, to increased limb stiffness for long periods after motor neuron activity has ceased. Activation of the common inhibitor motor neurons could also affect the strength of co-contraction through modulation of the rate of muscle relaxation. Co-contraction of antagonistic tibial muscles is important for postural stability (Hoyle and Burrows 1973) but also provides an important mechanism for load compensation in active movements (Zakotnik et al. 2006).
Posture dependence of motor activity
At a femoro-tibial angle of 70°, the passive torques generated by the inactive extensor and flexor tibiae muscles balance out (Burrows and Horridge 1974; Zakotnik et al. 2006). Movements away from this neutral posture require active torque, but movements toward it may be assisted by the passive forces. Both flexor and extensor motor neurons contribute more to movement away from the neutral posture than to movement toward it. For example, SETi fires more during extension movements in posterior trials than in anterior trials because the limb posture is generally more extended and requires larger torques to overcome the strong passive forces intrinsic to the limb mechanics (Zakotnik et al. 2006). Conversely, SA flexor motor neurons are more active during flexion movements in anterior trials, reflecting the increased torque required to flex an already flexed limb further away from the neutral posture.
The difference in limb postures used for anterior and posterior scratches governs the orientation of the tibia relative to gravity: in anterior trials, the levated coxo-trochanteral joint lifts the tibia so that it is almost horizontal, whereas in posterior trials the depressed coxo-trochanteral joint means that the tibia moves through angles somewhat closer to vertical. As a consequence, gravity assists tibial flexion more in anterior trials than in posterior trials. In summary, Fig. 6 shows that SA flexor recruitment shifts toward postures for which the assisting effect of gravity on tibial flexion decreases and therefore may require more neural activity for equivalent joint angle excursion.
Measurements made by Burrows and Horridge (1974; their Fig. 9) suggest that the passive restoration torque toward the neutral posture is 0.4 x 10–6 Nm/°. Modeling results from Zakotnik et al. (2006) suggest the larger value of 0.98 x 10–6 Nm/°. Given that the mass of the tibia and tarsus is 23 mg and the center of mass is 15 mm distal to the femoro-tibial joint, passive joint torque is sufficient to move the tibia toward the neutral posture from nearly any limb orientation. Indeed, model calculations by Zakotnik et al. (2006) suggest that even a loaded tibia can be extended by passive forces alone provided the joint is sufficiently flexed. In line with this prediction, we sometimes recorded extension of the femoro-tibial joint in the absence of any extensor motor activity (Fig. 5). These passive movements were almost always of low amplitude (a few degrees) although they could be up to 32° if the joint was initially flexed. Stronger movements reported by Berkowitz and Laurent (1996) in the absence of recorded extensor motor activity could have been driven by undetected activity of SETi in their study (see Recruitment of fast and slow motor neurons), but led to the hypothesis that such movements might result from energy transfer from elevation of the femur driven by movements of the basal joints. Our results in Fig. 4 suggest that there was a similar passive flexion from extended postures, but we cannot rule out the possibility that unrecorded flexor motor neurons contributed to these flexion movements.
There was strong posture dependence in the firing of FETi and LA flexor motor neurons. In anterior trials, FETi was more likely to fire at relatively flexed femoro-tibial angles than at extended angles. In posterior trials, FETi was more likely to fire near the neutral posture. In both cases, the posture at which FETi was most likely to fire corresponded to the prevalent turning point of the cyclic movement, i.e., the switch from flexion to extension. In anterior trials, LA flexors tended to fire at relatively extended angles, compared with the overall distribution of joint angles in these trials. Again this indicates that LA flexor motor neurons are preferentially recruited during early flexion, i.e., at the switch from extension to flexion. Activity in FETi and LA flexor tibiae motor neurons was also correlated with coxo-trochanteral joint angles, reflecting consistent patterns of interjoint coordination.
Recruitment of fast and slow motor neurons
Patterns of activity in metathoracic SETi during scratching are broadly similar to those observed during walking. During scratching, each burst of activity in SETi driving a single tibial extension contains on average 4–7 spikes and during the swing phase of a step SETi fires 3–10 spikes (Burns and Usherwood 1979). In contrast, although metathoracic FETi is recruited during scratching, as suggested by Berkowitz and Laurent (1996), it is not active during walking (Burns and Usherwood 1979). The patterns of spikes in some of the extensor EMGs illustrated by Berkowitz and Laurent (1996) and the characteristic cross-talk onto the flexor channel suggest very strongly that in some of their recordings (e.g., Figs. 6A, 8A, and 9A), only FETi activity was detected, whereas SETi activity remained undetected (cf. their Fig. 8A in which SETi is clear).
FETi is well known to be recruited during fast or forceful behaviors in locusts, such as defensive kicking, jumping (Burrows 1995; Heitler and Burrows 1977a,b; Hoyle and Burrows 1973), and righting (Faisal and Matheson 2001). We now show that FETi is also used in an ongoing movement, scratching, that does not involve extensor and flexor co-contraction or contact of the leg with another object. FETi can fire up to nine times during a single scratch but generally produces only one or two spikes per tibial extension. During righting, FETi fires 1–5 times (Faisal and Matheson 2001), and during the co-contraction phase of a kick FETi fires between 2 and 50 times (Burrows 1995). Our data therefore demonstrate that FETi can be recruited in a more varied fashion than previously demonstrated and is used during movements of moderate velocity: that of the distal tibia during posterior scratches was 8 ± 5 cm s–1, whereas that during a kick was 248 ± 131cm s–1 (medians ± interquartile range). FETi of the hind leg is unique among all the motor neurons (and different even to the homologous FETi of the other legs) in that it makes central excitatory output synapses onto all of the antagonist flexor tibiae motor neurons. This is thought to be a specialization for the generation of the motor pattern for kicking and jumping in which the hind leg tibial muscles are co-activated powerfully. During a kick, each spike in FETi causes a monosynaptic excitatory postsynaptic potential (EPSP) in all of the flexor tibiae motor neurons (see Burrows 1996). This usually results in several spikes in the slow flexor motor neurons and sometimes spikes in the fast flexor motor neurons (Burrows et al. 1989). Our EMG recordings provide no evidence that FETi spikes are followed closely by bursts of flexor activity that would be expected if the central synapse was operating at high gain during scratching (see e.g., Fig. 4A). The specialized synaptic connectivity between FETi and the flexor motor neurons, and their characteristic patterns of high-frequency firing in kicking and jumping, are therefore not apparent in scratching, suggesting that there may be context-dependent modulation of the central synapse. In cockroaches, FETi is recruited in a wide variety of behaviors including walking, searching, kicking and running (Hustert and Gnatzy 1995; Pearson 1972; Tryba and Ritzmann 2000; Watson and Ritzmann 1998b), but here there is no known central synapse with antagonistic flexor motor neurons (R. Ritzmann, personal communication).
In scratching, FETi was generally recruited when SETi was firing at relatively high frequencies, and, similarly, the LA flexor tibiae motor neurons were generally recruited when the SA flexor motor neurons were firing most rapidly. Similar patterns are observed for FETi recruitment in the running cockroach (Watson and Ritzmann 1998b) and for fast flexor tibiae recruitment in the walking stick insect (Gabriel et al. 2003) and can be explained by a degree of common synaptic input to agonistic slow and fast motor neurons coupled with a generally higher spiking threshold in the fast motor neurons (Burrows and Horridge 1974; Gabriel et al. 2003; Hoyle and Burrows 1973). Activation of FETi or LA flexor motor neurons during a scratch does not "swamp" ongoing movements driven by SA motor activity, so the resulting behavior remains smooth and well aimed. This is also the case for cockroach walking (Watson and Ritzmann 1998b).
In jumping and kicking, FETi becomes active only once the tibia has been locked into a fully flexed position. Its spiking activity is then enhanced by positive feedback from tibial campaniform sensilla signaling cuticular deformations resulting from the co-contractions of the antagonistic tibial muscles (Burrows 1996). During scratching, however, FETi is recruited without full tibial flexion and always without recorded co-activation of the tibial flexor motor neurons. Although FETi generally spikes at relatively flexed femoro-tibial joint angles, it is unlikely that the same positive feedback pathways are activated during scratching because forces at the joint will be relatively small while the tibia is free to move. We hypothesize that a lack of positive feedback permits FETi to fire single spikes rather than the high-frequency bursts generally seen during kicks and jumps.
Effect of motor neuron activity on the kinematics of scratching
Our data demonstrate that the magnitude and velocity of tibial movement are more strongly dependent on variations in the number of SA motor neuron spikes and the duration of motor bursts than on modulation of the firing frequency. This concurs with the observation that time constants of muscle activation dynamics are long compared with interspike intervals (Zakotnik 2006), resulting in a temporal integration of spike number. The numbers of spikes per burst were generally smaller than would be required to reach maximum contraction forces. For example, the median numbers in our experiments ranged from 4 to 9 spikes per burst (Table 1), as opposed to >50 SETi spikes needed to reach plateau isometric force at 50 spikes s–1 firing frequency (Zakotnik 2006). The spike-number dependence of slow invertebrate muscle has been demonstrated for an individual stick insect flexor tibiae motor neuron (Bässler 1984) and has been clearly distinguished from spike-frequency dependence in a muscle of the stomatogastric system (Morris and Hooper 1997). In walking cockroaches, there is a linear relationship between the firing frequency of a slow motor neuron and the velocity of joint movement (Watson and Ritzmann 1998a). A likely explanation for this difference is the low spike numbers per burst during scratching and the associated lack of saturation of contraction force.
When FETi or LA flexor tibial motor neurons were recruited during scratching, they had pronounced effects on the femoro-tibial joint angle and angular velocity. In the running cockroach, FETi is recruited to drive high-amplitude extensions but is not necessary to achieve high velocity movements, which can also be driven by SETi (Watson and Ritzmann 1998b). In our study, however, the velocity change driven by bursts of SETi spikes alone was substantially lower than that achieved by those bursts associated with the recruitment of FETi (controlling for SETi firing frequency). Activity of the common inhibitory motor neurons could enhance rapid tibial movements if their patterns of activity during scratching were timed appropriately (see e.g., Wolf 1990).
Our analysis of hind limb kinematics and motor activity demonstrates an important role for posture-dependant mechanisms in aimed limb movements and demonstrates how biomechanical properties of the musculo-skeletal system and differential recruitment of motor neurons interact to produce a stimulus-specific motor pattern.
|
|
GRANTS |
|---|
|
|
|
ACKNOWLEDGMENTS |
|---|
|
|
|
FOOTNOTES |
|---|
Present address and address for reprint requests and other correspondence: T. Matheson, Dept. of Biology, University of Leicester, University Road, Leicester LE1 7RH, UK (E-mail: tm75{at}le.ac.uk)
|
|
REFERENCES |
|---|
|
Bässler U. A movement generated in the peripheral nervous system: rhythmic flexion by autotomized legs of the stick insect Cuniculina impigra. J Exp Biol 111: 191–199, 1984.
Bässler U, Stein W. Contributions of structure and innervation pattern of the stick insect extensor tibiae muscle to the filter characteristics of the muscle-joint system. J Exp Biol 199: 2185–2198, 1996.[Abstract]
Bennet-Clark HC. The energetics of the jump of the locust Schistocerca gregaria. J Exp Biol 63: 53–83, 1975.
Berkowitz A, Laurent GJ. Local control of leg movement and motor patterns during grooming in locusts. J Neurosci 16: 8067–8078, 1996.
Burns MD, Usherwood PNR. The control of walking in orthoptera. II. Motor neuron activity in normal free walking animals. J Exp Biol 79: 69–98, 1979.
Burrows M. Physiological and morphological properties of the metathoracic common inhibitory neuron of the locust. J Comp Physiol 82: 59–78, 1973.[CrossRef]
Burrows M. Motor patterns during kicking movements in the locust. J Comp Physiol [A] 176: 289–305, 1995.[Medline]
Burrows M. The Neurobiology of an Insect Brain. Oxford, UK: Oxford Univ. Press, 1996.
Burrows M, Horridge GA. The organisation of inputs to motoneurons of the locust metathoracic leg. Philos Trans R Soc Lond B Biol Sci 269: 49–94, 1974.[CrossRef][Web of Science][Medline]
Burrows M, Hoyle G. Neural mechanisms underlying behaviour in the locust Schistocerca gregaria. III. Topography of limb motoneurons in the metathoracic ganglion. J Neurobiol 4:167–186, 1973.[CrossRef][Web of Science][Medline]
Burrows M, Watson AHD, Brunn DE. Physiological and ultrastructural characterization of a central synaptic connection between identified motor neurons in the locust. Eur J Neurosci 1: 111–126, 1989.[CrossRef][Web of Science][Medline]
Duch C, Pflüger H-J. Motor patterns for horizontal and upside-down walking and vertical climbing in the locust. J Exp Biol 198: 1963–1976, 1995.[Abstract]
Dürr V, Matheson T. Graded limb targeting in an insect is caused by the shift of a single movement pattern. J Neurophysiol 90: 1754–1765, 2003.
Ewing AW, Manning A. Some aspects of the efferent control of walking in three cockroach species. J Insect Physiol 12: 1115–1118, 1966.[CrossRef][Web of Science][Medline]
Faisal AA, Matheson T. Coordinated righting behaviour in locusts. J Exp Biol 204: 637–648, 2001.[Abstract]
Field LH, Burrows M. Reflex effects of the femoral chordotonal organ upon leg motor neurones of the locust. J Exp Biol 101: 265–285, 1982.
Gabriel JP, Scharstein H, Schmidt J, Büschges A. Control of flexor motoneuron activity during single leg walking of the stick insect on an electronically controlled treadwheel. J Neurobiol 56: 237–251, 2003.[CrossRef][Web of Science][Medline]
Guschlbauer C, Scharstein H, Büschges A. The extensor tibiae muscle of the stick insect: biomechanical properties of an insect walking leg muscle. J Exp Biol 210: 1092–1108, 2007.
Hale JP, Burrows M. Innervation patterns of inhibitory motor neurons in the thorax of the locust. J Exp Biol 117: 401–413, 1985.
Heitler WJ, Burrows M. The locust jump. I. The motor program. J Exp Biol 66: 203–219, 1977a.
Heitler WJ, Burrows M. The locust jump. II. Neural circuits of the motor program. J Exp Biol 66: 221–241, 1977b.
Hoyle G. The anatomy and innervation of locust skeletal muscle. Proc Roy Soc Lond B Biol Sci 143: 281–292, 1955a.[Medline]
Hoyle G. Neuromuscular mechanisms of a locust skeletal muscle. Proc Roy Soc Lond B Biol Sci 143: 343–367, 1955b.[Medline]
Hoyle G. Distributions of nerve and muscle fiber types in locust jumping muscle. J Exp Biol 73: 205–233, 1978.
Hoyle G, Burrows M. Neural mechanisms underlying behaviour in the locust Schistocerca gregaria. I. Physiology of identified motorneurons in the metathoracic ganglion. J Neurobiol 4: 3–41, 1973.[CrossRef][Web of Science][Medline]
Hustert R, Gnatzy W. The motor program for defensive kicking in crickets: performance and neural control. J Exp Biol 198: 1275–1283, 1995.[Web of Science][Medline]
Krauthamer A, Fourtner J. Locomotory activity in the extensor and flexor tibiae of the cockroach Periplaneta americana. J Insect Physiol 24: 813–819, 1978.[CrossRef][Web of Science]
Larsen GS, Frazier SF, Fish SE, Zill SN. Effects of load inversion in cockroach walking. J Comp Physiol [A] 176: 229–238, 1995.[Medline]
Matheson T. Hindleg targeting during scratching in the locust. J Exp Biol 200: 93–100, 1997.[Abstract]
Matheson T. Contralateral coordination and retargeting of limb movements during scratching in the locust. J Exp Biol 201: 2021–2032, 1998.[Abstract]
Matheson T, Dürr V. Load compensation in targeted limb movements of an insect. J Exp Biol 206: 3175–3186, 2003.
Morris LG, Hooper SL. Muscle response to changing neuronal input in the lobster (Panulirus interruptus) stomatogastric system: spike number vs. spike frequency dependent domains. J Neurosci 17: 5956–5971, 1997.
Page KL, Matheson T. Wing hair sensilla underlying aimed hindleg scratching of the locust. J Exp Biol 207: 2691–2703, 2004.
Pearson KG. Central programming and reflex control of walking in the cockroach. J Exp Biol 56: 173–193, 1972.
Phillips CE. An arthropod muscle innervated by nine excitatory motor neurones. J Exp Biol 88: 249–258, 1980.
Sasaki K, Burrows M. Innervation pattern of a pool of nine excitatory motor neurons in the flexor tibiae muscle of a locust hind leg. J Exp Biol 201: 1885–1893, 1998.
Snodgrass RE. The thoracic mechanism of a grasshopper, and its antecedents. Smith Misc Coll 82: 1–111, 1929.
Tryba AK, Ritzmann RE. Multi-joint coordination during walking and foothold searching in the Blaberus cockroach. I. Kinematics and electromyograms. J Neurophysiol 83: 3323–3336, 2000.
Usherwood PNR, Grundfest H. Peripheral inhibition in skeletal muscle of insects. J Neurophysiol 28: 487–518, 1965.
Watson JT, Ritzmann RE. Leg kinematics and muscle activity during treadmill running in the cockroach, Blaberus discoidalis. I. Slow running. J Comp Physiol [A] 182: 11–22, 1998a.[Web of Science][Medline]
Watson JT, Ritzmann RE. Leg kinematics and muscle activity during treadmill running in the cockroach, Blaberus discoidalis. II. Fast running. J Comp Physiol [A] 182: 23–33, 1998b.[Web of Science][Medline]
Wolf H. Activity patterns of inhibitory motor neurons and their impact on leg movements in tethered walking locusts. J Exp Biol 152: 281–304, 1990.
Zakotnik J. Biomechanics and Neural Control of Targeted Limb Movements in an Insect (PhD dissertation). Bielefeld, Germany: University of Bielefeld, 2006.
Zakotnik J, Dürr V. Motion analysis using stochastic optimisation and posture disambiguation. In: Proc. 3rd Int. Symp. Adaptive Motion in Animals and Machines (AMAM2005), edited by Witte H. Ilmenau, Germany, 2005.
Zakotnik J, Matheson T, Dürr V. A posture optimisation algorithm for model-based motion capture of movement sequences. J Neurosci Methods 135: 43–54, 2004.[CrossRef][Web of Science][Medline]
Zakotnik J, Matheson T, Dürr V. Co-contraction and passive forces facilitate load compensation of aimed limb movements. J Neurosci 26: 4995–5007, 2006.
This article has been cited by other articles:
|
|
 |
|
  S. L. Hooper, C. Guschlbauer, M. Blumel, P. Rosenbaum, M. Gruhn, T. Akay, and A. Buschges Neural Control of Unloaded Leg Posture and of Leg Swing in Stick Insect, Cockroach, and Mouse Differs from That in Larger Animals J. Neurosci., April 1, 2009; 29(13): 4109 - 4119. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. L. Page and T. Matheson Functional Recovery of Aimed Scratching Movements after a Graded Proprioceptive Manipulation J. Neurosci., March 25, 2009; 29(12): 3897 - 3907. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
TOP
ABSTRACT
INTRODUCTION
) were positioned on the thorax anterior to the metathoracic coxa (1), the thorax posterior to the prothoracic coxa (2), the base of the wing on its midline (3), the distal end of the wing on its midline (4), the metathoracic proximal femur (5), the metathoracic distal femur (6), the metathoracic proximal tibia (7), and the metathoracic distal tibia (8). Tactile stimuli were applied to anterior and posterior regions (
, means ± SE stimulus positions). Greyscale shading indicates the likelihood that a given position within the leg's workspace was reached (probability per mm2, see scale bar). 
, stimulus sites (those in Ai and Ci are partly obscured by the movement trajectory). Aii–Cii: time courses of joint angular change for the femoro-tibial, coxo-trochanteral, and thoraco-coxal joints (top to bottom). Each point represents an angle in 1 video frame. 
in Aii, times at which the fast extensor tibiae motor neuron (FETi) spiked.
) in inset at right]. When FETi was active, it caused cross-talk onto both flexor recordings (*). The slow extensor motor neuron contributed to only 3 of the 6 extensions (spikes in 2nd trace), so the remaining 3 extensions (
