Kinematic and EMG Determinants in Quadrupedal Locomotion of...
Transcript of Kinematic and EMG Determinants in Quadrupedal Locomotion of...
In press – Journal of Neurophysiology
Kinematic and EMG Determinants in Quadrupedal
Locomotion of a Non-Human Primate (Rhesus) Contributors:
Grégoire Courtine 1, Roland R. Roy 2, John Hodgson 1, 2, Heather McKay 3,
Joseph Raven1, Hui Zhong 1, Hong Yang 4, Mark H. Tuszynski 4, 5 and V.
Reggie Edgerton 1, 2
Affiliations: 1. Department of Physiological Science, University of California, Los Angeles, California.
2. Brain Research Institute, University of California, Los Angeles, California.
3. California National Primate Research Center (CNPRC), University of California, Davis,
California.
4. Department of Neurosciences, University of California, San Diego, La Jolla, California.
5. Veterans Affairs Medical Center, San Diego, California.
Running title:
Kinematic and EMG patterns during Rhesus locomotion
CA Correspondence: V. Reggie Edgerton, Ph.D.
Dept. of Physiological Science
University of California, Los Angeles
405 Hilgard Ave.
Los Angeles, CA 90095-1527.
310-825-1910 (phone)
310-267-2071 (fax)
Kinematic and EMG patterns during rhesus monkey locomotion
p.2
Acknowledgments This work was supported by the National Institutes of Health (Grant Number: RO1-NS42291) and the California Roman
Reed Bill.
Courtine et al.
p.3
Abstract
We hypothesized that the activation patterns of flexor and extensor muscles and the resulting kinematics of the forelimbs
and hindlimbs during locomotion in the Rhesus would have unique characteristics relative to other quadrupedal mammals.
Adaptations of limb movements and in motor pool recruitment patterns in accommodating a range of treadmill speeds
similar to other terrestrial animals in both the hindlimb and forelimb were observed. Flexor and extensor motor neurons
from motor pools in the lumbar segments, however, were more highly coordinated than in the cervical segments. Unlike the
lateral sequence characterizing subprimate quadrupedal locomotion, non-human primates use diagonal coordination
between the hindlimbs and forelimbs, similar to that observed in humans between the legs and arms. Although there was a
high level of coordination between hindlimb and forelimb locomotion kinematics, limb-specific neural control strategies
were evident in the inter-segmental coordination patterns and limb endpoint trajectories. Based on limb kinematics and
muscle recruitment patterns, it appears that the hindlimbs, and notably the distal extremities, contribute more to body
propulsion than the forelimbs. Furthermore, we found adaptive changes in the recruitment patterns of distal muscles in the
hindlimb and forelimb with increased treadmill speed that likely correlate with the anatomical and functional evolution of
hand and foot digits in monkeys. Changes in the properties of both the spinal and supraspinal circuitry related to stepping,
probably account for the peculiarities in the kinematic and EMG properties during non-human primate locomotion. We
suggest that such adaptive changes may have facilitated evolution toward bipedal locomotion.
Kinematic and EMG patterns during rhesus monkey locomotion
p.4
Introduction
Mammalian locomotion is characterized by repetitive and stereotypic limb movements associated with highly-structured
patterns of muscle activity. Examination of the features of muscle synergies underlying stepping-related limb movements
has provided important information regarding the functions of the skeletal musculature (Roy et al. 1991b; Zajac et al. 2003;
Zernicke and Smith 1996), and the organization of the neural processes that drive motor neuron activity during locomotion
(Grillner 1981; Ivanenko et al. 2004b; Rossignol 1996). Likewise, locomotion kinematics has provided insight into the
neural strategies by which the CNS coordinates the oscillations of the limbs and the trunk during walking in both cats (Shen
and Poppele 1995) and humans (Courtine and Schieppati 2004; Lacquaniti et al. 2002).
There are relatively few data on the kinematic and electromyographic (EMG) determinants for non-human primate
locomotion (Mori et al. 1996; Recktenwald et al. 1999). For example, little is known about the pattern of hindlimb (HL)
and forelimb (FL) muscle activation during walking, nor about how the recruitment of these motor neuron pools is
modulated with respect to the speed of locomotion. Similarly, there is virtually no information on how the CNS of monkeys
coordinates the oscillation of HL and FL segments during gait, and adjusts the structure of inter-segmental coordination
patterns to increasing locomotor velocities.
Inter-species comparisons of the kinematic and EMG characteristics of locomotor control have highlighted many
similarities in the features of the motor program for walking among mammals, thereby supporting the idea of robust
conservation of the neural strategies that control terrestrial locomotion (Lacquaniti et al. 1999; Orlovskii et al. 1999). On
the other hand, important differences have been emphasized. Notably, erect bipedal stepping encompasses characteristics in
limb kinematics and muscle activity patterns that are unique to human locomotion (Capaday 2002). Interestingly,
laboratory-based anthropological studies show that non-human primate quadrupedalism exhibits a variety of features that
distinguish it from that observed in most other mammals (Larson 1998; Schmitt 2003). One view is that these alterations in
the organization of gait in primates necessitated adaptations in the underlying neurological control mechanisms, including
an increased dependence of spinal circuits upon supraspinal modulating commands (Capaday 2002; Duysens and Van de
Crommert 1998; Eidelberg et al. 1981; Fedirchuk et al. 1998; Nielsen 2002; Vilensky and O'Connor 1998). Nevertheless,
no conclusive evidence for non-human primate-specific neural control mechanisms for stepping has been obtained,
principally because of the lack of detailed information on the neuromechanics of locomotion under carefully controlled
conditions. This information is critical, however, since comparative features of the neural architecture for monkey and
human locomotion would have major implications for understanding the evolution toward bipedal locomotion in humans.
This information also would have direct relevance in the use of non-human primates to formulate interventions to enhance
motor recovery following spinal cord injury (SCI). The degree to which interventions developed in cats (Edgerton et al.
2001; Rossignol et al. 2004) and rats (Jones et al. 2001) will be effective in larger animals and in primates continues to be
an open question. The effectiveness of these interventions may depend upon the extent of the similarities in the functional
and anatomical organization of the motor infrastructure between lower and higher mammals (Edgerton and Roy 2002;
Tuszynski et al. 2002).
Courtine et al.
p.5
In the current study, we detail the spatial and temporal characteristics of the gait pattern in the intact Rhesus, and show how
these features are adjusted with increasing locomotor velocities. We document for the first time the kinematics of both the
HLs and FLs and their associated patterns of muscle activity during quadrupedal locomotion on a treadmill over a range of
speeds. Our objectives were 1) to compare the kinematic and EMG features of walking in a non-human primate with sub-
primate quadrupedal mammals and with humans; and 2) to provide a solid baseline to identify the effects of selective
interruptions of descending and ascending pathways on Rhesus locomotor control (Courtine et al. 2004), and the putative
recovery of motor function that can occur following selected interventions to enhance this recovery (Yang et al. 2004). We
hypothesized that the activation patterns of flexor and extensor muscles and the resulting kinematics of the FLs and HLs
during locomotion in the Rhesus would reflect some ability to decouple the interdependence of selected motor pools
relative to other quadrupedal mammals, and that these characteristics would be consistent with the evolution of bipedal
locomotion in primates. Although our results show that Rhesus monkeys share a number of similarities in the organization
of gait with other mammals, there are some unique features in their stepping-related kinematic and muscle activation
patterns. We suggest that these differences are related to adaptation of non-human primate gait to the arboreal environment,
and the evolutionary forces that led to the evolution of bipedalism in human primates.
Kinematic and EMG patterns during rhesus monkey locomotion
p.6
Materials and Methods
Experimental subjects and treadmill training protocol Six adult (7 to 15 years of age, 11.2 kg (range, 7-14 kg) body weight), male Rhesus monkeys (Macaca mulatta) were
studied. Each animal was trained to walk quadrupedally on a motor-driven treadmill. A plexiglass enclosure was used to
maintain the animal in position while allowing video recording of their movements. The initial training sessions were used
to acclimatize the animals to the treadmill environment without the belt moving. Subsequent sessions were used to train the
animals to locomote consistently at speeds of 0.45, 0.89, 1.34, and 1.79 m/s. Each animal was trained for a minimum of one
month before any locomotor data were collected. Each training session consisted of eight locomotor trials (2 repetitions at
each speed) with approximately a one-min rest period between each trial. The duration of each session was ~40 min. A
variety of food items were used as rewards after each locomotor trial.
Surgical Procedures
After the training period, the six clinically normal Rhesus were implanted with bipolar intramuscular EMG electrodes
under aseptic conditions. The Rhesus were housed individually in standard 4.3 or 6.1 square feet stainless steel cages. Prior
to any surgical procedures, the Rhesus were trained to wear a specially designed jacket that would protect exteriorized
instrumentation. EMG electrode arrays similar to those described by Hodgson et al. (2001) were purchased from a
commercial source (Model TK-12, Konigsberg Instruments, Pasadena, CA, USA). The EMG implants were manufactured
from teflon-coated multistrand stainless steel wire (32 gauge; Cooner Wire, Chatsworth, CA). The wires from the EMG
implants were embedded in silicon rubber in 1.5 mm silicon rubber tubes, and terminated in a small multipin connector (a
skin button) attached to the skin between the scapulae. The Rhesus were housed and all surgical procedures were
performed at the California National Primate Research Center (CNPRC, University of California, Davis, CA, USA). The
following anesthesia regimen was followed. Preoperative management consisted of food restriction for ~8 h. Induction of
anesthesia was with ketamine HCl (10 mg/kg i.m.). Atropine sulfate (0.04 mg/kg i.m.) was administered during induction.
A catheter was placed in either the saphenous or cephalic vein to supply fluids during the procedure, and a tracheal tube
was placed to give a free airway for gas anesthesia. Anaesthesia was maintained with isoflurane gas (1.25%) in 100%
oxygen delivered via a cuffed orotracheal tube. Throughout the surgery, a trained animal health technician monitored heart
rate, blood pressure, O2 saturation, CO2 expiration levels, core body temperature, respiratory rate, respiratory pressure, and
tidal volume using a surgical Ohmeda-Datex unit. Lactated Ringers solution (10 ml/kg/h) was administered at a continuous
infusion rate for the duration of anesthesia. Prior to any incisions being made, the depth of anesthesia was assessed by
checking heart rate, blood pressure, jaw tone, and toe-pinch response. Adjustments in the level of anesthesia were made as
needed.
Selected HL muscles were implanted in Rhesus #1, #2, and #3 and selected FL muscles were implanted in Rhesus #4, #5
and #6. Table 1 identifies the muscles implanted in each animal, and lists the main actions of each muscle as well as the
muscle abbreviations used throughout the text. Rhesus #4 rejected the implants, and only kinematic data (see below) were
recorded from this animal. A ~4-cm incision was made at the midline of the upper back on line with the caudal border of
Courtine et al.
p.7
the scapula. For the HL implants, skin incisions (~4-6 cm) were made over the bellies of the VL, the triceps surae and the
TA. The bellies of the VL, MG, Sol, FDL, FHL, EDL and/or TA in the right leg and the Sol, MG and/or TA in the left leg
were exposed as clearly as possible. Using a smooth rod, the EMG wires were routed subcutaneously from the back
incision to the appropriate locations in the HL. Bipolar intramuscular EMG electrodes were inserted into the medial,
midbelly of the Sol, the distal, medial deep region of the MG, the midbelly of the TA, the lateral midbelly of the FHL, the
lateral midbelly of the FDL, the lateral distal portion of the EDL, and/or the lateral, deep region of the VL using procedures
described in detail previously (Hodgson et al. 2001). For the FL implants, skin incisions were made over the bellies of the
Bic, Tri, palmaris longus, EDC and the thenar eminence. The wires were routed to the incision sites as described above
and EMG electrodes were implanted in the right medial midbelly of the medial head of the Tri, lateral midbelly of the long
head of the Bic, lateral midbelly of the FDS, and lateral midbelly of the FDP, and in the right and left lateral midbelly of
the EDC and midbelly of the FPB. The EMG wires were coiled near each implant site to provide stress relief. Back
stimulation through the skin button (see below) was used to verify the proper placement of the electrodes in each muscle. In
addition, the electrode placement was verified in a terminal experiment at the end of the study.
A small incision then was made approximately one inch lateral to the upper back incision. A skin button was passed
through the hole and the skin was sutured around the button. To provide stress relief, the wires were looped subcutaneously
near the skin button. All incision areas were irrigated liberally with warm, sterile saline and closed in layers, i.e., investing
fascia and then the skin. All closed incision sites were cleansed thoroughly with saline solution. Postoperative care
consisted of intensive monitoring until the monkeys regained their equilibrium and were able to sit upright. Analgesia was
provided by either oxymorphone (0.15 mg/kg, IM, TID) or buprenex (0.5–1.0 mg/kg im, TID). The analgesics were
initiated prior to completion of the surgery and continued for a minimum of 3-5 days. The monkeys were monitored closely
for food and water intake and were supplemented liberally with fresh fruit and vegetables on a daily basis until a proper
appetite resumed. Antibiotic therapy with cephazolin (20 mg/kg im, TID) or cephalexin (30 mg/kg, oral, BID) was initiated
preoperatively (given every two hours during the procedure) and continued for 5-10 days. Wound healing was monitored
closely by a dedicated veterinary and therapeutics staff. Initial care immediately after surgery consisted of monitoring the
transcutaneous exit sites for erythema or exudation. If deterioration of the exit sites was noted, the incision sites were
cleansed with dilute Novalsan solution (chlorohexadine) and topical antibiotics were used if deemed necessary by the
veterinarian staff. If necessary, systemic antibiotic therapy was initiated with the drug of choice from the case veterinarian.
All surgical and experimental procedures in these experiments were carried out using the principles outlined by the
Laboratory Animal Care (National Institutes of Health Publication 85-23, revised 1985) and were approved by the
Institutional Animal Care and Use Committee (IACUC). Testing protocols and data collection After a recuperation period of ~2-3 weeks, kinematic data and EMG activity were recorded under the same experimental
conditions as during the training procedures.
Kinematics. Video recordings (60 Hz) were made using one (Rhesus #1, #2, and #3) or two (Rhesus #4, #5 and #6)
cameras (Panasonic System Camera, WV D5100; Panasonic AG1280P Panasonic, Cypress, CA, USA) oriented
perpendicular to (one camera), or at 45o and 135o (two cameras) with respect to the direction of the locomotion, i.e., the
Kinematic and EMG patterns during rhesus monkey locomotion
p.8
animal’s sagittal plane. Before each testing session, a calibration device was placed in the treadmill and recorded. Non-
toxic white paint was used to mark shaved areas of skin overlying the following body landmarks (right side): for the HL,
the greater trochanter (GT), the knee joint (K), the malleolus (M), the 5th metatarsal (MT) and the outside tip (T) of the fifth
digit; for the FL, the head of the humerus (H), the elbow joint (E), the distal head of the ulna (U), the metacarpo-phalangeal
(MCP) joint, and the outside tip of the third digit (D) (see Fig. 3). The body was modeled as an interconnected chain of
rigid segments: GT-K for the thigh; K-M for the shank; M-MT for the foot; MT-T for the fifth digit; GT-H for the trunk; H-
E for the arm; E-U for the forearm; U-MCP for the hand; MCP-D for the third digit. In addition, the limb axis was defined
as the virtual line connecting the hip to the MT joint, and the shoulder to the MCP joint for the HL and FL, respectively.
EMG. Two telemeters (Model T-47, Konigsberg Instruments, Pasadena, CA, USA) were attached to the EMG skin button
and placed in two large pouches on the back of the monkey’s jacket. The telemeters weighed ~150 g and did not appear to
interfere with the performance of the locomotor task. Output from the telemetry receiver was recorded at 2 kHz on FM tape
(TEAC Model XR-510, TEAC, Montebello, CA, USA). A Society of Motion Picture and Television Engineers (SMPTE)
time code generator (model F30, Fast Forward Video, Irvine, CA, USA) was used to synchronize video frames with the
EMG signals recorded on FM tape for Rhesus #5 and #6.
Data processing Kinematics. Selected video recordings were digitized with a video grabber card and recorded to disc. The Motus software
(Peak Performance Technologies, Inc., Centennial, CO, USA) was used to automatically detect the centroid of the (x, y)
coordinates of the reflecting points attached to the skin of the monkey. We used these (x, y) coordinates to reconstruct the
trajectory of the limb and to calculate joint angles at the hip, knee, ankle, MT joint, shoulder, elbow, wrist, and MCP joint
(see Fig. 3). Flexion, plantarflexion (MT), ventro-flexion (wrist, MCP), and retraction (shoulder) were defined as a
decrease in the measured angle. The angle of each segment with respect to the direction of gravity (elevation angle) in the
sagittal plane also was computed. These angles were positive in the forward direction, i.e., when the distal marker crossed
the vertical line passing through the proximal marker.
Spatial and temporal features of the gait pattern. The gait cycle was defined as the time interval between two successive
paw contacts of one limb. Successive paw contacts were visually defined by the investigators with an accuracy of ± 1 video
frame. Ten or more successive, consistent HL and FL gait cycles were typically recorded from each animal at each
treadmill speed. Swing phase onset was set at the zero crossing of the rate of change of the elevation angle of the limb axis,
i.e., at the onset of forward oscillation (Borghese et al. 1996; Courtine and Schieppati 2004). Cycle duration was computed
for each limb and stance and swing phase durations were expressed as a percentage of the cycle duration. A gait diagram
was constructed for each trial and the timing of HL and FL displacements with respect to the right HL gait cycle was
determined (Fig. 1). Stride length for each limb was considered to be the linear spatial distance between the malleolus
position at successive paw contacts plus the cycle duration multiplied by the treadmill speed. Consequently, the measured
stride length directly reflected the actual forward displacement of the limb during a complete gait cycle. We also computed
the mean body (limb) speed during each gait cycle as the stride length divided by the cycle duration. We introduced this
difference between treadmill speed and mean body speed to take into account the variability in the limb movements (see
Courtine et al.
p.9
Fig. 2B and C) and the possible drifting of the animal on the moving treadmill belt. The location of the foot with respect to
the hand at foot contact (stance onset) was measured as the distance between the position of the HL (toe, T) and FL (finger,
F) endpoint markers.
Average angular waveforms. Each data set was time-interpolated over individual gait cycles on a time base with 100
points. Averages were constructed for all gait cycles at each speed for each animal.
Velocity-curvature power law. To compute the velocity-curvature relationship, limb endpoint spatial coordinates
corresponding to the swing phase of a given limb at a given speed were extracted and pooled. We performed a linear
regression analysis in log-log scales of the equation:
βω )()( tCKt ⋅=
where ω(t) and C(t) are the instantaneous values of the angular velocity and the path curvature of the limb endpoint,
respectively, K is a velocity gain factor that depends on overall movement duration, and β is the power exponent. In
logarithmic scales, a power function becomes a straight line whose slope corresponds to the exponent (Ivanenko et al.
2002).
Inter-segmental coordination. General procedures have been described elsewhere (Courtine and Schieppati 2004).
Briefly, the timing of HL and FL segment oscillations in the sagittal plane was determined through Fast Fourier
Transformation (FFT). The phase φ of the first order Fourier series component of each angle was taken as the timing of its
oscillation during a given gait cycle. This timing was expressed in percent with respect to the normalized gait cycle
duration (φ * 100/2π).
Principal component analysis. We used principal component (PC) analysis to quantify the spatio-temporal structure of the
inter-segmental coordination among body segments (Bianchi et al. 1998; Borghese et al. 1996; Courtine and Schieppati
2004). For each set of trial data, the analysis was performed by computing the covariance matrix A of the ensemble of
selected time-varying angles over the gait cycle, after subtraction of their respective mean values. The PCs were computed
from eigenvalues λj and eigenvectors Uj of A. The PCs were ordered according to the amount of data variance accounted
for by each component:
∑=
n
ii
j
1
λ
λ
The number of components required to account for data variance was used to provide insight into the complexity of the
spatio-temporal structure of the underlying inter-segmental coordination pattern. PC analysis was carried out on time-
varying elevations of the HL and FL segments, either separately (n = 4) or together (n = 8). To allow for direct comparisons
with data previously reported in humans (Borghese et al. 1996; Courtine and Schieppati 2004), we also applied the PC
analysis on thigh, leg and foot elevations angles for the HL, and arm, forearm, and hand elevation angles for the FL. The
variance accounted for by PC1 plus PC2 (Borghese et al. 1996) quantified the extent of planarity in the co-variation among
the HL or FL segment oscillations (Fig. 6).
Kinematic and EMG patterns during rhesus monkey locomotion
p.10
Time shift of limb oscillation with increases in treadmill speed. The cross-correlation function between pairs of time-
normalized waveforms was computed to quantify the time shift of kinematic parameters with treadmill speed by means of
the following formula:
∫∫∫ ∆+
=∆dttydttx
dtttytxRxy
)()(
)()()(
22
where x and y are the two waveforms (after subtraction of their mean values), and ∆ is the time lag between the two
signals. The highest positive correlation and its corresponding time lag were detected and expressed as a percentage of gait
cycle duration. Using this method, we calculated time delays of mean elevation angles (HL and FL) at 0.89, 1.34, and 1.79
m/s with respect to those at 0.45 m/s (Ivanenko et al. 2004b).
EMG. Raw EMG signals were band-pass filtered (30 Hz to 1 kHz), rectified, time-interpolated over a time base with 1000
points for individual gait cycles, and averaged. Onsets and ends of EMG burst activity of each muscle recorded during each
gait cycle were established at the points at which muscle activity exceeded and fell below, respectively, the mean activity
plus 1.5 SD recorded during a period (200 ms) when this muscle was least active (Courtine and Schieppati 2003). A mobile
average (40 ms width) was first applied on the signal to reduce the effects of signal oscillation (Courtine and Schieppati
2003). The time between the onset and end of an individual burst was considered the burst duration. The onset and ending
points of the EMG bursts were used to determine the relative timing of EMG activity recorded from different muscles. For
the HL, cycle period was calculated as the time between the onset points of successive bursts of EMG activity in the Sol
muscle. Activity of FL muscles was synchronized with kinematic data: cycle period corresponded to the time interval
during two successive paw contacts. Mean EMG amplitude was calculated as the integral of the muscle envelope divided
by the burst duration (Roy et al. 1991b). The EMG amplitude of each burst was normalized to the mean burst amplitude of
the same muscle when walking at 0.45 m/s, i.e., the slowest treadmill speed.
Statistical analysis. For each animal, we calculated the mean values and standard deviations (SD) of the different
parameters over all trials for each experimental condition. Repeated-measures ANOVA’s were used to test the effect of the
different conditions on the experimental parameters. The factors examined were the limb (HL, FL) and the treadmill speed
(0.45, 0.89, 1.34, and 1.79 m/s). Post-hoc differences were assessed by the Newman-Keuls test. Regression linear analyses
were performed to determine the relationships between variables and reported as correlation coefficients.
Courtine et al.
p.11
Results
The spatial and temporal characteristics of gait parameters, locomotion kinematics and muscle activation patterns were
analyzed in six adult Rhesus walking quadrupedally over a wide range of treadmill speeds. Gait, kinematic, and EMG
descriptors routinely used to describe cat and human walking were derived from recordings of Rhesus locomotion.
Spatial and temporal characteristics of the Rhesus gait pattern and their
relationship with treadmill speed
The animals used a diagonal footfall sequence when walking over the imposed range of speeds, i.e., the footfall of a FL
usually followed that of the contralateral HL. As a consequence, the stance phase of a diagonally-opposed pair of limbs
occurred approximately at the same time, and in general, either four or two limbs were simultaneously supporting body
weight at the slower and higher speeds, respectively (Fig. 1). An interesting consequence of this diagonal footfall sequence
was that the foot contacted the treadmill belt in close proximity to the ipsilateral hand. The distance between HL and FL
endpoints was approximately equal to zero at HL footfall for all speeds (Fig. 2D). Conversely, the FL contacted the
treadmill belt near the end of the HL stance phase, i.e., when the HL segments reached a backward position, and the
interlimb distance was substantial at this time (limb effect, p < 0.001).
Despite the fact that the Rhesus generally used a diagonal footfall sequence, the precise timing of FL footfall with respect
to HL footfall varied appreciably with speed (Fig. 2A). The interlimb timing was quantified by relating the time of footfall
of either limb relative to two successive footfalls of the right HL. With increased treadmill speed, the footfall of the
contralateral FL (black circles) and the ipsilateral FL (grey circles) was delayed with respect to the ipsilateral HL cycle
duration. For example, the footfall of the left FL could precede, coincide with, or occur consistently after the footfall of the
right HL footfall with increasing treadmill speed (Fig. 1A). However, the two FLs maintained a near-perfect out of phase
coupling over the range of speeds studied. Changes in the timing of footfall with increasing speeds were virtually identical
for the two FLs (left and right FL best-fitting lines are parallel in Fig. 2A). Likewise, contralateral HL footfalls (open
circles) occurred at half of the ipsilateral HL cycle duration (49 ± 3%) regardless of the actual body velocity. One Rhesus
(#5) used a lateral footfall sequence at the three faster speeds in which a FL footfall followed the ipsilateral HL footfall. In
Fig. 2A the triangle symbols represent the timing of ipsilateral FL footfall for this Rhesus.
Figure 2B depicts the relationship between the cycle duration and the mean body velocity for the right HL and FL. A
monotonic decrease in cycle duration accompanied an increase in mean body velocity (speed effect, p < 0.001), which was
nearly identical for the HLs and FLs (limb effect, p > 0.10). Accordingly, the HLs and FLs displayed similar stride lengths
under comparable treadmill speeds (Fig. 2C) (limb effect, p > 0.10).
In spite of the observation that HL and FL cycle duration and stride length progressed similarly, systematic differences
were detected between the duration of their stance phases (speed * limb interaction effect, p < 0.05). Except at the slowest
speed (p= 0.30), stance duration was significantly longer for the HL than the FL (Fig. 2E, all post hoc comparisons, p <
0.05). This difference between HL and FL duty factors is highlighted in Fig. 2F where the stance and swing durations are
plotted vs. the total duration of the cycle for Rhesus #1.
Kinematic and EMG patterns during rhesus monkey locomotion
p.12
Spatial and temporal characteristics of HL and FL kinematics and their
relationship with speed of locomotion Figure 3 displays the average (±SD) joint angle waveforms at each joint of the HL and FL recorded at each treadmill speed
in a representative Rhesus (#4). Each joint angle of the HL and FL changed cyclically within a single step and the time
course for the changes was consistent for the six Rhesus. We observed a gradual decrease in stance duration (Fig. 2E and F)
and a progressive lengthening of the stride (Fig. 2C) with increases in treadmill speed. These progressive changes in spatial
and temporal gait parameters were reflected in the graded adjustment of the timing and amplitude of the HL and FL joint
angles (Fig. 3). Changes in timing with increases in treadmill speed included an earlier maximal joint extension of all HL
and FL angles that paralleled the decrease in stance phase duration (compare Fig. 2E and F with Fig. 3). The range of joint
excursions increased as treadmill speed increased for both the distal and proximal segments of the HL and FL (Fig. 3). The
increase in amplitude of HL joint angles with increasing treadmill speed was significant at all joints (speed effect, p <
0.01). This modification was progressive and significant correlations were detected between the mean body velocity and
the joint angle amplitude at the hip (r = 0.82 ± 0.05), knee (r = 0.88 ± 0.05), ankle (r = 0.90 ± 0.06), and MT (r = 0.88 ±
0.07). In the FL, an increase in the amplitude of the joint angles with increased treadmill speed was mainly located at the
shoulder (speed effect, p < 0.01), and at the elbow, though to a lesser extent (speed effect p < 0.05). A significant
correlation between treadmill speed and amplitude of joint angle changes was observed only at the shoulder joint (r = 0.72
± 0.10).
Superimposed trajectories of the HL and FL endpoints during the swing phase for six consecutive step cycles at each
treadmill speed are shown in Fig. 4A. Treadmill speed-related increases in the length (speed effect, p < 0.01) and height
(speed effect, p < 0.01) of the path of the endpoint trajectories are seen for both the HL and FL. Ivanenko et al. (2002)
showed that in human locomotion, the foot trajectory obeys the so-called 2/3 power relationship between the instantaneous
curvature (C) and angular velocity (Ω), i.e., the exponent β of the Ω—C relationship is very close to two-thirds (see
Materials and Methods). We assessed whether the same relationship characterizes HL and FL endpoint trajectories in the
Rhesus. As in humans, the correlation value of the linear regression were very high for all Rhesus, regardless of the limb
and speed (r = 0.98 ± 0.01). The typical Ω—C relationship obtained from all steps performed by a representative Rhesus
(#4) at 0.89 m/s is depicted in Fig. 4B for the HL (left panel) and FL (right panel). The exponent β was very close to 2/3 for
both the HL and FL at all treadmill speeds (Fig. 4C). Nevertheless, β was significantly closer to 2/3 when regressions were
computed for the FL compared to the HL endpoint trajectories (limb effect, p < 0.05). This difference was small (mean β
HL minus mean β FL was 0.03 ± 0.02), but consistent across animals.
Inter-segmental coordination patterns between the HL and FL
We investigated the inter-segmental coordination among the HL and FL segments during quadrupedal walking in the
Rhesus as previously described for humans (Courtine and Schieppati 2004). Figure 5A displays the average (for 6 Rhesus)
waveforms of the HL and FL elevation angles and at each treadmill speed studied. In addition, the 3-D gait loops obtained
Courtine et al.
p.13
by plotting these elevation angles vs. each other are shown for Rhesus #4 in Figure 6A. Spatio-temporal modifications of
HL and FL segment oscillations with increasing speed were assessed by computing the cross-correlation function between
the mean angular waveforms at 0.89, 1.34, and 1.79 m/s relative to those at 0.45 m/s in each animal (see Methods and
Materials). Phase lag and correlation coefficients quantify the changes in timing and shape of the elevation angles with
increasing speed, respectively (Fig. 5B). The correlations (insets of Fig. 5B) were very high across speeds for all HL and
FL elevation angles, although r values tended (p < 0.1) to decrease for the distal segments of both limbs. Indeed, the
amplitude of distal segment backward oscillation generally increased at the highest treadmill speed. An increase in speed
was generally associated with a progressive lead in the time of maximal backward oscillation, i.e., end of stance. This shift
corresponded to the decrease in stance duration (filled squares in Fig. 5B), and was more pronounced for distal compared to
proximal segments.
Vertical lines joining the data points to the grid in the plots in Figure 6A indicate the distance between the 3-D gait loop
and the best-fitting plane obtained by linear regression: systematic deviations from a perfect plane were detected for both
the HL and FL segments. The mean (SD) value (for 6 Rhesus) of the index of planarity (see Materials and Methods) is
reported above each 3-D plot, where a value of 100 corresponds to a perfect plane. The size of the 3-D gait loop described
by HL elevation angles increased considerably at the higher speeds, as a consequence of the combined increase in
amplitude and phase lag among segment oscillations (Fig. 5A and B), whereas the spatial orientation and general shape
were unaffected.
The coupling between the oscillations of HL and FL segments was examined further through principal component (PC)
analysis (Courtine and Schieppati 2004). PC analysis was performed on time-varying elevation angles of the HL and FL
during complete gait cycles, either independently for each limb (n = 4), or simultaneously (n = 8). Mean values of the
variance accounted for by each PC are shown in Figure 6 B, C, and D: the higher the variance explained, the stronger is the
linear co-variation between spatio-temporal changes of segment angles (Bianchi et al. 1998). Four main results emerged
from this analysis: (1) the variance accounted for by PC1 was generally high when PC analysis was applied independently
on either the HL or FL datasets (Fig. 6B). PC1 explained more than 80% and PC1 plus PC2 about 98% of the variance in the
data. (2) The variance accounted for by PC1 was higher (limb effect, p < 0.005) for HL (86 ± 2%) than FL segments (77 ±
3%). (3) When PC analysis was performed simultaneously on the HL and FL segments, PC1 plus PC2 generally accounted
for nearly all of the variance in the data, i.e., mean of 91% (range, 85 to 94%) (Fig. 6C). (4) In this latter analysis (n=8),
the variance accounted for by PC1 plus PC2 increased linearly with increasing body velocity (speed effect, p < 0.0001) (Fig.
6D).
Amplitude and timing characteristics of HL and FL muscle activity and their
relationship with treadmill speed
The averaged EMG activity recorded from selected HL (Rhesus #1, #2, and #3) and FL (Rhesus #5 and #6) muscles (Table
1) is plotted vs. the normalized gait cycle duration at each treadmill speed (Fig. 7). Figure 8 shows how the characteristics
of the HL and FL muscle bursts, i.e. integrated (time-normalized integral) EMG activity and timing, are modulated with
changes in cycle duration.
Kinematic and EMG patterns during rhesus monkey locomotion
p.14
HL muscles. HL muscle activities were very consistent across the three Rhesus (#1, #2 and #3). The pattern of HL muscle
activity typically showed an alternate activation of extensor and flexor muscles during the stance and swing phase of gait,
respectively. The onset of extensor activity began slightly before the time of paw contact as reported previously in cats
(Rasmussen et al. 1978). The VL and MG generally were activated prior to Sol activation (Fig. 7). The duration of extensor
muscle bursts gradually decreased with a decrease in cycle and, therefore, stance duration (see Fig. 2E). The level of EMG
activity was unchanged in the Sol and slightly increased (p > 0.05) in the VL with increasing velocities. In contrast, the
integrated activity of the MG bursts increased as much as 6-fold (Rhesus #2) when locomoting at 1.79 vs. 0.45 m/s (p <
0.001).
FDL and FHL, which are close anatomical synergists, were differentially recruited during walking. The FHL muscle
produces toe flexion and plantarflexion torque at the ankle, and was activated concomitantly with the Sol muscle
throughout the stance phase of gait. The duration of the FHL EMG burst (Fig. 8) decreased with increasing velocities
(decreased stance phase). The integrated EMG activity of the FHL muscle sharply increased as a function of speed (p <
0.001). FDL burst onset did not coincide with the onset of the extensor muscles, i.e., the FDL was activated near early-mid
stance and ceased its activity around the time of toe-off. This pattern of activity was observed in both Rhesus #2 and #3.
Speed-related modulation of FDL EMG activity was substantial (Figs. 7 and 8); FDL burst integral increased (p < 0.001) as
much as 20-fold when walking at 1.79 vs. 0.45 m/s (Rhesus #3).
The TA was activated during the swing phase (Fig. 7). The duration of the TA burst was roughly invariant over the range of
speeds studied, whereas the EMG integral gradually increased with speed (p < 0.001) (Figs. 7 and 8). The EDL burst was
initiated with the TA burst at the onset of swing, but its activity persisted after TA burst extinction (Fig. 7). Therefore, the
EDL was co-activated with ankle extensors during ~10-20% of the cycle duration, depending on the Rhesus and the speed
of locomotion. EDL burst duration decreased modestly (p < 0.01) with a decrease in cycle duration (Fig. 8). The EMG
burst integral gradually increased (p < 0.001) with a decrease in cycle duration (Fig. 8).
FL muscles. FL muscle activity is shown for both Rhesus #5 and #6 to illustrate individual differences as well as
similarities. Both Rhesus exhibited the same reciprocal activation pattern between the Tri and Bic in the FL as between the
Sol/MG and TA in the HL. Activation of Tri began prior to (Rhesus #5) or at (Rhesus #6) paw contact, and persisted
throughout the stance phase. As observed in the ankle extensors, the EMG burst duration of the Tri gradually decreased
with a decrease in cycle duration. The integral of the Tri EMG burst gradually increased with an increase in treadmill speed
(p < 0.01) (Fig. 8). However, the speed-related increase in Tri recruitment was not as large as that observed in the MG
muscle (p < 0.05). Bic activity was initiated around the time of paw-off and terminated before (Rhesus #5) or at (Rhesus
#6) paw contact. As observed in the TA, the Bic burst duration was constant over the range of speeds studied (Fig. 8). In
turn, both Rhesus showed a sizeable increase (p < 0.001) in Bic activity with speed (Figs. 7 and 8).
We recorded the EMG activity of four muscles acting at the wrist, digits, and thumb. FDP and FDS, which both produce a
flexion of the wrist and digits, were activated throughout stance (Fig. 7). However, both the left (not shown) and right FDP
muscles exhibited an additional burst of activity starting immediately after paw-off. In Rhesus #5, we found a high level of
EMG activity in the FDS muscle during the stance phase, whereas the muscle was recruited modestly (low speed) or was
quiescent (high speed) during swing. In contrast, the FDS EMG activity recorded in Rhesus #6 was dramatically larger
during the swing phase than during the stance phase. The FPB is a thumb flexor. FPB activity was initiated just after paw
Courtine et al.
p.15
contact, and ceased before paw-off, i.e., when the wrist started flexing (Fig. 3), thereby raising the thumb off the treadmill
belt.
The EDC extends the wrist and digits. In Rhesus #5, EDC showed a modest, short burst occurring around paw contact and
was followed by a period of tonic activity that ended with swing onset (Fig. 7). This latter activity was particularly high at
the fastest treadmill speed. The recruitment of EDC began near the end of swing (90 ± 1.4%) in Rhesus #6, and ended
when the hand and digits were laid flat on the treadmill belt (20% and 12% of the cycle period when walking at 1.79 vs.
0.45 m/s; Fig. 3).
Characteristics of digit muscle activity were gradually modulated with increasing speed, both in timing and amplitude. All
digit muscles recorded in Rhesus #5 were active during the stance phase of gait (Fig. 7). Accordingly, the burst duration
decreased with increased treadmill speed (Fig. 8). In contrast, FDP (swing-related burst) and EDC burst durations were
invariant over the range of speeds studied in the Rhesus #6. The burst integrals were significantly (p < 0.05) correlated with
the cycle duration in all digit muscles of both animals. The level of EMG activity significantly increased as a function of
speed in all of the digit muscles (p < 0.001), particularly in the FDS and FDP during stance and swing, respectively (Fig.
8).
Coordination of muscle activity
We further investigated the temporal tuning of FL muscle activity by scrutinizing the relationships between the temporal
features of their bursts and the timing of FL and HL segment oscillations (Fig. 9). Temporal burst features related to the
end of stance were selected because modifications in stance duration reflected the speed-related changes in the temporal
structure of the gait pattern (Fig. 5B). The timing of HL and FL segment oscillations was computed through FFT analysis
(see Methods and Materials), and was expressed as a percent of the normalized gait cycle duration. An increase in the value
of the timing indicates that the segment oscillation peaked earlier with respect to the normalized gait cycle duration. This
typically corresponds to a decrease in stance duration, i.e., an increase in the speed of movement (Fig. 5B). Fig. 9 shows the
relationship between the temporal features related to the end of stance for all bursts from the right FL muscles and the
timing of FL (top) and HL (bottom) segment oscillations. These relationships were computed from gait cycles recorded
from Rhesus #5; however, similar relationships were observed for Rhesus #6. The timing of FL muscle activity was
significantly (p < 0.01) correlated with the timing of all FL and HL segment oscillations. The relationships associated with
each HL and FL segment are depicted with different symbols to emphasize the lag between oscillations of the different
segments. The results indicate that any modification in FL muscle timing was coordinated on a cycle-to-cycle basis with
similar changes in the timing of all HL and FL segments.
Finally, we explored how the CNS coordinates the tuning of the motor patterns to increase locomotor velocity. Figure 10
shows the modulation of the burst integral of each muscle in the HL relative to the burst integral of the Sol and TA and of
each muscle in the FL relative to the burst integral of the Bic and Tri. The Sol and TA and Tri and Bic muscles were
selected as a reference for the HL and FL, respectively, because similar reciprocal activation and speed-related modulation
of burst characteristics were detected for each extensor-flexor muscle pair. Burst integral of each muscle was normalized to
Kinematic and EMG patterns during rhesus monkey locomotion
p.16
its mean value computed during walking at the slowest treadmill speed (0.45 m/s). These values were not time-normalized
since each burst integral was computed during the same gait cycle. Significant correlations (p < 0.05) were detected
between the burst integral of the right Sol and TA and those of the right and left thigh and shank muscles. The relationships
tended to be less robust when considering the distal HL muscles. In contrast, the correlations between the activities of
individual FL muscles were very low for Rhesus #5 and inconsistent (i.e. low or high) for Rhesus #6.
Courtine et al.
p.17
Discussion We examined the spatial and temporal characteristics of gait parameters, locomotion kinematics, and associated patterns of
HL and FL muscle activity during quadrupedal treadmill locomotion over a wide range of speeds in the Rhesus monkey.
Our results point out a number of similarities, but also some differences, in the kinematic and EMG determinants sustaining
Rhesus locomotion compared to sub-primate mammals and to humans. The adaptation process of non-human primates to
an arboreal environment thus involved a number of changes in gait control including alterations to muscle recruitment and
the associated patterns of interlimb and intralimb coordination. Substantial changes either in the properties of the stepping-
related spinal neuronal circuitry or in the level of supraspinal control, or both necessarily accompanied these modifications
of the locomotor program (Capaday 2002; Duysens and Van de Crommert 1998; Nielsen 2003; Vilensky and O'Connor
1997).
Interlimb coordination
One of the most distinctive aspects of primate quadrupedal walking is the frequent use of diagonal sequence footfalls (Fig.
1) in combination with diagonal-couplet interlimb timing (Fig. 2A). Most other quadrupedal mammals use a lateral
sequence footfall pattern (Hildebrand 1967). Evolutionary correlates of primate-specific interlimb coordination have been
addressed in anthropological studies. Biomechanical analyses suggest that the diagonal sequence gait increases body
stability in monkeys due to the posterior location of their center-of-mass (Kimura 1985; Schmitt 2003; Vilensky 1989).
Furthermore, Cartmill et al. (2002) argued that monkeys use a diagonal sequence footfall pattern to place the grasping
hindfoot in a protracted position and on an already-tested support at touchdown: the distance between the ipsilateral foot
and hand is roughly null at HL contact (Fig. 2D). Consequently, the supporting hindfoot is advantageously located
underneath the animal’s center of mass when the contralateral hand strikes the ground on an untested, arboreal support.
Interestingly, humans exhibit a similar diagonal-coupled leg-to-arm coordination when creeping, swimming, or walking
erect on two legs (Wannier et al. 2001).
Investigations in cats (Miller et al. 1975) and rats (Butt et al. 2002; Butt and Kiehn 2003) disclose some neural elements
responsible for interlimb coordination during locomotion. These studies show that short-range commissural interneurons
ensure left-right coordination, whereas long-range propriospinal interneurons connect lumbar and cervical enlargements
and assist coupling between the HL and FL (Grillner 1981). In the current study, we generally observed a high correlation
in the modulation of the EMG activity in the left and right limbs, although this relationship was stronger between the
lumbar compared to the cervical motor pools (Fig. 10). Moreover, the temporal features of the proximal and distal FL
muscle recruitment patterns were highly coordinated with the timing of the oscillations of all the segments of both the FL
and the HL (Fig. 9). Finally, principal component analysis applied simultaneously on elevation angles of all HL and FL
segments revealed a high degree of coupling in the generation of limb oscillations during Rhesus locomotion (Fig. 6C and
D). Thus similar neural mechanisms could operate in the spinal cord of sub-primates and primates to ensure inter-limb
coordination during quadrupedal walking. Indeed, the existence of long projecting propriospinal neurons coupling lumbar
and cervical enlargements has been demonstrated not only in non-human primates (Molenaar and Kuypers 1978) but also
humans (Nathan et al. 1996). Accordingly, recent studies also provided evidences for a strong coupling in the control of the
Kinematic and EMG patterns during rhesus monkey locomotion
p.18
two leg movements (Courtine and Schieppati 2004; Ting et al. 2000) and, though to a less extent, between the neuronal
processes generating leg and arm oscillations during human locomotion (Dietz et al. 2001; Zehr and Duysens 2004).
Nevertheless, similar mechanisms for interlimb coordination in mammals do not account for two important specific
features of primate locomotion: 1) the emergence of a diagonal coupling pattern, and 2) the high versatility in the control of
limb movements. Indeed, walking in the primate species is characterized by frequent changes in interlimb coupling, e.g., to
grasp a supporting branch or an object while locomoting (see Fig. 5 in D'Aout et al. 2004). Interestingly, juvenile monkeys
preferentially use lateral gait when walking in an arboreal environment, and then shift to a diagonal gait after maturation of
the descending tracts (Dunbar and Badam 1998). Furthermore, interruption of the descending spinal pathways results in
severe disruption of interlimb coupling during locomotion (Vilensky et al. 1992). It seems that such flexibility in interlimb
coordination and associated postural regulation (Mori et al. 2004) requires a significant contribution from supraspinal
control mechanisms (Drew et al. 2004).
The theoretical analysis of coupled oscillator-based interlimb coordination during quadrupedal locomotion stipulates that
versatility in limb movements can be achieved easily by adjusting the phase relationships between the neural oscillators
controlling HL and FL movements (Schoner et al. 1990). It is plausible, therefore, that supraspinal commands modulate the
coupling between propriospinal neuronal networks to regulate interlimb coordination during locomotion. For example,
Jankowska et al. (2003) showed that lumbar commissural interneurons are monosynaptically activated from the ipsilateral
reticular formation in cats. Indeed, cerebellar contributions to interlimb coordination via reticulospinal neurons occur in
both cats (Armstrong 1988) and humans (Morton and Bastian 2004).
Intralimb coordination
Lacquaniti and co-workers (Lacquaniti et al. 2002) demonstrated the oscillation of lower limb segments with respect to the
direction of gravity do not evolve independently of each other during walking in humans. On the contrary, a kinematic law
of planar co-variation among lower limb segment oscillations characterize intralimb coordination pattern during a variety
of locomotor tasks (Bianchi et al. 1998; Borghese et al. 1996; Courtine and Schieppati 2004). In comparison with human
gait, the elevation angles of thigh, shank, and foot segments did not evolve close to a plane during non-human primate
locomotion, nor did the elevation angles of the FL segments (Fig. 5A). Furthermore, the 3-D gait loop size described by HL
elevation angles increased greatly with speed, as a consequence of larger segment oscillations (Fig 5A), whereas its spatial
orientation changed minimally in the Rhesus. In contrast, the plane of angular co-variation systematically rotates and the 3-
D gait loop shape varies little for the production of higher velocities in humans (Bianchi et al. 1998). This corresponds to a
parametric tuning in the phase-relationship of inter-segmental coordination (Courtine and Schieppati 2004) that is used by
the nervous system to optimize pendulum-like movements, and thereby limit the overall energy expenditure in humans
(Bianchi et al. 1998). The current results, therefore, suggest that the strategy by which the CNS achieves inter-segmental
coordination in nonhuman primates and adapts its spatio-temporal structure to increase speed differ somewhat from the
kinematic principles that operate in human gait control.
Monkeys do not achieve optimum inverted pendulum-type gait (D'Aout et al. 2004), nor do human toddlers (Ivanenko et al.
2004a) during their first steps. Development of pendulum mechanisms in infants correlates with the emergence of planar
Courtine et al.
p.19
co-variation among lower limb oscillations (Ivanenko et al. 2004a). We, thus, conclude that control of bipedal, erect
walking movements likely required further constraints on the control of inter-segmental coupling to stabilize the vertical
trunk (Hirasaki et al. 2004) and to optimize energy-saving pendulum movements (Ivanenko et al. 2004a). Such differences
in intralimb coordination strategy must be taken into consideration when studying bipedal walking in normally quadrupedal
animals (D'Aout et al. 2002; Mori et al. 2001) since we have to assume that the human is not a monkey walking on two
legs, and that refinement in the organization of motor control centers has undoubtedly taken place during the evolution
toward habitual bipedalism (Nielsen 2003).
Speed-related changes in gait parameters and muscle activity
Modulation of spatial and temporal characteristics of gait and muscle activity with respect to speed has been thoroughly
studied in some mammals, especially in cats (Grillner 1981) and humans (Nilsson et al. 1985). These studies highlight the
common organizational principles by which the stepping-related motor program smoothly adapts to an increase in
locomotor velocity. Our results show that these same neural strategies operate during Rhesus locomotion. As in other
mammals, an increase in body speed was associated with a monotonic decrease and increase in cycle period duration and
stride length, respectively, which was virtually identical for the HL and FL (Fig. 2; Mori et al. 1996; Vilensky 1983).
Moreover, an increase in treadmill speed typically involved a graded decrease in the EMG burst duration of extensor
muscles that paralleled a decrease in stance duration (Figs. 2F and 8). In contrast, the durations of flexor muscle bursts and
of the swing phase remained almost constant over the range of speeds studied (Fig. 2F and 8). This speed-related
modulation was observed in both the HL and FL muscles, and a similar reciprocal activation pattern was detected between
the Sol/MG and TA muscles in the HL, and the Tri and Bic muscles in the FL. These observations are consistent with the
idea that the ankle and elbow joints share a similar function during locomotion in quadrupeds (English 1978; Rossignol
1996). Another similarity between the Rhesus and other mammals was the differential recruitment of slow vs. fast muscles
with respect to speed of locomotion. For example, the MG muscle, comprised of a high proportion of fast, fatigable motor
units (Roy et al. 1991a), was more heavily recruited at the faster than the slower speeds of locomotion, whereas the
predominantly slow Sol muscle was already heavily recruited at the slower treadmill speeds (Figs. 7 and 8). Similar
observations have been reported in the Rhesus (Recktenwald et al. 1999) as well as in rats (Roy et al. 1991b; de Leon et al.
1994) cats (English 1984; Pierotti et al. 1989), and humans (Nilsson et al. 1985). This coordinated, speed-related
modulation of HL and FL muscle activity may arise from similar mechanisms in Rhesus and in cats, i.e., velocity-
dependent tuning of spinal circuits via the brainstem tonic commands and the phasic afferent input signaling hip extension
(Grillner and Rossignol 1978) and loading of the leg (Duysens and Pearson 1980). For example, Shik et al. (1966) showed
that electrical stimulation of brainstem nuclei in decerebrate cats evokes quadrupedal locomotor activity and that the
stepping rate depends on the intensity of the stimulation. Eidelberg et al. (1981) similarly detected a ‘positive site’ for
stimulation in the posterior subthalamic region that elicited locomotor movements in monkeys. Stronger stimulation
intensities increased the stepping rate, and even provoked a shift from a walking to a galloping gait.
Kinematic and EMG patterns during rhesus monkey locomotion
p.20
Control of the support and propulsion phases of locomotion
In the current study, we observed a variety of differences in the operative principles regulating speed-related adjustments of
HL and FL movements during Rhesus locomotion. For example, an increase in treadmill speed typically involved a larger
decrease in stance duration, and thus extensor burst duration, in the FL than the HL (Fig. 2E). Moreover, the increase in
extensor muscle recruitment was greater in the HL than the FL. This substantial, speed-related tuning of HL EMG activity
was most noticeable in the predominantly fast leg muscles (see above) as well as in the muscles functioning at the distal
joints such as the FHL and FDL (Fig. 8). In the cat, the FDL exhibits a phasic burst at swing onset to produce the very early
plantarflexion of the digits needed to clear the paw from the ground (Fleshman et al. 1984; O'Donovan et al. 1982). In the
Rhesus, the FDL muscle was recruited during more than one-half of the stance phase, and its activity increased as much as
20-fold at the faster compared to the slower treadmill speeds. This modulation of ankle and digit muscle activity was
reflected in the monotonic increase in the excursion of the distal joints (Fig. 3), as well as in the gradual lag of the distal
with respect to the proximal segment oscillations (Fig. 5B) with increased treadmill speed. In contrast, speed-related
changes in the amplitude of the FL oscillations were limited, and mainly located at the shoulder joint (Fig. 3; Larson 1998).
All of these differences between the HL and FL are aspects of the same adaptive processes, and provide evidence that the
HL muscles generate a majority of the force necessary to support and propel the body mass during quadrupedal locomotion
in the Rhesus. Kimura (1985) likewise concluded from kinetic recordings that the HLs carry most of body weight during
walking in non-human primates, whereas the FLs primarily steer the animal. Such specialization of the HL for propulsion
and support may result from evolutionary pressure to reduce compressive forces on the FL (Larney and Larson 2004;
Schmitt 1999). It is also conceivable that this active use of the HL digits along with the generation of substantial forces to
keep up with faster treadmill speeds may require more contribution from supraspinal control systems (Mori et al. 1996).
Control of limb oscillation
Differences in the recruitment of HL vs. FL muscles in Rhesus compared to other quadrupedal mammals, were detected not
only during the support phase, but also during the period of limb oscillation. For instance, the speed-related increase in
EMG amplitude was greater in the Bic than its counterpart, the TA (Fig. 8). The compliant gait used by monkeys to reduce
peak locomotor stresses on the FL is reflected in substantial elbow yield during stance (Larson et al. 2000; Schmitt 1999).
Therefore, the increased level of recruitment of the Bic in the beginning of swing may be linked to clearing the paw from
the ground while the HL muscles are generating forces to propel the body forward (Fig. 3). Similarly, the recruitment of the
FDP early in swing (Fig. 7) contributes to the rapid and pronounced MCP flexion that follows paw lift (Fig. 3). The
emergence of FDP activation during the swing phase of primate locomotion thus correlates with the development of FL
digits (Okada 1985). Similar anatomical modifications of the foot digits may have necessitated adaptations in distal HL
muscle activity patterns. Indeed, we found that the EDL in the Rhesus is not strictly activated during the swing phase of
gait as has been observed in other quadrupedal mammals (Trank et al. 1996). Instead, the EDL has a prolonged burst of
activity and, similar to the EDC in the FL, is co-activated with extensor muscles during early stance. This activation pattern
of physiological flexors (Rho et al. 1999) of the HL (EDL) and FL (EDC) probably prevents the slender digits from
dragging along the walking surface until there is complete loading of the foot and hand. Interestingly, activation of
dorsiflexors at the beginning of stance is required during human walking to achieve heel strike first (Nilsson et al. 1985).
Courtine et al.
p.21
The emergence of prolonged activation in distal flexor muscles following foot contact during monkey locomotion,
therefore, may have created a favorable context for the development of bipedal gait mechanisms.
Comparison with cat locomotion Investigations performed in cats have provided most of our knowledge on the neuronal control of locomotion in mammals
(reviews in Drew et al. 2004; Edgerton et al. 2004; Orlovskii et al. 1999; Pearson 2004; Rossignol 1996). Therefore, it is
important to emphasize the differences in the kinematic and EMG determinants underlying cat vs. Rhesus quadrupedal
stepping. Although cats and Rhesus share a number of common gait features, there are three important differences: 1) the
inter-limb coupling; 2) the different modulation of HL vs. FL gait features with speed; and 3) the EMG patterns of HL and
FL distal muscles. Indeed, cats, as most mammals, use a lateral footfall sequence during walking (Halbertsma et al. 1976).
Moreover, contrary to Rhesus (Figs. 2 and 8), the durations of the flexor and extensor phases change to the same extent in
the HL and FL with increasing speed in the cat (Halbertsma et al. 1976). Accordingly, speed-related increases in the
amplitude of HL joint angle changes and the level of HL muscle activity are larger in Rhesus (present results) than in cats
(Halbertsma et al. 1976; Rossignol, 1996; Jiang and Drew 1996), particularly in the distal extremities. In the same vein,
vertical reaction forces are larger on the FL than the HL during cat locomotion (Manter 1938; Lavoie et al. 1995), whereas
the inverse ratio characterizes non-human primate stepping (Kimura, 1985). Similar to that observed in monkeys, however,
the cat HLs provide propelling forces for a longer period than the FLs, and the FLs are used mainly for braking (Lavoie et
al. 1995). This substantial contribution of HL muscles to body propulsion in Rhesus at higher speeds is reflected in the
dramatic modulation of distal muscle activity, such as the FDL and FHL muscles, which is not observed in cats (Fleshman
et al. 1984; O'Donovan et al. 1982). Finally, flexor (FDL) and extensor (EDL) muscles of the HL show different
recruitment patterns in Rhesus vs. cats (Fleshman et al. 1984; O'Donovan et al. 1982; Trank et al. 1996). Likewise, whereas
the EDC exhibits a phasic burst at the swing to stance transition during locomotion in the cat (Drew 1993) as observed in
Rhesus, significant differences were noted in the recruitment of distal flexor muscles of the FL. The FDP muscle is
activated only during the support phase during cat locomotion, whereas a second burst emerges during the swing phase in
the Rhesus. Accordingly, the amplitude of flexion of the MCP joint at swing onset is not as large in cats (Drew and Kably,
cited in Rossignol 1996) as in Rhesus. These changes in the recruitment pattern of distal muscles of the HL and FL during
Rhesus locomotion are interesting since they are associated with an increased representation of the same muscles in the
motor cortex (Park et al. 2004). We discuss in the next section that such adaptive modifications may correlate with an
increase in the level of supraspinal control of gait in the Rhesus vs. the cat.
Adaptation in neurobiological control mechanisms
We showed that inter-muscular coordination tended to be less consistent when considering the activity of the distal muscles
of the HL and FL (Fig. 10), i.e., the muscles showing primate-specific activation. This would be expected if the cortical
input superimposes its neural drive upon the ongoing, spinally-generated motor activity in order to adjust the amplitude of
the motor pool activity of the distal musculature in a step-to-step basis, as originally proposed by Kuypers (1978). We
found that the generation of distal HL and FL trajectories generally obeys the relationship between instantaneous curvature
and angular velocity (known as the two-thirds power law) similar to that observed in humans (Ivanenko et al. 2002;
Lacquaniti et al. 1983) (Fig. 4). This law is derived from principles of optimum endpoint control (Harris and Wolpert
1998), and is reflected in the neuronal activity of the primate motor cortex during manual tracking (Schwartz and Moran
Kinematic and EMG patterns during rhesus monkey locomotion
p.22
1999). The existence of such rules is consistent with the notion that the motor cortex may actively control the activity of
distal muscles during primate gait. Accordingly, Georgopoulos and Grillner (1989) postulated that the same neural
mechanisms are involved in voluntary reaching and accurate positioning of the limb during locomotion in primates.
Increased cortical control of stepping-related limb movements may have been critical for successful travel in a
discontinuous, complex small-branch arboreal environment (Larson et al. 2000). Furthermore, this increased “voluntary”
control of distal limb movements may have created favorable conditions for the development of fine manipulatory ability
independent of locomotion (Georgopoulos and Grillner 1989), and the evolution toward bipedalism. Nonetheless, only
direct recordings of stepping-related pyramidal neuron activity could provide the opportunity to address this hypothesis
(Drew et al. 2002).
Conclusions
The organization of gait control mechanisms in Rhesus show a number of similarities with other terrestrial animals,
suggesting that common principles operate during stepping among a wide range of mammals. However, the non-human
primate exhibits differences in interlimb and intralimb coupling, hindlimb and forelimb EMG patterns as well as propulsive
and support systems compared to sub-primate animals. Some of the major differences are: 1) the frequent use of diagonal-
couplet interlimb timing; 2) the more important propulsive role of the HL compared to the FL with increase in locomotor
velocity; 3) the novel recruitment pattern of some distal HL and FL muscles; and 4) the limb-specific neural control
strategies in the inter-segmental coordination patterns and limb endpoint trajectories. The current analyses of kinematic and
EMG determinants of Rhesus locomotion provide a baseline for future comparisons of neuromotor properties following
spontaneously occurring dysfunctions and selected experimental interventions.
Courtine et al.
p.23
References
Armstrong DM. The supraspinal control of mammalian locomotion. J Physiol 405: 1-37, 1988.
Bianchi L, Angelini D, Orani GP, and Lacquaniti F. Kinematic coordination in human gait: relation to mechanical
energy cost. J Neurophysiol 79: 2155-2170, 1998.
Borghese NA, Bianchi L, and Lacquaniti F. Kinematic determinants of human locomotion. J Physiol 494: 863-879,
1996.
Butt SJ, Harris-Warrick RM, and Kiehn O. Firing properties of identified interneuron populations in the mammalian
hindlimb central pattern generator. J Neurosci 22: 9961-9971, 2002.
Butt SJ and Kiehn O. Functional identification of interneurons responsible for left-right coordination of hindlimbs in
mammals. Neuron 38: 953-963, 2003.
Capaday C. The special nature of human walking and its neural control. Trends Neurosci 25: 370-376, 2002.
Cartmill M, Lemelin P, and Schmitt D. Support polygons and symmetrical gaits in mammals. Zool J Linn Soc 136: 401-
420, 2002.
Courtine G, Hodgson JA, Raven J, Roy RR, McKay H, Bernot T, Yang H, Tuszynski MH, and Edgerton VR. Effects
of a unilateral corticospinal tract (CST) lesion or a complete cervical hemisection (HS) on locomotor performance in a
primate. Washington, DC: Society for Neuroscience, 2004, Program No. 654.657.
Courtine G and Schieppati M. Human walking along a curved path. II. Gait features and EMG patterns. Eur J Neurosci
18: 191-205, 2003.
Courtine G and Schieppati M. Tuning of a basic coordination pattern constructs straight-ahead and curved walking in
humans. J Neurophysiol 91: 1524-1535, 2004.
D'Aout K, Aerts P, De Clercq D, De Meester K, and Van Elsacker L. Segment and joint angles of hind limb during
bipedal and quadrupedal walking of the bonobo (Pan paniscus). Am J Phys Anthropol 119: 37-51, 2002.
D'Aout K, Vereecke E, Schoonaert K, De Clercq D, Van Elsacker L, and Aerts P. Locomotion in bonobos (Pan
paniscus): differences and similarities between bipedal and quadrupedal terrestrial walking, and a comparison with other
locomotor modes. J Anat 204: 353-361, 2004.
de Leon R, Hodgson JA, Roy RR, and Edgerton VR. Extensor- and flexor-like modulation within motor pools of the rat
hindlimb during treadmill locomotion and swimming. Brain Res 654: 241-250, 1994.
Dietz V, Fouad K, and Bastiaanse CM. Neuronal coordination of arm and leg movements during human locomotion. Eur
J Neurosci 14: 1906-1914, 2001.
Drew T, Jiang W, and Widajewicz W. Contributions of the motor cortex to the control of the hindlimbs during
locomotion in the cat. Brain Res Brain Res Rev 40: 178-191, 2002.
Drew T, Prentice S, and Schepens B. Cortical and brainstem control of locomotion. Prog Brain Res 143: 251-261, 2004.
Drew T. Motor cortical activity during voluntary gait modifications in the cat. I. Cells related to the forelimbs. J
Neurophysiol 70: 179-199, 1993.
Dunbar DC and Badam GL. Development of posture and locomotion in free-ranging primates. Neurosci Biobehav Rev
22: 541-546, 1998.
Duysens J and Pearson KG. Inhibition of flexor burst generation by loading ankle extensor muscles in walking cats.
Brain Res 187: 321-332, 1980.
Kinematic and EMG patterns during rhesus monkey locomotion
p.24
Duysens J and Van de Crommert HW. Neural control of locomotion; The central pattern generator from cats to humans.
Gait Posture 7: 131-141, 1998.
Edgerton VR, Leon RD, Harkema SJ, Hodgson JA, London N, Reinkensmeyer DJ, Roy RR, Talmadge RJ,
Tillakaratne NJ, Timoszyk W, and Tobin A. Retraining the injured spinal cord. J Physiol 533: 15-22, 2001.
Edgerton VR and Roy RR. Paralysis recovery in humans and model systems. Curr Opin Neurobiol 12: 658-667, 2002.
Edgerton VR, Tillakaratne NJ, Bigbee AJ, de Leon RD, and Roy RR. Plasticity of the spinal neural circuitry after
injury. Annu Rev Neurosci 27: 145-167, 2004.
Eidelberg E, Walden JG, and Nguyen LH. Locomotor control in macaque monkeys. Brain 104: 647-663, 1981.
English AW. An electromyographic analysis of compartments in cat lateral gastrocnemius muscle during unrestrained
locomotion. J Neurophysiol 52: 114-125, 1984.
English AW. An electromyographic analysis of forelimb muscles during overground stepping in the cat. J Exp Biol 76:
105-122, 1978.
Fedirchuk B, Nielsen J, Petersen N, and Hultborn H. Pharmacologically evoked fictive motor patterns in the acutely
spinalized marmoset monkey (Callithrix jacchus). Exp Brain Res 122: 351-361, 1998.
Fleshman JW, Lev-Tov A, and Burke RE. Peripheral and central control of flexor digitorum longus and flexor hallucis
longus motoneurons: the synaptic basis of functional diversity. Exp Brain Res 54: 133-149, 1984.
Georgopoulos AP and Grillner S. Visuomotor coordination in reaching and locomotion. Science 245: 1209-1210, 1989.
Grillner S. Control of locomotion in biped, tetrapod and fish. In: Handbook of physiology, The nervous system II, edited by
Edition B: Bethesda MD:American physiological physiology, 1981, p. 1179-1236.
Grillner S and Rossignol S. On the initiation of the swing phase of locomotion in chronic spinal cats. Brain Res 146: 269-
277, 1978.
Halbertsma J, Miller S, Van Der Meche F, and R. M. Herman SG, P. S. G. Stein, and D. G. Stuart,. Basic programs
for the phasing of flexion and extension movements of the limbs during locomotion. In: Neural control of locomotion,
edited by Hermna RM, Grillner S, Stein P and Stuart DG. New York: Plenum Press, 1976, p. 489-517.
Harris CM and Wolpert DM. Signal-dependent noise determines motor planning. Nature 394: 780-784, 1998.
Hildebrand M. Symmetrical gait of primates. Am J Phys Anthropol 26: 119-130, 1967.
Hirasaki E, Ogihara N, Hamada Y, Kumakura H, and Nakatsukasa M. Do highly trained monkeys walk like humans?
A kinematic study of bipedal locomotion in bipedally trained Japanese macaques. J Hum Evol 46: 739-750, 2004.
Hodgson JA, Wichayanuparp S, Recktenwald MR, Roy RR, McCall G, Day MK, Washburn D, Fanton JW,
Kozlovskaya I, and Edgerton VR. Circadian force and EMG activity in hindlimb muscles of rhesus monkeys. J
Neurophysiol 86: 1430-1444, 2001.
Ivanenko YP, Dominici N, Cappellini G, Dan B, Cheron G, and Lacquaniti F. Development of pendulum mechanism
and kinematic coordination from the first unsupported steps in toddlers. J Exp Biol 207: 3797-3810, 2004a.
Ivanenko YP, Grasso R, Macellari V, and Lacquaniti F. Two-thirds power law in human locomotion: role of ground
contact forces. Neuroreport 13: 1171-1174, 2002.
Ivanenko YP, Poppele RE, and Lacquaniti F. Five basic muscle activation patterns account for muscle activity during
human locomotion. J Physiol 556: 267-282, 2004b.
Jankowska E, Hammar I, Slawinska U, Maleszak K, and Edgley SA. Neuronal basis of crossed actions from the
reticular formation on feline hindlimb motoneurons. J Neurosci 23: 1867-1878, 2003.
Courtine et al.
p.25
Jiang W and Drew T. Effects of bilateral lesions of the dorsolateral funiculi and dorsal columns at the level of the low
thoracic spinal cord on the control of locomotion in the adult cat. I. Treadmill walking. J Neurophysiol 76: 849-866, 1996.
Jones LL, Oudega M, Bunge MB, and Tuszynski MH. Neurotrophic factors, cellular bridges and gene therapy for spinal
cord injury. J Physiol 533: 83-89, 2001.
Kimura T. Bipedal and quadrupedal walking of primates: Comparative dynamics. In: Primate Morphophysiology,
Locomotor Analyses and Human Bipedalism, edited by Kondo S. Tokyo: Tokyo University Press, 1985, p. 81-104.
Kuypers HG. The motor system and the capacity to execute highly fractionated distal extremity movements.
Electroencephalogr Clin Neurophysiol 34: 429-431, 1978.
Lacquaniti F, Grasso R, and Zago M. Motor patterns in walking. News Physiol Sci 14: 168-174, 1999.
Lacquaniti F, Ivanenko YP, and Zago M. Kinematic control of walking. Arch Ital Biol 140: 263-272, 2002.
Lacquaniti F, Terzuolo C, and Viviani P. The law relating the kinematic and figural aspects of drawing movements. Acta
Psychol (Amst) 54: 115-130, 1983.
Larney E and Larson SG. Compliant walking in primates: elbow and knee yield in primates compared to other mammals.
Am J Phys Anthropol 125: 42-50, 2004.
Larson SG. Unique aspects of quadrupedal locomotion in nonhuman primates. In: Primate Locomotion: Recent Advances,
edited by Strasser E, Fleagle JG, Rosenberger AL and McHenry HM. New York: Plenum Press, 1998, p. 157-173.
Larson SG, Schmitt D, Lemelin P, and Hamrick M. Uniqueness of primate forelimb posture during quadrupedal
locomotion. Am J Phys Anthropol 112: 87-101, 2000.
Lavoie S, McFadyen B, and Drew T. A kinematic and kinetic analysis of locomotion during voluntary gait modification
in the cat. Exp Brain Res 106: 39-56, 1995.
Lawrence DG and Kuypers HG. The functional organization of the motor system in the monkey. I. The effects of
bilateral pyramidal lesions. Brain 91: 1-14, 1968.
Manter JT. The dynamics of quadrupedal walking. J Exp Biol 15: 522-540, 1938
Miller S, Van Der Burg J, and Van Der Meche F. Coordination of movements of the kindlimbs and forelimbs in
different forms of locomotion in normal and decerebrate cats. Brain Res 91: 217-237, 1975.
Molenaar I and Kuypers HG. Cells of origin of propriospinal fibers and of fibers ascending to supraspinal levels. A HRP
study in cat and rhesus monkey. Brain Res 152: 429-450, 1978.
Mori F, Nakajima K, Tachibana A, Takasu C, Mori M, Tsujimoto T, Tsukada H, and Mori S. Reactive and
anticipatory control of posture and bipedal locomotion in a nonhuman primate. Prog Brain Res 143: 191-198, 2004.
Mori F, Tachibana A, Takasu C, Nakajima K, and Mori S. Bipedal locomotion by the normally quadrupedal Japanese
monkey, M. Fuscata: strategies for obstacle clearance and recovery from stumbling. Acta Physiol Pharmacol Bulg 26: 147-
150, 2001.
Mori S, Miyashita E, Nakajima K, and Asanome M. Quadrupedal locomotor movements in monkeys (M. Fuscata) on a
treadmill: kinematic analyses. Neuroreport 7: 2277-2285, 1996.
Morton SM and Bastian AJ. Cerebellar control of balance and locomotion. Neuroscientist 10: 247-259, 2004.
Nathan PW, Smith M, and Deacon P. Vestibulospinal, reticulospinal and descending propriospinal nerve fibres in man.
Brain 119: 1809-1833, 1996.
Nielsen J. Motoneuronal drive during human walking. Brain Res Brain Res Rev 40: 192-201, 2002.
Nielsen JB. How we walk: central control of muscle activity during human walking. Neuroscientist 9: 195-204, 2003.
Kinematic and EMG patterns during rhesus monkey locomotion
p.26
Nilsson J, Thorstensson A, and Halbertsma J. Changes in leg movements and muscle activity with speed of locomotion
and mode of progression in humans. Acta Physiol Scand 123: 457-475, 1985.
O'Donovan MJ, Pinter MJ, Dum RP, and Burke RE. Actions of FDL and FHL muscles in intact cats: functional
dissociation between anatomical synergists. J Neurophysiol 47: 1126-1143, 1982.
Okada M. Primate bipedal walking: Comparative kinematics. In: Primate Morphophysiology, Locomotor Analyses and
Human Bipedalism, edited by Kondo S. Tokyo: Tokyo University Press, 1985, p. 47-58.
Orlovskii GN, Deliagina TG, and Grillner S. Neuronal Control of Locmotion: From Mollusc to Man. Oxford: Oxford
University Press, 1999.
Park MC, Belhaj-Saif A, and Cheney PD. Properties of primary motor cortex output to forelimb muscles in rhesus
macaques. J Neurophysiol 92: 2968-2984, 2004.
Pearson KG. Generating the walking gait: role of sensory feedback. Prog Brain Res 143: 123-129, 2004.
Pierotti DJ, Roy RR, Gregor RJ, and Edgerton VR. Electromyographic activity of cat hindlimb flexors and extensors
during locomotion at varying speeds and inclines. Brain Res 481: 57-66, 1989.
Rasmussen S, Chan AK, and Goslow GE, Jr. The cat step cycle: electromyographic patterns for hindlimb muscles during
posture and unrestrained locomotion. J Morphol 155: 253-269, 1978.
Recktenwald MR, Hodgson JA, Roy RR, Riazanski S, McCall GE, Kozlovskaya I, Washburn DA, Fanton JW, and
Edgerton VR. Effects of spaceflight on rhesus quadrupedal locomotion after return to 1G. J Neurophysiol 81: 2451-2463,
1999.
Rho MJ, Lavoie S, and Drew T. Effects of red nucleus microstimulation on the locomotor pattern and timing in the intact
cat: a comparison with the motor cortex. J Neurophysiol 81: 2297-2315, 1999.
Rossignol S. Neural control of stereotypic limb movement. In: Handbook of Physiology. Exercise: Regulation and
Integration of Multiple Systems: Bethesda, MD: Am. Physiol. Soc., 1996, p. 173-216.
Rossignol S, Bouyer L, Langlet C, Barthelemy D, Chau C, Giroux N, Brustein E, Marcoux J, Leblond H, and
Reader TA. Determinants of locomotor recovery after spinal injury in the cat. Prog Brain Res 143: 163-172, 2004.
Roy RR, Bodine-Fowler SC, Kim J, Haque N, de Leon R, Rudolph W, and Edgerton VR. Architectural and fiber type
distribution properties of selected rhesus leg muscles: feasibility of multiple independent biopsies. Acta Anat (Basel) 140:
350-356, 1991a.
Roy RR, Hutchison DL, Pierotti DJ, Hodgson JA, and Edgerton VR. EMG patterns of rat ankle extensors and flexors
during treadmill locomotion and swimming. J Appl Physiol 70: 2522-2529, 1991b.
Schmitt D. Compliant walking in primates. J Zool 247: 149-160, 1999.
Schmitt D. Insights into the evolution of human bipedalism from experimental studies of humans and other primates. J Exp
Biol 206: 1437-1448, 2003.
Schoner G, Jiang WY, and Kelso JA. A synergetic theory of quadrupedal gaits and gait transitions. J Theor Biol 142:
359-391, 1990.
Schwartz AB and Moran DW. Motor cortical activity during drawing movements: population representation during
lemniscate tracing. J Neurophysiol 82: 2705-2718, 1999.
Shen L and Poppele RE. Kinematic analysis of cat hindlimb stepping. J Neurophysiol 74: 2266-2280, 1995.
Shik ML, Severin FV, and Orlovskii GN. Control of walking and running by means of electric stimulation of the
midbrain. Biofizika 11: 659-666, 1966. (In Russian)
Courtine et al.
p.27
Ting LH, Kautz SA, Brown DA, and Zajac FE. Contralateral movement and extensor force generation alter flexion
phase muscle coordination in pedaling. J Neurophysiol 83: 3351-3365, 2000.
Trank TV, Chen C, and Smith JL. Forms of forward quadrupedal locomotion. I. A comparison of posture, hindlimb
kinematics, and motor patterns for normal and crouched walking. J Neurophysiol 76: 2316-2326, 1996.
Tuszynski MH, Grill R, Jones LL, McKay HM, and Blesch A. Spontaneous and augmented growth of axons in the
primate spinal cord: effects of local injury and nerve growth factor-secreting cell grafts. J Comp Neurol 449: 88-101, 2002.
Vilensky JA. Gait characteristics of two macaques, with emphasis on relationships with speed. Am J Phys Anthropol 61:
255-265, 1983.
Vilensky JA. Locomotor behavior and control in human and non-human primates: comparisons with cats and dogs.
Neurosci Biobehav Rev 11: 263-274, 1987.
Vilensky JA. Primate quadrupedalism: how and why does it differ from that of typical quadrupeds? Brain Behav Evol 34:
357-364, 1989.
Vilensky JA, Moore AM, Eidelberg E, and Walden JG. Recovery of locomotion in monkeys with spinal cord lesions. J
Mot Behav 24: 288-296, 1992.
Vilensky JA and O'Connor BL. Stepping in humans with complete spinal cord transection: a phylogenetic evaluation.
Motor Control 1: 284–292, 1997.
Vilensky JA and O'Connor BL. Stepping in nonhuman primates with a complete spinal cord transection: old and new
data, and implications for humans. Ann N Y Acad Sci 860: 528-530, 1998.
Wannier T, Bastiaanse C, Colombo G, and Dietz V. Arm to leg coordination in humans during walking, creeping and
swimming activities. Exp Brain Res 141: 375-379, 2001.
Yang H, McKay H, Hodgson JA, Bernot T, Raven J, Courtine G, Lu P, Blesch A, Havton LA, Roy RR, Edgerton
VR, and Tuszynski MH. BDNF and NT3 gene delivry in models of primate spinal cord injury (SCI). Washington, DC:
Society for Neuroscience, 2004, Program No. 106.114.
Zajac FE, Neptune RR, and Kautz SA. Biomechanics and muscle coordination of human walking: Part II: lessons from
dynamical simulations and clinical implications. Gait Posture 17: 1-17, 2003.
Zehr EP and Duysens J. Regulation of arm and leg movement during human locomotion. Neuroscientist 10: 347-361,
2004.
Zernicke and Smith JL. Biomechanical insights into neural control of movement. In: Handbook of Physiology. Exercise:
Regulation and Integration of Multiple Systems: Bethesda, MD: Am. Physiol. Soc., 1996, p. 293-330.
Kinematic and EMG patterns during rhesus monkey locomotion
p.28
Legends
Figure 1. Footfall patterns Gait diagrams obtained from a typical Rhesus (#1) locomoting at 0.45, 0.89, 1.34, 1.79 m/s, from top to bottom,
respectively. Five complete gait cycles of the right hindlimb (HL) are represented for each treadmill speed. Each box
indicates the time during which a given limb is in contact with the ground (stance phase). The histograms on the left show
the percentage (± SD) of time during which 1, 2, 3, or 4 limbs are simultaneously in contact with the ground during a
complete right HL gait cycle.
Figure 2. Spatial and temporal features of the gait pattern A. Relationship between the mean body velocity and the time that the contralateral (left) HL (LHL), ipsilateral (right)
forelimb (RFL), and contralateral (left) FL (LFL) contact the ground relative to right HL contact expressed as a percentage
of the right (ipsilateral) HL cycle duration. The shaded area indicates the duration during which the right HL is in contact
with the ground (stance). The triangles depict data from Rhesus #5 (see text for details). B. Relationship between mean
body velocity and cycle duration for the right HL and FL obtained from a representative Rhesus (Rhesus #1). Mean
correlation coefficient values (±SD) computed independently for the HL and FL in each Rhesus (n=6) are shown. C.
Relationship between mean body velocity and stride length for Rhesus #1. Mean correlation coefficient values (±SD)
computed independently for the HL and FL in each Rhesus (n=6) are shown. D. Mean values of the horizontal distance
between the endpoints of the right HL and FL at the time of contact (see inset). Bars with asterisk indicate significant
differences between HL and FL E. Mean values of the stance phase duration for the HL and FL expressed as a percentage
of cycle duration for each treadmill speed. Significant differences between HL and FL are reported. F. Relationship
between cycle duration and the duration of the stance and swing phases for the HL and FL for Rhesus #1. Mean correlation
coefficients values (±SD) computed independently for the HL and FL in each Rhesus (n=6) are shown (p < 0.05 for all
relationships).
Courtine et al.
p.29
Figure 3. Time course of hindlimb and forelimb joint angle changes during the gait cycle
The mean waveforms (thick line ± SD, thin lines) of each joint angle recorded in a representative Rhesus (#4) during each
complete step cycle (n=10) at each treadmill speed is plotted vs. the normalized cycle duration. The horizontal bars at the
bottom indicate stance phase duration at each speed. The stick diagrams at the top show the successive HL and FL
positions during the stance and swing phases reconstructed from a representative step cycle at 0.79 m/s. The conventions
used to compute angles are shown in the insets at the top. The top and bottom of the vertical histograms at the right of each
joint plot show the mean (±SD) maximum and minimum joint angle positions, respectively, at each treadmill speed for six
Rhesus.
Figure 4. Control of limb endpoint trajectories
A. Superimposed endpoint trajectories of the HL (tip of the fifth toe) and FL (tip of the fifth finger) during the swing phase
for 6 consecutive step cycles at each treadmill speed for Rhesus #4. The plots are anisotropic with the vertical scale being
expanded relative to the horizontal scale. The top diagrams display the typical schematic decomposition of HL and FL
movements during the swing of gait. Endpoint trajectory (continuous line) is also shown, the scale being isotropic B.
Relationship (in logarithmic scales) between the angular velocity and curvature of HL (left) and FL (right) endpoints in the
same animal. All samples corresponding to the swing phases at 0.89 m/s were pooled in these examples. All linear
regressions were significant (p < 0.001). Linear regression analysis (black line) was performed to estimate the exponent β
from the slope. The grey line indicates the two-third slope, i.e. β = 0.66. C. Mean exponent (β) values (±SD) of the angular
velocity–curvature relationship for the HL and FL at each treadmill speed for six Rhesus are shown. The horizontal dotted
line designates the two-third value. Asterisks indicate a lower value in the FL than the HL at p < 0.05.
Figure 5. Characteristics of segment angular oscillations with increasing speed A. The mean waveforms (for 6 Rhesus) of HL and FL elevation angles (with respect to the direction of gravity) computed
from all step cycles at each treadmill speed are plotted vs. the normalized cycle duration. The circles identify the point of
maximal backward oscillation for each segment and point out the phase-relationships between the different limb segments.
B. Mean values (for 6 Rhesus) of the shift in timing of HL and FL segment oscillations with increasing treadmill speed are
shown. These values correspond to the phase lag required to provide the best fit between the angular waveforms at 0.89,
1.34, and 1.79 m/s relative to those at 0.45 m/s. Changes in the relative duration of the stance phase with increasing speed
are represented by the filled squares. Increase in speed corresponds to a decrease in stance duration (Fig. 2), which is
reported as a percent of change in cycle duration. The maximal positive value of r obtained from each cross-correlation
function was averaged across speeds and animals for each segment and the values (±SD) are shown to the right of the plot
(all linear regressions were significant, p < 0.001).
Figure 6. Pattern of inter-segmental coordination in Rhesus gait A. The 3-D gait loops obtained by plotting thigh, shank, and foot elevation angles for the HL or arm, forearm, and hand
elevation angles for the FL vs. each other at the different treadmill speeds are shown for Rhesus #4. The mean value for
each angular coordinate has been subtracted. The time of stance (St) and swing (Sw) onset is indicated by the black circles.
The arrows indicate the direction along which the 3-D gait loop evolves during stance and swing. The grids identify the
best-fitting planes whose intersections with the cubic wire frame provide information about its spatial orientation. The
Kinematic and EMG patterns during rhesus monkey locomotion
p.30
planarity of the 3-D gait loops was computed as the percent variance accounted for by PC1 plus PC2. The mean values (±
SD) (for 6 Rhesus) of the planarity index are reported above each 3-D plot. The vertical lines connecting the data points to
the grid indicate the distance between the 3-D gait loops and the best-fitting plane. B. Mean values (SD) (for 6 Rhesus) of
the variance accounted for by each principal component (PC) for HL or FL segment oscillations (all treadmill speeds
pooled) are shown. Asterisks indicate significant differences between the HL and FL values. C. Mean (SD) values (for 6
Rhesus) of the variance accounted for by each PC (PC5 to PC8 are summed) when PC analysis was applied to HL and FL
segment oscillations simultaneously (all treadmill speeds pooled). The cumulative variance accounted for by each PC is
shown above each histogram bar. D. The variance accounted for by PC1 plus PC2 (PC applied on HL and FL segments
simultaneously) is plotted vs. the mean body velocity. All the extracted gait cycles from the 6 Rhesus are reported. The
linear regression was significant (p < 0.05).
Figure 7. EMG activity of HL and FL muscles Mean (rectified) EMG activity of HL and FL muscles plotted vs. the normalized gait cycle duration at each treadmill speed
(as shown for the EDL). HL EMG traces are the ensemble average of all gait cycles performed by Rhesus #3 (knee and
ankle joints) and #1 (MT joint). HL waveforms are aligned at the onset of the right Sol EMG burst. Mean EMG activities
for the right FL muscles are illustrated for both Rhesus #5 and #6. FL waveforms are aligned at paw contact of the right FL.
The horizontal bars at the bottom of the FL traces indicate the mean duration of stance and swing phase for each Rhesus.
Figure 8. Temporal and spatial characteristics of EMG bursts (Typical relationships between EMG burst duration (top) and integral (bottom) for the HL and FL muscles vs. the cycle
duration are shown. HL-related data were obtained from Rhesus (#1 and #3). FL-related relationships were obtained from
Rhesus #5. In addition, diamond data points illustrate the temporal modulation of the EDC and swing-related FDP EMG
bursts with respect to cycle duration for Rhesus #6.
Figure 9. Temporal coordination between muscle activity and oscillations of HL and FL
segments The timing of FL (top) and HL (bottom) segment oscillations is plotted vs. the temporal features related to the end of stance
of all right FL muscles for all step cycles from Rhesus #5. This corresponds to the timing of burst termination for all
muscles but the Bic for which burst onset was related to stance end (Fig. 7). Timing of segment oscillations were computed
as the phase φ of the first order Fourier series component of the angle, and expressed as a percent of the normalized gait
cycle duration. The minimum value recorded in FL segments was set at zero, and all HL and FL values were modified
accordingly. The relative timing between the oscillation of HL and FL segments is preserved. Similar relationships were
detected in Rhesus #6 (data not shown). All the linear regression were significant (p < 0.05)
Figure 10. Coordination between the EMG activity of HL and FL muscles
The EMG burst integral (non-time normalized) for each HL muscle is computed for each gait cycle and plotted vs. the right
Sol and TA EMG burst integral during the same gait cycle for the HL (top), and plotted vs. the right Tri and Bic EMG burst
integral for the FL (bottom). These relationships were consistent among the Rhesus in the HL, and representative plots are
displayed. These relationships, however, were less consistent in the FL and data are shown for both Rhesus #5 and #6.
Courtine et al.
p.31
Mean (± SD, if n > 1) correlation coefficient (r) values are reported. The EMG burst integral of each muscle from each
Rhesus was normalized to its mean value computed during walking at the slowest treadmill speed, i.e., at 0.45 m/s. Left and
right muscles are identified by open and filled symbols, respectively. FHL and EDL, FDS and EDC were recorded from the
right muscles only, and are represented by open and filled circles, respectively.
Kinematic and EMG patterns during rhesus monkey locomotion
p.32
HINDLIMB Abb. Main action Monkey
Vastus lateralis VL Knee extension # 3 (LR)
Medial Gastrocnemius MG Plantarflexion / Knee flexion # 1 # 2 # 3 (all LR)Soleus Sol Plantarflexion # 1 (R) # 2 (R) # 3 (LR)Tibialis Anterior TA Dorsiflexion / inversion # 1 # 2 # 3 (all LR)
Flexor digitorum longus FDL 2nd - 5th toes flexion / Weak plantarflexion # 1 # 2 (R)Extensor digitorum longus EDL Toe extension / Weak dorsiflexion # 1 # 2 (R)Flexor hallucis longus FHL Big toe flexion / Plantarflexion # 1 # 2 (R)
FORELIMB Abb. Main action MonkeyTriceps brachii medial head Tri Forearm extension # 5 # 6 (R)Biceps long head Bic Forearm flexion and supination / Arm flexion # 5 # 6 (R)
Extensor digitorum communis EDC Medial digit extension / Wrist extension # 5 # 6 (R)Flexor digitorum profundus FDP Distal PH flexion / Assist proximal PH and wrist flexion # 5 # 6 (LR)Flexor digitorum superficialis FDS Middle-proximal PH digit flexion / Wrist flexion # 5 # 6 (R)Flexor pollicis brevis profundus FPB Proximal PH thumb flexion # 5 # 6 (LR) L, left; R, right; Abb., abbreviations; PH, phalange
Table 1. Summary of muscles studied and their primary actions.
Courtine et al.
p.33 FIG. 1
0.89
m/s
1.34
m/s
0.45
m/s4
21
34
21
3
Lim
b in
co
ntac
t
Left
Rig
ht
1 s
1.79
m/s
0
50
100
% o
f tim
e
HLFL
Kinematic and EMG patterns during rhesus monkey locomotion
p.34
FIG. 2
Swing
0
0.2
0.4
0.6
0.8
0.5 0.7 0.9 1.1Cycle duration (s)
Stance FHLFL
r = 0.99 ± 0.01
r = 0.69 ± 0.18
Swing
0
0.2
0.4
0.6
0.8
0.5 0.7 0.9 1.1Cycle duration (s)
Stance FHLFL
r = 0.99 ± 0.01
r = 0.69 ± 0.18
RFL
Con
tact
tim
e (%
) ALHLRFLLFL
distance
-200
20406080
0.45 0.89 1.34 1.790.45 0.89 1.34 1.79Inte
r-lim
b di
stan
ce a
t co
ntac
t (cm
)
D+
Treadmill speed (m/s)St
ance
dur
atio
n(%
of c
ycle
dur
atio
n)0.45 0.89 1.34 1.79
0
25
50
75
HLHLFLFL
E
Dur
atio
n (s
)
0.4
0.8
1.2
0 0.5 1 1.5 2
Mean body velocity (m/s)
Strid
e le
ngth
(m) C
0.4
0.8
1.2
0 0.5 1 1.5 2
Cyc
le d
urat
ion
(s)
B
-200
20406080
0.5 1 1.5 2
#5
HLFL
HL : r = - 0.97 ± 0.01FL : r = - 0.97 ± 0.01
HL : r = 0.98 ± 0.01FL : r = 0.97 ± 0.01
* *
* *
* * *
Courtine et al.
p.35
FIG. 3
MCPwrist
30
ankle
hip
Hindlimb Forelimb
0 25 50 75 100
30
70
110
Hip
(deg
)
50
100
150
Kne
eM
T
60
110
160
Ankl
e
Cycle duration (%)
100
150
200
Sho
ulde
rE
lbow
MC
PW
rist
0 25 50 75 1000 25 50 75 100
Flex
Flex
120
70
Ret
r
280
200
120
0.450.891.341.79
knee
MT
Stance Swing
240
170
100 Vent
roFl
ex
shoulderelbow
Cycle duration (%)
Stance Swing
Trea
dmill
spee
d (m
/s)
260
210
160 Vent
roFl
exFl
exFl
exFlex
Flex
Flex
Flex
Flex
Flex
Kinematic and EMG patterns during rhesus monkey locomotion
p.36
FIG. 4
Angu
lar v
eloc
ity(ra
d/s)
10-3 1 102 104
10-1
10
103
Curvature (m-1)
β = 0.68r = 0.99β = 0.68r = 0.99
Hindlimb Forelimb
β = 0.71r = 0.99
β = 0.66
FLHL
A
B
0
0.2
0.4
0.6
0.80.66
β
0.45 0.89 1.34 1.79Treadmill speed (m/s)
C
0.45
0.89
1.34
1.792 cm2 cm
4 cm
4 cm
β = 0.66
* * * *
Courtine et al.
p.37
FIG. 5
Hindlimb ForelimbE
leva
tion
angl
e (d
eg)
Forw
ard
Backw
ardFo
rwar
dB
ackward
Cycle duration (%) Cycle duration (%)
ArmForearm
HandFinger
ThighShankFootToe-100
-500
50100
-60-30
04590
0 25 50 75 100 0 25 50 75 100
0
5
10
15
0
6
12
18
0.8
1
0.8
1StanceStance
0.45 0.89 1.791.34Treadmill speed (m/s)
Pha
se la
g
(% o
f cyc
le)
r val
ue
0.45 0.89 1.791.34Treadmill speed (m/s)
A
B
0.45 m/s
0.89 m/s
1.79 m/s
1.34 m/s
r val
ue
Kinematic and EMG patterns during rhesus monkey locomotion
p.38
FIG. 6
Foot
Thigh (deg)Shank -4040
60-50
60
-60 Han
d
Arm (deg)Forearm -40
4070
-70
80
-80
stance
swin
g
0
50
100
Var
ianc
eac
coun
ted
for (
%)
PC1 PC2 PC3 PC4
HLFLHLFL
B
0
50
100
PC1 PC2 PC3 PC4
60%
91%
97%99%100%
C
70
80
90
100
0 0.5 1 1.5 2
D
Mean body velocity (m/s)
PC1 plus PC2
Hindlimb Forelimb
0.45 m/s
0.89 m/s
1.79 m/s
1.34 m/s
Sw
St
stan
ce
swin
g
A
PC5-8
97.8 ± 0.66 98.7 ± 0.50
98.0 ± 0.56 98.5 ± 0.34
97.9 ± 0.52 98.0 ± 0.52
98.1 ± 0.57 97.9 ± 0.4
St
Sw
Var
ianc
eac
coun
ted
for (
%)
Var
ianc
eac
coun
ted
for (
%)
*
*
r = 0.80
Courtine et al.
p.39
FIG. 7
ED
LVL
Sol
TAFH
LFD
LM
G
Hindlimb Forelimb
Tri
Bic
Right Left # 5 (right) # 6(right)
KN
EEA
NK
LED
IGIT
S
ELB
OW
DIG
ITS
0Cycle duration (%)
50 1000Cycle duration (%)
50 100
SolonsetSolonset
0Cycle duration (%)
50 1000Cycle duration (%)
50 100
Paw contactPaw contact
ED
CE
DC
FDS
FDP
FPB
STANCE SWING0.45 m/s0.89 m/s
1.79 m/s1.34 m/s
Kinematic and EMG patterns during rhesus monkey locomotion
p.40
FIG. 8
# 6TriTri
FHL
FHL
Knee
Hindlim
bForelim
bTIM
ING
AMPLITU
DE
Digits
ElbowD
igits
Cycle duration (s)
Burst duration (s)
Cycle duration (s)
Normalized Burst integral
VLVLM
GM
GS
LS
L
TATA
EDL
EDL
FDL
FDL
Bic
Bic
ED
CE
DC
FDP
FDP
00.20.40.60.8
0.40.6
0.81
Ankle
# 5
0 1 2 3
0 1 2 3 4 5
0 2 4 6 8
0 2 4
0 1 2 3
0.40.6
0.81
0 2 4 6VLVL
MG
MG
SL
SL
TATA
FHL
FHL
EDL
EDL
FDL
FDL
TriTriB
icB
ic
ED
CE
DC
FPB
FPB
FDP
FDP
FDS
FDS
FPB
FPB
FDS
FDS
EDC
EDC
FDP
FDP
KneeD
igitsElbow
Digits (#5)
Ankle
TIMIN
G
AMPLITU
DE
Courtine et al.
p.41
FIG. 9
Tim
ing
of s
egm
ent o
scill
atio
n (%
of c
ycle
dur
atio
n)
FDS FDP EDC FPB
Temporal burst feature (% of cycle duration)
01020304050
30 80 30 80 35 85 20 65 20 80 25 65
Bic TriForelimb
HindlimbFDS FDP EDC FPB Bic TriFDS FDP EDC FPB Bic Tri
20
40
60
80
30 80 30 80 35 85 20 65 20 80 25 65
Arm ForearmHand FingerArm ForearmHand Finger
Thigh ShankFoot ToeThigh ShankFoot Toe
Kinematic and EMG patterns during rhesus monkey locomotion
p.42
FIG. 10
Hindlimbbu
rst i
nteg
ral
Right Sol burst integral
0
1
2
3
0
1
2VLTA
Right
Left
0
2
4
6
0.4 0.8 1.20
1
2
0
1
2
3FHL-EDL FDLMG
r = - 0.58 ± 0.1 r = - 0.74 ± 0.09 r = - 0.62 ± 0.04
r = - 0.67 ± 0.1
r = - 0.52 ± 0.02
r = 0.88
r = 0.93
r = - 0.81 ± 0.09
r = - 0.78 ± 0.09
0
1
2
0.4 1.4 2.40
2
4
6
0.4 1.9 3.40
1
2
3
0.4 1.9 3.40
1
2
0.4 1.9 3.40
1
2
3
0.4 1.9 3.4Right TA burst integral
VLTA FHL-EDL FDLMGr = 0.53 ± 0.04
r = 0.52 ± 0.23
r = - 0.71 ± 0.02
r = - 0.83
r = - 0.88
r = 0.81 ± 0.04
r = 0.65 ± 0.1
r = 0.46 ± 0.2
Forelimb
Right Tri burst integral
FDS-EDCFDPBic
FDS-EDCFDPBic
0
5
10
0 1 20
5
10
0 1 20
1
2
0 1 2
0
4
8
0.5 1 1.50
1
2
3
0.5 1 1.50
1
2
0.5 1 1.5
r = -0.17
r = 0.13 r = -0.15 r = 0.07
r = -0.03
r = 0.18 FPB
0
1
2
0.5 1 1.5
r = -0.12 r = -0.02
r = 0.86 r = -0.06
r = 0.83
r = 0.21 FPB
0
1
2
3
0 1 2
r = 0.81 r = 0.78
# 5
# 6
0
1
2
3
0 4 80
5
10
0 4 80
1
2
0 4 80
1
2
3
0 4 8
0
1
2
3
0 4 80
1
2
0 4 80
1
2
0 4 8
0.4 0.8 1.2 0.4 0.8 1.2 0.4 0.8 1.2 0.4 0.8 1.2
burs
t int
egra
lbu
rst i
nteg
ral
burs
t int
egra
lbu
rst i
nteg
ral
burs
t int
egra
l
FDS-EDCFDPTri
FDS-EDCFDPTri
FPB
FPB
0
1
2
0 4 8
Right Bic burst integral
# 5
# 6
r = -0.17
r = 0.13 r = -0.35r = 0.01
r = 0.19
r = 0.04 r = -0.27 r = -0.64
r = -0.1 r = -0.06
r = - 0.37
r = -0.05 r = -0.39 r = - 0.04