J Neurophysiol 93: 1127–1135, 2005;
doi:10.1152/jn.00615.2004.
Invited Review
Sensory Control and Organization of Neural Networks Mediating
Coordination of Multisegmental Organs for Locomotion
Ansgar Büschges
Zoological Institute, University of Cologne, Cologne, Germany
Submitted 17 June 2004; accepted in final form 28 October 2004
INTRODUCTION
Locomotor patterns are often rhythmic or cyclic, consisting
of two main phases: a power stroke of the locomotor apparatus
that yields propulsion of the organism relative to its environment, may it be water, air, or the ground, and a return stroke
that re-establishes the starting position for the locomotor apparatus to generate the next power stroke. In walking, the
power stroke is the stance phase of a leg, and the return stroke
is the swing phase. In almost all rhythmically active locomotor
systems, it is clear that a complex rhythmic motor pattern is
generated by neural networks within the CNS, called central
pattern generators (CPGs) (Grillner 1985, 2003; Marder and
Bucher 2001; Pearson 1993, 2004), thus paralleling the situation in other rhythmic motor systems (Marder and Calabrese
1996). Numerous examples show that a functional motor
program not only depends on centrally generated basic rhythmic motor output but is also dependent on sensory signals
reporting the actual movements from the periphery (Clarac et
al. 2000; Fouad et al. 2002; Grillner and Wallén 2002; Pearson
2000, 2004). Sensory feedback can influence the magnitude
and the timing of motor activity (Fig. 1). This general scheme
based on one internal rhythm generating mechanism suffices
for the description of motor outputs that keep a given relation
in timing between their output elements, e.g., the muscles
driving a locomotor organ.
Address for reprint requests and other correspondence: A. Büschges, Dept.
Animal Physiol./Zool. Institute, Univ. of Cologne, 50923 Koeln, Germany
(E-mail: [email protected]).
www.jn.org
However, the situation is more complex when several or
multi-segmented locomotor organs have to be controlled. For
instance, walking is based on coordinated movement sequences of the legs, which themselves are multi-segmented. An
insect leg, for example (Fig. 2A), consists of more than five
segments, i.e., the coxa, the trochanter, the femur, the tibia, and
the tarsal segments, which are driven by ⬎12 muscles. To
generate steps, a specific task-dependent spatiotemporal sequence of muscle activation has to be generated so that the
movements of the leg segments are coordinated (Bässler and
Büschges 1998; Pearson 2000). Until 10 years ago, research
did not often focus on the special issues arising from controlling a multi-segmented limb. Individual influences of sensory
feedback from the limb in the generation of the locomotor
output were identified, but an overall picture of sensory-central
interaction, which could account for the generation of a coordinated walking motor output of a multi-segmented limb, was
lacking (e.g., Bässler and Büschges 1998). Two main topics
needed to be addressed. 1) How is coordination of motoneuron
activity for a multi-segmented locomotor organ generated? 2)
Does the basic organization of the central pattern generating
networks correspond to the segmentation of the locomotor
organs? Can the central neural networks, i.e., the CPG, be
regarded as one entity that is somehow reorganized for generating a range of different motor outputs (e.g., see Pearson
2004) or are there “modules” that can be coordinated differentially according to the desired motor output? This article will
review recent findings on these two issues from the stick insect
walking system and place them in a broader context together
with recent data on cat walking and lamprey swimming.
CONTROL OF THE MULTI-JOINTED STICK INSECT
MIDDLE LEG
Each of the six legs of a stick insect is controlled by an
individual leg controller, i.e., individual neural networks in the
nervous system necessary for generating a motor output for one
leg. These are located in the segmental ganglion, i.e., in the
prothoracic ganglion for the front legs, in the mesothoracic
ganglion for the middle legs, and in the metathoracic ganglion
for the hind legs (Bässler 1993; Cruse 1990; Foth and Bässler
1985; Wendler 1977). The interactions of these leg controllers
follow certain behavioral rules that generate coordinated gaits
for walking (Cruse 1990), but the neural basis of these rules is
not known. In contrast to mammalian limbs, arthropod limbs
do not move in only one plane, but instead, due to the
construction of the limbs, limb movements occur in the horizontal and vertical planes (for recent summary, see Ritzmann
et al. 2004). The morphology of the stick insect leg is shown in
Fig. 2A. The stick insect middle leg has three main leg joints:
0022-3077/05 $8.00 Copyright © 2005 The American Physiological Society
1127
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
Büschges, Ansgar. Sensory control and organization of neural networks mediating coordination of multisegmental organs for
locomotion. J Neurophysiol 93: 1127–1135, 2005; doi:10.1152/
jn.00615.2004. It is well established that locomotor patterns result
from the interaction between central pattern generating networks in
the nervous system, local feedback from sensory neurons about
movements and forces generated in the locomotor organs, and coordinating signals from neighboring segments or appendages. This
review addresses the issue of how the movements of multi-segmented
locomotor organs are coordinated and provides an overview of recent
advances in understanding sensory control and the internal organization of central pattern generating networks that operate multi-segmented locomotor organs, such as a walking leg. Findings from the
stick insect and the cat are compared and discussed in relation to new
findings on the lamprey swimming network. These findings support
the notion that common schemes of sensory feedback are used for
generating walking and that central neural networks controlling multisegmented locomotor organs generally encompass multiple central
pattern generating networks that correspond with the segmental structure of the locomotor organ.
Invited Review
1128
A. BÜSCHGES
the thoraco-coxal (TC-) joint enables back and forth movements, the coxa-trochanteral (CTr-) joint enables levation and
depression of the leg, and the femur-tibia (FTi-) joint enables
flexion and extension of the tibia. In the stick insect, there is no
trochanter-femur joint because these two segments are fused.
For simplicity, one could compare the FTi-joint with the
knee-joint of a vertebrate hind limb and describe the TC- and
CTr-joints as an outer and inner part of a joint, respectively,
that functionally corresponds to the hip joint of a vertebrate
hind limb. The neural networks controlling the individual leg
joints of each leg are located within the segmental ganglion of
a given leg. By means of tactile stimulation of head or abdomen (Bässler and Wegener 1983; Büschges et al. 2004) as well
as by the application of muscarinic agonists such as pilocarpine
(Büschges et al. 1995) to deafferented thoracic ganglia, these
networks can be activated and generate rhythmic activity in
antagonistic motoneuron pools of each leg joint. The activity of
antagonistic motoneuron pools of each leg joint is alternating,
reflecting the output of central pattern generating networks for
each leg joint. Importantly, no reliable cycle-to-cycle coupling
seems to be present between the motoneurons controlling
different leg joints (Büschges et al. 1995) (Fig. 2), suggesting
that the individual joint CPGs can operate rather independently
from each other (summary in Bässler and Büschges 1998).
Although some premotor interneurons within the CPGs have
been identified, the detailed network topology of the CPGs is
not known (Büschges 1995).
In recent years, detailed information has accumulated on
how sensory signals contribute the coordination of antagonistic
motoneuron pools that innervate one leg joint, termed intrajoint control, as well as to the coordination of the motor
activity between leg joints, termed interjoint control. Information on this issue was often gathered from preparations in
which leg proprioceptors were selectively stimulated, and their
influence on rhythmic leg motoneuron activity was monitored
when the locomotor system was activated either by tactile
stimulation or by application of pharmacological agents (e.g.,
Bässler 1993; Hess and Büschges 1999). Today, it is clear that
signals from leg receptors, reporting movement of a leg segment and load or strain on the cuticle, have access to the CPGs
of the individual leg joints and that they can induce transitions
in the activity of the individual CPGs (Akay et al. 2001; Hess
and Büschges 1999). Examples from the stick insect are given
J Neurophysiol • VOL
93 • MARCH 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
FIG. 1. Schematic drawing of the organization of the principal sensorycentral interactions in locomotor pattern generation. A central pattern generating network (CPG) depicted in a square box with 2 neurons A and B,
interacting with each other, generates a basic rhythmic output that drives sets
of antagonistic motoneurons pools and muscles (combined in square boxes and
marked with letters in italics). Feedback about the resulting limb movement is
provided by sense organs. Sensory feedback can contribute either to the control
of magnitude of motor output or to the control of timing of motor activity.
in Fig. 2, B and C. The scheme depicts each joint CPG as a box
(symbolized by 2 interneurons with mutual interactions) driving two antagonistic motoneuron pools (box). The first example shows movement and position signals from the FTi-joint
affecting the CPG governing the CTr-joint. Sensory signals
from the transducer of the knee joint, the femoral chordotonal
organ (fCO), that report extension of the tibia, can switch the
CPG of the CTr-joint from levator to depressor phase (Bucher
et al. 2003; Hess and Büschges 1999) (Fig. 2B). The reverse is
true for flexion signals from the FTi-joint. The second example
shows the influence of load signals from the trochanteral
campaniform sensillae (trCS) on the CPG of the TC-joint. This
CPG can be switched to the retractor phase by an increase in
load (Fig. 2C) (Akay et al. 2004), while the opposite switch
occurs following a decrease in load. Interestingly, in mammalian walking, movement and load signals from the limb induce
transitions between phases of motor activity (e.g., Grillner and
Rossignol 1978; Hiebert and Pearson 1999; Hiebert et al.
1996). A recent study on the stick insect has indicated that
sensory information may act through two independent neural
pathways or channels, with one pathway initiating transitions
and the second pathway determining the direction of transition
(Bässler et al. 2003).
In summary, our results indicate that the leg muscle control
system of the stick insect leg contains CPGs for each leg joint
that can be active independently. The CPGs receive input from
sense organs on the leg that can determine their phase of
activity, and thereby, can contribute to the coordination of their
actions. Figure 2D summarizes the known sensory-central
interactions identified in the stick insect middle leg by arranging them in the sequence for generation of the stepping motor
pattern typical for stepping movements in the single middle leg
preparation (Fischer et al. 2001). The scheme spans the second
half of the swing phase of the leg to the first half of the next
swing phase and divides the step into a sequence of four states
(1 to 4). It is important to note that the scheme only includes
those influences that affect timing of motor activity and does
not include the control of motor activity magnitude (cf. Fig. 1
and Bässler and Büschges 1998). The sequence of events in
brief. 1) During the swing phase of a stick insect, middle leg
levator trochanteris (Lev) and extensor motoneurons (Ext) are
active together with protractor coxae motoneurons (Pro). Sensory feedback about extension of the FTi-joint initiates depressor trochanteris (Dep) and terminates levator trochanteris (Lev)
motoneuron activity and is thus the first event in initiating
stance phase. This causes the leg to be set down (Bucher et al.
2003; Hess and Büschges 1999). 2) Ground contact of the leg
and the subsequent increase in load induces a transition in
activity in the CPG governing the TC-joint initiating retractor
coxae (Ret) activity and terminating Pro activity. This would
cause backward movement at the TC-joint when the leg is free
to move. Load signals affect knee joint motoneurons by exciting flexor tibiae (Flx) and inhibiting Ext motoneurons (Akay et
al. 2001, 2004). 3) During stance, flexion signals from the
FTi-joint reinforce flexor activity (Bässler 1986; Knop et al.
2001). 4) At a particular flexed position of the FTi-joint,
position signals induce a transition in the FTi-CPG. Flx activity
terminates, and Ext activity resumes (Bässler 1986, 1988).
Flexion of the FTi-joint can induce transition of activity in the
CTr-joint CPG as well, initiating Lev activity and terminating
Dep activity (Bucher et al. 2003; Hess and Büschges 1999). As
MOTOR PATTERN GENERATION FOR LOCOMOTION
J Neurophysiol • VOL
93 • MARCH 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
FIG. 2. Influences of proprioceptive signals from the stick insect middle leg on the timing of activity of the CPGs of the 3 main leg joints. A: schematic representation of the stick insect leg and its main
leg joints. B: movement and position signals of the knee joint (FTi-joint) have access to the CPG of the adjacent CTr-joint. CPG of each leg joint is symbolized by 2 interacting neurons in a box according
to motoneuron pools driven, e.g., retractor (Ret) and protractor (Pro) motoneurons in the case of the inner hip joint (TC-joint). Each CPG is connected schematically by arrows with 2 antagonistic sets of
motoneurons and muscles that they innervate (combined as depicted by boxes; Pro, Ret, Dep, Lev, Flx, Ext). Active elements are shown as filled symbols, inactive elements are depicted as open symbols.
Movement signals from the knee-joint can induce transitions in activity of the CPG of the CTr-joint, as exemplified for extension signals switching the CTr-CPG to the depressor phase of activity. C: signals
about changes in load affect the activity of the CPG of the TC-joint. These signals can induce a transition in activity of this CPG; e.g., increased load switches the activity of the TC-CPG from protractor to
retractor phase of activity. D: diagram summarizing all known sensory influences on the timing of motor activity for intra- and interjoint coordination in single middle leg stepping (for a detailed description,
see Akay et al. 2004; Ekeberg et al. 2004). Filled symbols denote active elements/neurons; open symbols denote inactive elements/neurons. Sensory influences on the CPGs are either excitatory (⫹) or inhibitory
(⫺). Description of the sequence of events is organized in a state-like fashion moving from the 2nd row (1) on the left to the 2nd to last row on the right (4), after which the state of the 1st row would follow
again. To exemplify this sequencing of events states, 4 and 1 are repeated at the left and right margin.
Invited Review
1129
Invited Review
1130
A. BÜSCHGES
J Neurophysiol • VOL
framework is for example also sufficient to simulate stepping
movements of front legs with only minor modifications.
How do these results on stick insect walking pattern generation compare with the known organization of other locomotor
systems, especially other walking systems? Two issues need to
be discussed. Do general principles emerge for sensory control
for walking pattern generation by comparing the results from
the stick insect to other legged systems? Is the fine structure of
the stick insect walking pattern generating system reflected in
other walking systems or other locomotor systems in general?
SENSORY CONTROL OF WALKING PATTERN
GENERATION: MOTOR PATTERN GENERATION IN
THE CAT HIND LIMB
In a variety of other legged animals, information has been
gathered on the sensory control of walking motor patterns,
among those are invertebrates, such as the cockroach, crayfish,
lobster, and locust, and vertebrates such as the mudpuppy,
mouse, rat, and humans (for reviews, see Fouad et al. 2002;
Orlovsky et al. 1999; Prochazka 1996). Presently, the most
complete knowledge of the role of sensory signals in timing
motor activity in the stepping cycle that can be compared with
the stick insect derive from previous and recent findings on the
cat hind leg.
It has been known, since the original work by Brown (1911),
that the cat lumbar region of the spinal cord can generate the
basic alternating activity in flexor and extensor muscles. Later
studies by Grillner and Zangger (1979, 1984) showed that the
motor output of the deafferented lumbar spinal cord resembled
the motor output for stepping (“fictive locomotion”; recent
summaries by Orlovsky et al. 1999; Pearson 2000). The motor
output generated expresses a well-coordinated action of the
various leg motoneuron pools of hip, knee, ankle, and foot and
was thus related to the action of one CPG situated in the lumbar
spinal cord (e.g., Pearson and Rossignol 1991; for review, see
Orlovsky et al. 1999; Rossignol 1996; but see also Yakovenko
et al. 2002). This spinal motor pattern can be drastically
modified by sensory feedback that changes its magnitude and
timing by signals from movement and force measuring sense
organs in the muscles (for review, see Orlovsky et al. 1999;
Pearson 2000). At present, the most complete picture exists for
sensory signals about position and load from the leg that can
contribute to the patterning of motor activity by influencing the
timing of the motor output in the stepping cycle. They have
been analyzed to an extent that the present knowledge can
account for the generation of a complete step cycle in the cat
hind leg.
The mechanisms influencing step cycle timing will be reviewed here in the sequence that they are activated during
stepping (Fig. 3A). 1) During leg swing, flexor motoneurons
activate the leg flexor muscles and move the leg forward.
Recent results of Pearson et al. (2003) provide evidence that
the transition from leg swing to stance, i.e., from flexor to
extensor activity is initiated by movement signals from the hip
joint. The receptors have not been identified yet, but receptors
in the hip joint or hip extensor muscles are most likely involved
(K. G. Pearson, personal communication). 2) After touch down
of the hindleg, extensor activity during leg stance is reinforced
by position and load signals from group Ia and Ib afferents of
ankle extensor muscle spindles and tendon organs (Pearson and
93 • MARCH 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
a consequence, the leg is lifted off the ground, and the accompanying decrease in load reinforces ongoing Ext activity (Akay
et al. 2001) and induces a transition in activity of the TC-joint
CPG, leading to activation of the Pro and inactivation the Ret
activity during the swing phase (Akay et al. 2004). When the
leg is free to move in the horizontal plane, as in forward
walking in vivo, position and load signals from the leg contribute to the stance-to-swing transition: When coxal sensors
signal a rather posterior position of the leg, while load on the
leg is decreasing, the leg is lifted off of the ground. This
influence has so far only been shown on the behavioral level
(Cruse 1985). Data from semi-intact walking animals on a
double-tread wheel indicate that, in forward stepping position,
signals from the leg and load signals are weighted against each
other so that a decrease in load can lead to a premature
initiation of leg swing. It should be emphasized that the outline
of the sequencing of events shown in Fig. 2D does not claim to
contain the complete set of sensory influences acting in the
control of timing motor activity in the stepping cycle of the
stick insect leg, although they have been proven to be sufficient
for this task (see next 3 paragraphs). Deficits exist in our
understanding of the sensory control of the swing-stance transition in detail. It is known to include a targeting of the leg’s
tarsus toward the position of the anterior leg’s tarsus in vivo
(Cruse 1985). In addition, there is the question of which sense
organs at the coxa report the relevant movement information
that is used for initiating the stance-swing transition, such as
the hair plates, the hair rows, an internal muscle receptor organ,
or any combination of them.
It becomes obvious from the above description that the
sensory-central interactions affecting timing of activity in leg
motoneurons are very specific. For example, sensory signals
reporting movement have only been found to contribute to
interjoint coordination between two of three leg joints and in
one direction only, i.e., from the knee joint (FTi-joint) to the
outer hip joint (CTr-joint) (Hess and Büschges 1999). No
influence of movement signals was found for the opposite
direction (Akay et al. 2001). In addition, no evidence for
movement signals in interjoint timing control has yet been
detected between any of the other leg joints (Akay et al. 2001).
A similar situation exists for load signals from the leg that
contribute to coordination of motoneuron activity (Akay et al.
2001, 2004).
From the number and variety of sensory-central interactions,
it is not possible to assess the sufficiency of this information
and its explanatory value by a purely qualitative examination
of the mechanisms. One tool in physiological research that has
proved important for the study of locomotor systems is computer simulation. In the past 10 years, simulations have paralleled and guided electrophysiological studies of locomotor
mechanisms in the spinal cord of the lamprey (e.g., Ekeberg
and Grillner 1999; Grillner et al. 2000) as well as neurobiological research on rhythmic nonlocomotor systems (e.g.,
Marder and Prinz 2002). We tested the sufficiency of the
sensory-central interactions described above with a three-dimensional (3D) neuro-mechanical simulation of the stick insect
based on its morphology (Ekeberg et al. 2004). Implementation
of the identified interactions (Fig. 2D) in the 3D neuro-mechanical simulation verified that a simulated neural controller
using these principles was able to generate coordinated stepping movements of a middle leg and that its conceptual
Invited Review
MOTOR PATTERN GENERATION FOR LOCOMOTION
1131
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
FIG. 3. A: diagram summarizing all known sensory influences on the timing of motor activity for intra- and interjoint coordination in hind leg stepping of the cat.
Spinal CPG is depicted by 2 neurons connected schematically with each other by 2 inhibitory synapses (filled circles) and with 2 antagonistic sets of motoneurons and
muscles that they innervate (combined as depicted by boxes: Flx, Ext). Active elements are shown as filled symbols; inactive elements are depicted as open symbols.
Sensory influences on the CPG are either excitatory (⫹) or inhibitory (⫺). Description of the sequence of events is organized in a state-like fashion moving from the
2nd row (1) on the left to the 2nd to last row on the right (3), after which the state of the 1st row would follow again. To exemplify this sequencing of events, states
3 and 1 are repeated at the left and right margin of the scheme. The 2 lines below the figure give sensory signals and receptors providing them, except for the case in
which they have not yet been identified (not ident.). B: comparison of sensory influences identified that influence timing of the motor output for walking movements
in the cat and the stick insect: top, sensory signals for stance-to-swing transition; middle, sensory signals for swing to stance transition; bottom, sensory signals in stance
control. Sensory parameters are given for the cat (left) and the stick insect (right). Middle: schematical representation is given that holds for both systems. In these
schemes, central neural networks generating stance and swing phase motor activity are simplified and lumped together to form 2 states, i.e., stance (st) and swing (sw).
Sensory influences are indicated and identified by arrows that either initiate (⫹) or terminate (⫺) 1 of 2 phases of motor output.
J Neurophysiol • VOL
93 • MARCH 2005 •
www.jn.org
Invited Review
1132
A. BÜSCHGES
Collins 1993; Whelan et al. 1995). 3) The termination of stance
and the transition to swing is initiated simultaneously by
position signals from the hip joint and load signals from the
ankle joint (Duysens and Pearson 1980; Grillner and Rossignol
1978; Hiebert et al. 1996; Pearson et al. 1992; Whelan et al.
1995). Unloading of the ankle extensor muscle reported by
group Ib afferents occurs at the time that the contralateral leg
initiates its stance phase and results in a decrease of the load
signals to the ipsilateral extensor half center. Simultaneously,
while the leg approaches its posterior extreme position in the
step, the flexor muscles, e.g., of the hip and ankle are further
lengthened, which is reported by their group I afferents. The
resulting sensory signals lead to an inhibition of the extensor
half center, with those from hip flexors being probably the
most relevant ones. Both load and position signals are the most
significant ones known to contribute to the termination of
extensor and initiation of flexor activity for the next swing
phase. They seem to be weighted against each other with load
being the decisive factor. This is apparent from the fact that
prolonged load signals from the leg, evoked by electrical
stimulation of group Ib afferents in the ankle extensor, can lead
to a marked prolongation of stance and shift of the posterior
extreme position (Whelan et al. 1995).
A comparison of the role of sensory signals in the timing of
motor activity in the stick insect middle leg and the cat hind leg
control systems reveal significant similarities (Fig. 3B). In both
systems, movement signals from the leg determine the transition from swing to stance by controlling the phase of central
pattern generating networks. In addition, the transition from
stance to swing is initiated by sensory signals reporting first,
unloading of the leg, and second, position of the leg, by
signaling a retracted position of the leg. Finally, in both
animals, stance phase activity is reinforced by both load and
position signals. From these results, the following general
principles emerge for the sensory control of motor output
timing during walking. The power stroke in walking (stance
phase) is reinforced and timed with respect to the biomechanical demands of the walking system to support the body (load)
and with respect to movement of the leg (position). The return
J Neurophysiol • VOL
stroke (swing phase) is timed primarily with respect to the
position of the leg relative to the body.
At present, differences between the two systems are apparent in the extent to which modularity of the central pattern
generating networks has been shown. While in the stick insect,
evidence suggests the existence of multiple central pattern
generating networks, only indirect evidence has been shown
for the cat walking system. However, recently published data
may shed some new light on this issue.
ORGANIZATION OF THE CENTRAL PATTERN
GENERATORS FOR LOCOMOTION: NEW INSIGHTS
FROM THE CAT AND THE LAMPREY
How does the modularity of the stick insect walking pattern
generating system with CPGs for each leg joint compare with
the composition of other locomotor systems? The general
scheme of sensory-central interactions outlined in the INTRODUCTION and Fig. 1 contains one central pattern generating
network as the basic kernel. It is now clear, however, that in a
variety of legged locomotor systems, the central network can
be composed of multiple central pattern generators, each controlling a subunit of the locomotor system, such as a joint,
segment, or muscle of a locomotor organ. For example, in the
mudpuppy cervical spinal cord, a modular organization has
been shown for the central pattern generating networks controlling flexor and extensor motoneuron pools of the forelimb
(Cheng et al. 1998), with each hemisegment containing one
individual CPG. In the turtle spinal cord, evidence suggests a
modular organization of the central pattern generating networks for the control of segmental motoneuron pools of the
segments controlling the hind limb (e.g., Stein et al. 1995),
generating different types of scratching patterns. Together with
the data from the stick insect, it seems that central neural
networks governing legged motor systems may exhibit a modular structure. This situation would verify the “unit burst
generator” (UBG) concept for locomotor networks formulated
by Grillner ⬎20 years ago (Grillner 1981) (Fig. 4). This
concept was based on conclusions drawn from data on motor
93 • MARCH 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
FIG. 4. A: schematic summary of results presented by Cangiano and Grillner (2003) on the modular organization of the lamprey spinal central pattern
generating network for swimming. Each hemisegment contains its own pattern generator, L and R, respectively, that is connected with the contralateral
hemisegment by mutual inhibition and with the caudal and rostral segments by intersegmental pathways (arrows). According to the apparent role of glutamatergic
excitation within the hemisegmental CPGs, each of these networks is presented with 2 interneuron pools exciting each other. B: schematic representation of the
unit burst generator (UBG) concept formulated by Grillner (1981). Each synergist muscle group is assumed to be driven by its own rhythm generating network.
Individual units controlling functional antagonistic motoneurons of 1 joint are connected with each other by mutual inhibition as well as with the other UBGs
in various ways.
Invited Review
MOTOR PATTERN GENERATION FOR LOCOMOTION
J Neurophysiol • VOL
was thought to rely crucially on mutual glycinergic inhibition
between antagonistic centers on both sides of the spinal cord
(see Grillner 2003) together with sensory feedback from the
intraspinal stretch receptors (edge cells). Their overall sensory
activity that is evoked by bending of the cord is able to entrain
the activity of the swimming CPG (recent summary in Grillner
and Wallén 2002). Generation of a coordinated motor output
between segments in the longitudinal direction of the cord is
thought to be based on the action of intersegmental projections
of the spinal interneurons (Buchanan 1999; Grillner et al. 2000;
Matsushima and Grillner 1992) and the interaction with the
sensory feedback (Grillner and Wallén 2002).
Cangiano and Grillner (2003) have now shown that each
hemisegment of the lamprey spinal cord contains a neural
network capable of generating basic rhythmic motor activity
(Fig. 3) and thus represents a UBG. They showed this by using
different levels of longitudinal lesion of the spinal cord and
electrical stimulation of individual hemisegments. While maintaining rhythmicity on both sides of the spinal cord, progressive surgical sagittal lesioning of axons crossing the midline of
the spinal cord led to a gradual increase in burst frequency,
establishing a continuum between the intact spinal cord and the
hemispinal cord preparation. It is currently assumed that hemisegmental CPGs operate on the basis of interconnected ipsilateral excitatory glutamatergic interneurons that include previously identified morphological classes of excitatory interneurons (e.g., Buchanan and Grillner 1987). Rhythmicity in the
hemispinal cord persists after removal of glycinergic inhibition, and it is currently assumed that inactivation of excitatory
interneurons during rhythmic activity arises from their intrinsic
properties (Cangiano and Grillner 2003). These results show
that each hemisegment is capable of generating rhythmic motor
activity and that the contralateral glycinergic inhibition underlies the alternation of activity on both sides of the spinal cord
as well as burst frequency reduction. Further questions now
arise as to whether coupled CPGs for each pool of segmental
motoneurons, i.e., the pools innervating the dorsal, the ventral
myotomal muscles, and the dorsal fin motoneurons (Aoki et al.
2001; Shupliakov et al. 1992) exist, and if so, how their activity
is coordinated. Finally, what role does sensory feedback play at
the level of pattern generation within a given segment.
Apart from the similar modular structure of the segmental
spinal CPGs of the walking stick insect and the swimming
lamprey, how do these data relate to our present understanding
of the stick insect walking system? In the lamprey spinal cord,
it is clear that interaction via central pathways plays a key basic
role for coordinating the action of the segmental CPGs. In the
stick insect leg control system, this prominent action of central
pathways does not appear to exist. This may be potentially
attributable to the fact that the stick insect walking system is
operating at rather slow cycle frequencies and does not require
a fast central predictive component for pattern generation on a
cycle-to-cycle basis. Pattern generation for walking over uneven ground or climbing on twigs may also be much more
dependent on “on-line” sensory feedback action to cope with
the nonhomogenous substrate compared with swimming in
water. To settle this issue, additional new approaches are
needed for the stick insect walking system that allow the
investigation of potential central coordinating pathways and
mechanisms by “driving” the walking system over a broader
range of frequency of operation.
93 • MARCH 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
pattern generation in the cat hind leg. According to the UBG
concept, each group of motoneurons may be controlled by its
own central pattern generating network.
Presently, however, direct evidence for modularity is still
lacking in the cat spinal cord system where the UBG concept
was initially postulated. Except for evidence that suggests that
the cat locomotor CPG is formed by two half-centers on each
side of the spinal cord inhibiting each other (see also Orlovsky
et al. 1999), the detailed organization and circuitry of the CPG
for the cat hind limb is not known. Bouyer and Rossignol
(2003a,b) analyzed the role of cutaneous input from the cat
hind paw on the generation of the stepping motor output. They
report that, in spinal cats, denervation of the hind paw caused
a permanent deterioration of the motor pattern. While flexor
and extensor muscles of hip and knee are still activated
(postspinalization) in an alternating fashion during stance and
swing, flexor and extensor muscles of the ankle and foot were
abnormally co-activated during stance. This is a marked difference compared with the activity of these muscles during
intact walking or following postdeafferentation regeneration,
because their activity is completely shifted into the same
(stance) phase of the step cycle. However, on generating fictive
locomotion in the deafferented spinal cord, alternating motor
activity is produced, indicating that the abnormal motor output
during walking is not the result of a permanent reorganization
of that CPG that is activated during fictive locomotion.
These results from the spinal walking cat were unexpected
because none of the previously identified sensory-central interactions can explain a simultaneous in-phase activation of
ankle and toe extensor and flexor muscles during stance (see
Pearson 2000). This may indicate that afferent feedback not
only interacts with the presumed single hind limb CPG, but
that it may exert a much more powerful influence on the motor
output by uncoupling motoneuron activity from CPG drive.
Therefore it may exhibit an organization similar to the schemes
described above for the stick insect. These results highlight the
importance of determining the organization of the cat spinal
walking CPG. One approach might be to combine differential,
competitive activation of the known powerful sensory influences of movement and load signals from leg joints (hip and
ankle joints; Grillner and Rossignol 1978; Hiebert et al. 1996;
Whelan et al. 1995) to challenge this CPG. In addition to such
approaches, fundamental new insights into the organization of
the mammalian hind limb CPG can be expected from recent
investigations combining electrophysiological studies of mammalian spinal networks for locomotion with the use of developmental markers for neuron populations in neonatal rodents
(Butt and Kiehn 2003).
The number of locomotor systems in which a modular
organization has been proven is steadily increasing. In the
lamprey spinal cord, evidence for the existence of hemisegmental central pattern generators was recently presented. The
undulatory swimming movement of the lamprey is generated
by a rostral-to-caudal wave of rhythmic activation of dorsal
and ventral myotomal muscles and fin musculature (for description of basic cell types, see Rovainen 1974). The basic
motor pattern of the myotomal musculature is based on a CPG
that generates alternating left-right activity in motoneuron
pools innervating the myotomal musculature on both sides of
each segment (recent summaries in Buchanan 2001; Grillner et
al. 2000; Orlovsky et al. 1999). Pattern generation of the CPG
1133
Invited Review
1134
A. BÜSCHGES
ACKNOWLEDGMENTS
I thank R. A. DiCaprio, A. ElManira, S. Grillner, K. G. Pearson, T. Mentel,
H. Scharstein, and J. Schmidt for comments on earlier versions of the
manuscript. In particular, I thank T. Akay, D. Bucher, and D. Hess, whose
passion for research was the basis of new findings on the stick insect walking
system reported here.
GRANTS
This work was supported by various Deutsche Forschungs Gemeinschaft
grants (Bu857), the Institute for Advanced Study Berlin, and the Boehringer
Ingelheim Foundation.
REFERENCES
Akay T, Bässler U, Gerharz P, and Büschges A. The role of sensory signals
from the insect coxa-trochanteral joint in controlling motor activity of the
femur-tibia Joint. J Neurophysiol 85: 594 – 604, 2001.
Akay T, Haehn S, Schmitz J, and Büschges A. Signals from load sensors
underlie interjoint coordination during stepping movements of the stick
insect leg. J Neurophysiol 92: 42–57, 2004.
Aoki F, Wannier T, and Grillner S. Slow dorsal-ventral rhythm generator in
the lamprey spinal cord. J Neurophysiol 85: 211–218, 2001.
Bässler U. Afferent control of walking movements in the stick insect Cuniculina impigra. II. Reflex reversal and the release of swing phase in the
restrained foreleg. J Comp Physiol A 158: 351–362, 1986.
Bässler U. Functional principles of pattern generation for walking movements
of stick insect forelegs: the role of the femoral chordotonal organ afferences.
J Exp Biol 136: 125–147, 1988.
Bässler U. The walking- (and searching-) pattern generator of stick insects, a
modular system composed of reflex chains and endogenous oscillators. Biol
Cybern 69: 305–317, 1993.
Bässler U and Büschges A. Pattern generation for stick insect walking
movements-multisensory control of a locomotor program. Brain Res Rev 27:
65– 88, 1998.
Bässler U, Sauer AE, and Büschges A. Vibration signals from the FT-joint
can induce phase-transitions in motoneuron pools of the stick insect walking
system. J Neurobiol 56: 125–138, 2003.
Bässler U and Wegener U. Motor output of the denervated thoracic ventral
nerve cord in the stick insect, Carausius morosus. J Exp Biol 105: 127–145,
1983.
Bouyer LJG and Rossignol S. Contribution of cutaneous inputs from the
hindpaw to the control of locomotion. I. Intact cats. J Neurophysiol 90:
3625–3639, 2003a.
Bouyer LJG and Rossignol S. Contribution of cutaneous inputs from the
hindpaw to the control of locomotion. II. Spinal cats. J Neurophysiol 90:
3640 –3653, 2003b.
Brown TG. The intrinsic factors in the act of progression in the mammal. Proc
R Soc Lond B 84: 308 –319, 1911.
J Neurophysiol • VOL
Buchanan JT. Commissural interneurons in rhythmic generation and intersegmental coupling in the lamprey spinal cord. J Neurophysiol 81: 2037–
2045, 1999.
Buchanan JT. Contributions of identifiable neurons and neuron classes to
lamprey vertebrate neurobiology. Prog Neurobiol 63: 441– 446, 2001.
Buchanan JT and Grillner S. Newly identified ‘glutamate interneurons’ and
their role in locomotion in the lamprey spinal cord. Science 236: 312–314,
1987.
Bucher D, Akay T, DiCaprio RA, and Büschges A. Interjoint coordination
in the stick insect leg-control system: the role of positional signaling.
J Neurophysiol 89: 1245–1255, 2003.
Büschges A. Role of local nonspiking interneurons in the generation of
rhythmic motor activity in the stick insect. J Neurobiol 27: 488 –512, 1995.
Büschges A, Ludwar BC, Bucher D, Schmidt J, and DiCaprio RA.
Synaptic drive contributing to rhythmic activation of motoneurons in the
deafferented stick insect walking system. Eur J Neurosci 19: 1856 –1862,
2004.
Büschges A, Schmitz J, and Bässler U. Rhythmic patterns in the thoracic
nerve cord of the stick insect induced by pilocarpine. J Exp Biol 198:
435– 456, 1995.
Butt SJB and Kiehn O. Functional identification of interneurons responsible
for left-right coordination of hindlimbs in mammals. Neuron 38: 953–963,
2003.
Cangiano L and Grillner S. Fast and slow locomotor burst generation in the
hemi-spinal cord of the lamprey. J Neurophysiol 89: 2931–2942, 2003.
Cheng J, Stein RB, Jovanovic K, Yoshida K, Bennett DJ, and Han Y.
Identification, localization and modulation of neural networks for walking in
the mudpuppy (Necturus macalatus) spinal cord. J Neurosci 18: 4295– 4304,
1998.
Clarac F, Cattaert D, and Le Ray D. Central control components of a
‘simple’ stretch reflex. Trends Neurosci 23: 199 –208, 2000.
Cruse H. Which parameters control the leg movement of a walking insect? II.
The start of the swing phase. J Exp Biol 116: 357–362, 1985.
Cruse H. A quantitative model of walking incorporating central and peripheral
influences. Biol Cybern 39: 131–136, 1990.
Duysens J and Pearson KG. Inhibition of flexor burst generation by loading
ankle extensor muscles in walking cats. Brain Res 187: 312–332, 1980.
Ekeberg Ö, Blümel M, and Büschges A. Dynamic simulation of insect
walking. Arthropod Struct Dev 33: 287–300, 2004.
Ekeberg Ö and Grillner S. Simulations of neuromuscular control in lamprey
swimming. Phil Trans R Soc Lond B 354: 895–902, 1999.
Fischer H, Schmidt J, Haas R, and Büschges A. Pattern generation for
walking and searching movements of a stick insect leg: 1. Intra- and
interjoint coordination of motor activity in a single middle leg preparation.
J Neurophysiol 85: 341–353, 2001.
Foth E and Bässler U. Leg movements of stick insects walking with five legs
on a tread wheel and with one leg on a motor-driven belt (I and II). Biol
Cybern 51: 319 –324, 1985.
Fouad K, Fischer H, and Büschges, A. Comparative locomotor systems. In:
Comprehensive Handbook of Psychology, Volume 3: Biological Psychology,
edited by Gallaghar M and Nelson R. New Jersey: Wiley, 2002, p. 109 –137.
Grillner S. Control of locomotion in bipeds, tetrapods and fish. In: Handbook
of Physiology—The Nervous System II. Bethesda, MD: Am Physiol Soc.,
1981, p. 1179 –1236.
Grillner S. Neurobiological bases of rhythmic motor acts in vertebrates.
Science 228: 143–149, 1985.
Grillner S. The motor infrastructure: from ion channels to neuronal networks.
Nature Rev 4: 573–586, 2003.
Grillner S, Cangiano L, G-Y Hu, Thompson R, Hill R, and Wallén P. The
intrinsic function of a motor system-from ion channels to network and
behavior. Brain Res 886: 224 –236, 2000.
Grillner S and Rossignol S. On the initiation of the swing phase of locomotion in chronic spinal cats. Brain Res 146: 269 –277, 1978.
Grillner S and Wallén P. Cellular bases of a vertebrate locomotor systemsteering, intersegmental co-ordination and sensory control. Brain Res Rev
40: 92–106, 2002.
Grillner S and Zangger P. On the central generation of locomotion in the low
spinal cat. Exp Brain Res 34: 241–261, 1979.
Grillner S and Zangger P. The effect of dorsal root transection on the efferent
motor pattern in the cat’s hindlimb during locomotion. Acta Physiol Scand
120: 393– 405, 1984.
Hess D and Büschges A. Role of proprioceptive signals from an insect
femur-tibia joint in patterning motoneuronal activity of an adjacent leg joint.
J Neurophysiol 81: 1856 –1865, 1999.
93 • MARCH 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
Two conclusions can be drawn about the neural networks
controlling multi-jointed locomotor organs. 1) In animal walking, general principles for the control of the stance phase and
the step-phase transitions can be proposed from studying the
cat and the stick insect. 2) Evidence from systems controlling
multi-segmented locomotor organs support the notion that
central neural networks contain multiple CPGs corresponding
with the segmental structure of the locomotor system. Why is
this proposition significant? One striking characteristic of all
locomotor systems, including those that operate multi-jointed
locomotor organs, is their ability to generate a highly flexible
and adaptable motor output, enabling the animal, for example,
to walk forward or backward or to turn. We need to know the
internal organization of the central pattern generating machinery to be able to develop experimental approaches to study the
underlying neural mechanisms responsible. This will allow us
eventually to target and analyze those neural mechanisms that
allow their task-specific flexibility of motor outputs.
Invited Review
MOTOR PATTERN GENERATION FOR LOCOMOTION
J Neurophysiol • VOL
Pearson KG, Ramirez JM, and Jiang W. Entrainment of the locomotor
rhythm by group Ib afferents from ankle extensor muscles in spinal cats. Exp
Brain Res 90: 557–566, 1992.
Pearson KG and Rossignol S. Fictive motor patterns in chronic spinal cats
J Neurophysiol 66: 1874 –1887, 1991.
Prochazka A. Proprioceptive feedback and movement regulation. In: Handbook of Physiology. Exercise: Regulation and Integration of Multiple
Systems. New York: American Physiological Society, 1996, p. 89 –127.
Ritzmann RE, Quinn RD, and Fischer M. Convergent evolution and
locomotion through complex terrain by insects, vertebrates and robots.
Arthropod Struct Dev 33: 361–379, 2004.
Rossignol S. Neural control of stereotypic limb movements. In: Handbook of
Physiology. New York: Oxford Univ Press, 1996, chapt. 5, p. 173–216.
Rovainen CM. Synaptic interactions of identified nerve cells in the spinal cord
of the sea lamprey. J Comp Neurol 145: 189 –206, 1974.
Shupliakov O, Wallén P, and Grillner S. Two types of motorneurons
supplying dorsal fin muscles in lamprey and their activity during fictive
locomotion. J Comp Neurol 321: 112–123, 1992.
Stein PSG, Victor JC, Field EC, and Currie SN. Bilateral control of
hindlimb scratching in the spinal turtle: contralateral spinal circuitry contributes to the normal ipsilateral motor pattern of fictive scratching. J Neurosci 15: 4343– 4355, 1995.
Wendler G. Erzeugung und kontrolle koordinierter bewegungen bei Tieren—
Beispiele an Insekten. In: Kybernetik, edited by Hauske G and Butenandt E.
München: Oldenbourg, 1977, p. 11–34.
Whelan PJ, Hiebert GW, and Pearson KG. Stimulation of the group I
extensor afferents prolongs the stance phase in walking cats. Exp Brain Res
103: 20 –30, 1995.
Yakovenko S, Mushahwar V, VanderHorst V, Holstege G, and Prochazka
A. Spatiotemporal activation of lumbosacral motoneurons in the locomotor
step cycle. J Neurophysiol 87: 1542–1553, 2002.
93 • MARCH 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
Hiebert G and Pearson KG. The contribution of sensory feedback to the
generation of extensor activity during walking in decerebrate cats. J Neurophysiol 81: 758 –770, 1999.
Hiebert G, Whelen PJ, Prochazka A, and Pearson KG. Contributions of
hindlimb flexor muscle afferents to the timing of phase transitions in the cat
step cycle. J Neurophysiol 75: 1126 –1137, 1996.
Knop G, Denzer L, and Büschges A. The role of central bistable networks in
generating reflex reversal in the insect leg muscle control system. J Neurophysiol 86: 3065–3068, 2001.
Marder E and Bucher D. Central pattern generators and the control rhythmic
movements. Curr Biol 11: 986 –996, 2001.
Marder E and Calabrese R. Principles of rhythmic motor pattern generation.
Physiol Rev 76: 687–717, 1996.
Marder E and Prinz AA. Modeling stability in neuron and network function:
the role of activity in homeostasis. Bioessays 24: 1145–1154, 2002.
Matsushima T and Grillner S. Neural mechanisms of intersegmental coordination in lamprey: local excitability changes modify the phase coupling
along the spinal cord. J Neurophysiol 67: 373–388, 1992.
Orlovsky GN, Deliagina TG, and Grillner S. Neuronal Control of Locomotion. Oxford: Oxford University Press, 1999.
Pearson KG. Common principles of motor control in vertebrates and invertebrates. Annu Rev Neurosci 16: 265–297, 1993.
Pearson KG. Neural adaptation in the generation of rhythmic behavior. Ann
Rev Physiol 62: 723–753, 2000.
Pearson KG. Generating the walking gait: role of sensory feedback Prog
Brain Res 143: 123–129, 2004.
Pearson KG and Collins DF. Reversal of the influence of group Ib afferents
from plantaris on activity in medial gastrocnemius muscle during locomotor
activity. J Neurophysiol 79: 1009 –1017, 1993.
Pearson KG, Donelan JM, and McVea D. Initiation of Stance Phase Activity
in Walking Cats: A Possible Role for Hip Afferents. Washington, DC:
Society for Neuroscience, 2003.
1135