Direct evidence supporting the contribution of upper limb motion on the generation of locomotive motor output in humans is still limited. Here, we aimed to examine the effect of upper limb motion on locomotor-like muscle activities in the lower limb in persons with spinal cord injury (SCI). By imposing passive locomotion-like leg movements, all cervical incomplete (n = 7) and thoracic complete SCI subjects (n = 5) exhibited locomotor-like muscle activity in their paralyzed soleus muscles. Upper limb movements in thoracic complete SCI subjects did not affect the electromyographic (EMG) pattern of the muscle activities. This is quite natural since neural connections in the spinal cord between regions controlling upper and lower limbs were completely lost in these subjects. On the other hand, in cervical incomplete SCI subjects, in whom such neural connections were at least partially preserved, the locomotor-like muscle activity was significantly affected by passively imposed upper limb movements. Specifically, the upper limb movements generally increased the soleus EMG activity during the backward swing phase, which corresponds to the stance phase in normal gait. Although some subjects showed a reduction of the EMG magnitude when arm motion was imposed, this was still consistent with locomotor-like motor output because the reduction of the EMG occurred during the forward swing phase corresponding to the swing phase. The present results indicate that the neural signal induced by the upper limb movements contributes not merely to enhance but also to shape the lower limb locomotive motor output, possibly through interlimb neural pathways. Such neural interaction between upper and lower limb motions could be an underlying neural mechanism of human bipedal locomotion.
The relative phase of motion between upper and lower limbs during human bipedal walking is very similar to that observed between forelimbs and hindlimbs during quadrupedal walking. It is well recognized that the neural pathway connecting the cervical and lumbar cord (called the “propriospinal pathway”) has an important role in generating coordinated interlimb motion in animals (Cazalets and Bertrand 2000; Forssberg et al. 1980; Gernandt and Megirian 1961; Gernandt and Shimamura 1961; Miller et al. 1973, 1975; Skinner et al. 1980). Several previous investigations have suggested that such a pathway is also preserved in humans (Calancie 1991; Meinck and Piesiur-Strehlow 1981; Nathan et al. 1996; Sarica and Ertekin 1985). Moreover, recent electrophysiological studies have reported that the amplitude of the interlimb cutaneous reflex is modulated by the phase of walking (Balter and Zehr 2007; Dietz et al. 2001; Haridas and Zehr 2003; Zehr and Haridas 2003; Zehr and Kido 2001; Zehr et al. 2001) and is altered among different locomotor tasks (Sakamoto et al. 2006; Wannier et al. 2001; for review see Dietz 2002). These findings suggest that the interlimb neural pathway is likely to contribute to human bipedal locomotion (Dietz 2002; Zehr and Duysens 2004).
Recently, Ferris and colleagues reported that the lower limb muscle activity during cyclic stepping with a recumbent stepping device was significantly increased for healthy subjects when the active arm movements were performed concurrently (Huang and Ferris 2004; Kao and Ferris 2005). Although their results support the view that interlimb coupling contributes to the enhancement of muscle activity during human locomotion, it is likely that the augmentation of lower limb muscle activity was merely induced by increased descending commands concurrent with voluntary upper limb movements such as that seen when a Jendrassik maneuver was applied (Gassel and Diamantopoulos 1964). To conclude a contribution of interlimb coordination to the altered lower limb muscle activities, the effect of the descending neural command should be differentiated.
In the present study, we attempted to obtain more valid evidence for the contribution of interlimb coordination on the locomotive motor output by examining the effect of passive upper limb movement on the locomotor-like lower limb muscle activity of individuals with spinal cord injury (SCI). Considering that the locomotor-like muscle activity is regarded to have originated from the central pattern generator (CPG) located in the spinal cord (Dietz et al. 1995; Dimitrijvic et al. 1998; Kawashima et al. 2005a), the muscle activity can be utilized as a tool to examine the characteristics of the spinal neural system. Importantly, if the involuntary locomotor-like EMG activity is affected by passive upper limb motion that is also involuntary, the result may be attributed to upper-limb–induced neural inputs mediated by an interlimb pathway rather than to the descending neural command.
We observed the locomotor-like muscle activity when additional upper arm motions were imposed in individuals with incomplete SCI at the cervical segments. Considering previous findings that interlimb reflex is still preserved after a cervical SCI (Calancie 1991; Calancie et al. 1996, 2002), we hypothesized that the upper limb motions, even if imposed passively, should influence locomotor-like muscle activity through an interlimb neural pathway in incomplete cervical SCI subjects. Additionally, we conducted the same experiment for individuals with complete SCI at the thoracic level, serving as a control group, who have no neural connections between the segments controlling upper and lower limbs. Since neuronal signals associated with upper limb motion are unable to reach the spinal cord segments innervating lower limb muscles in these patients, the neural input induced by upper arm motion should not affect the lower limb locomotor-like muscle activity for these subjects. Parts of the present study were previously presented in abstract form (Kawashima et al. 2005b).
Seven male SCI subjects (age 27.7 ± 7.5 yr) with incomplete cervical spinal cord injury participated in this study. In all subjects, ≥6 mo had passed since their injuries. These subjects had injuries at the third or more proximal cervical segments and were classified as C or D according to the American Spinal Injury Association Impairment Scale (ASIA; Maynard et al. 1997). Residual motor function of the six key muscles was assessed using the manual muscle test (MMT). Three of seven patients had a marked lateral dominance of the motor and sensory paralysis and the other four patients showed a similar extent of paralysis in both sides. As a control group, five male SCI subjects (27.1 ± 11.9 yr) with thoracic spinal cord injury between Th5 and Th12 (ASIA A or B) participated. All thoracic SCI subjects were judged clinically as motor complete and were unable to produce voluntary contraction in lower limb muscles. Regardless of the type of injury, all subjects had spasticity in their lower limbs, but none of the subjects took antispasticity medication. The characteristics of the patients are summarized in Table 1. Each subject gave written informed consent to the experimental procedures, which were conducted in accord with the Helsinki Declaration of 1975 and approved by the ethics committee of the National Rehabilitation Center for Persons with Disabilities (Tokorozawa, Japan).
Passive leg movement
The experimental apparatus used in this study was described previously (Kawashima et al. 2005a). Briefly, to impose locomotion-like movements on their legs, we used an apparatus (Fig. 1 A) originally developed for physical exercise for persons with disabilities (EasyStand Glider 6000, Altimate Medical, Morton, MN). This apparatus enables persons with SCI to stand securely by immobilizing their trunk and pelvis using front and back pads and by preventing hyperextension of the knee joint using a knee pad. It also enables the subjects to swing their legs by moving the handle connected to the foot plate. In this study, the SCI subjects were placed standing on the device (Fig. 1A) and the passive reciprocal leg movements were accomplished by moving the handle connected to the leg frame. The leg motion was measured using the signal obtained from an electrical goniometer (Goniometer System, UK Biometrics, Tyne and Wear, UK) attached to the lateral aspect of the device and was continuously monitored with an oscilloscope. With regard to the frequency of the leg swing motion, Drills (1958) reported that the mean value of cadence is 112 steps/min (varying from 78 to 144). Based on these values, we set 1 Hz (60 steps/min) as the motion frequency in this study. An experimenter moved the handle back and forth in a sinusoidal manner while keeping pace with the tempo of a metronome. This handle movement could induce approximately ±15 and ±9° motion in the hip and ankle joints, respectively (these values of range of joint motion depend on the subject's lower limb length). This range of motion of each joint is similar to the data for normal walking provided by Winter (1991).
First, the subjects were placed in a standing posture with support and we confirmed that the cervical SCI subjects could safely keep the passive upright standing posture without any abnormal sensations and orthostatic hypotension. Then, to habituate them to the imposed passive leg motion, the continuous movement was applied for 3 min.
The subjects were asked to relax their legs throughout the experiments. The following three different upper limb conditions were adopted. 1) Rest: the subjects placed both hands on a fixed bar and kept the upper limbs relaxed. 2) Passive: both hands were strapped to the handle using an elastic bandage; thus the upper limbs were moved passively with the motion of the handle. The subjects kept the upper limbs relaxed. 3) Active: both hands were strapped to the handle and the subjects tried to move their upper limbs voluntarily with the motion of the handle. To keep the target range of motion and frequency, both the upper and lower limb motions were induced by an experimenter so that the subjects conducted gentle arm motion during the Active condition. All three conditions were performed without unloading of body weight. Each session was conducted for 1 min and data from the last 30 s were used for later analysis. To ensure reproducibility of the data, the experiment was repeated twice under each condition in a randomized order.
Using a bipolar electrode, the surface electromyography (EMG) signal was recorded from the bilateral soleus (SOL), tibialis anterior (TA), and in both anterior (aDel) and posterior deltoid (pDel) muscles. Special care was taken for rejection of any artifacts in the EMG recording. For example, to reduce the resistance of the body surface, the skin was washed with scrub gel and wiped with sandpaper. After careful preparation of the skin, the electrodes were attached with double-sided adhesive tape. The EMG signal was amplified and band-pass filtered between 20 and 450 Hz (Bagnoli-8 EMG System, Delsys, Boston, MA).
A three-dimensional motion-analysis system (Vicon 370, Oxford Metrics, Oxford, UK) was used to analyze upper and lower limb movements. In all, 13 markers were attached to the subject's skin at the following sites: the vertex, both sides of the acromium (SHO), the lateral aspects of the hip (HIP), and ankle (AKL) joints, the top of the great toe (TOE), the protrusion of the ulna at the elbow (ELB), and the wrist joint (Fig. 1A). We defined the shoulder angle as the angle formed by the ELB, SHO, and HIP; the hip angle as the angle formed by the SHO, HIP, and AKL; and the ankle angle as that formed by the KNE, AKL, and TOE. Furthermore, the actual load applied to each foot sole was measured using four load cells (LMA-A-1KN, Kyowa Hakko Kogyo, Tokyo, Japan) placed under the four corners of the stainless steel foot plate (Fig. 1B). During the experiment, all data were continuously monitored by Power Lab software (Chart ver. 5, AD Instruments, Grand Junction, CO) and were digitized at 1 kHz for later analysis.
The ranges of hip, ankle, and shoulder movements were calculated from the data obtained by the three-dimensional motion analysis system. The load applied on each foot sole was quantified by the sum of the forces from the four load cells.
As mentioned in our previous article (Kawashima et al. 2005), the EMG activity often demonstrated gradual decay after the beginning of the motion, which is possibly explained by the exhaustion or fatigue of the spinal motoneurons, as previously examined by Dietz and Müller (2004). Therefore for the EMG analysis, the data of the first 30 cycles were discarded and it took about 30 s (i.e., 30 cycles) to stabilize. The signal was full-wave rectified after subtracting the offset level that was detected by the averaged value of the raw EMG signal. The magnitude of the locomotor-like EMG activity for each subject was quantified as the integrated value of the EMG waveform averaged over 30 cycles. The duration over which the muscle was active was calculated using the following criterion: the muscle was active when its averaged EMG signal consistently exceeded the level of resting EMG activity (mean value plus 3SD).
To examine phase modulation of the EMG activities, the mean amplitude of the EMG signal in each forward-swing phase and backward-swing phase was calculated. According to muscle activation patterns during ordinary bipedal walking, the Sol muscle is expected to activate in the backward-swing phase and disappear in the forward-swing phase (Winter 1991). Thus we used the following equation as an index for phase modulation of the Sol EMG activity A larger value of this index indicates a more suitable EMG modulation pattern during the locomotion cycle.
Values are given as the means ± SE. Two-way repeated-measures (RM) ANOVA was used to identify the effects of group and upper limb motion condition on the magnitude of the locomotor-like EMG activity. One-way RM ANOVA was used to compare the difference of the EMG activity in each forward- and backward-swing phase and modulation index among three conditions (Rest, Passive, and Active). If the result of ANOVA showed statistical significance, a multiple comparison (Bonferroni's method) was applied to identify differences between conditions. Significance was accepted at P < 0.05.
Rhythmic EMG activity during reciprocating passive leg movements
As shown in Fig. 1C, passive leg movements induced rhythmic EMG activity in the lower limb muscles. In the cervical incomplete SCI group, EMG activity in the Sol muscle was observed in all subjects during the backward-leg-swing phase corresponding to the stance phase in normal locomotion. The TA muscle showed EMG activity during the forward-leg-swing phase in five of seven subjects corresponding to the swing phase in normal locomotion. In the thoracic complete SCI group, EMG activity was present in the Sol but not in the TA muscle. These EMG activity patterns were almost identical to those observed in our previous study (Kawashima et al. 2005a).
Differences in EMG magnitude among the three conditions
Figure 2 shows the typical waveforms of the joint angle, load applied on the foot sole, and EMG activity for the three experimental conditions. As shown in this figure (top), the EMG magnitude in the Sol and TA muscles was different among the experimental conditions in the cervical incomplete SCI subjects. In contrast, the thoracic complete SCI subject showed similar EMG levels of these muscles in the three experimental conditions (bottom). Concerning EMG activity in upper limb muscles, one thoracic complete and three cervical incomplete SCI subjects showed EMG activity in the aDel muscle during the forward-arm-swing phase in the Passive arm motion condition. EMG activity in upper limb muscles was present in all subjects in the Active condition.
Figure 3 summarizes the magnitude of the integrated EMG over one cycle in the three conditions. Since data from both legs were included, the total number of samples per group was twice the number of subjects. No obvious change in the EMG magnitude among the three experimental conditions was observed for the thoracic complete SCI subjects (right panels). On the other hand, there were considerable differences in the EMG magnitude among the conditions in the cervical incomplete SCI subjects, although the modulation pattern (increase or decrease) varied from subject to subject. The results of two-way ANOVA indicate statistical significance of the effect of group [F(1,22) = 9.439, P < 0.05], but no significant effect of conditions [F(2,44) = 2.449, n.s.]. The intersubject variability was reflected in a larger coefficient of variance of Sol EMG of the cervical incomplete SCI subjects in both Passive and Active conditions, compared with those of the thoracic SCI (incomplete vs. complete: 56.13 vs. 18.72 in the Passive condition; 65.48 vs. 18.90 in the Active condition). Due to this intersubject variability, there was no significant difference in the relative magnitude of the EMG activity in the Sol muscle (averaged value) between the two groups (incomplete vs. complete: 137.3 ± 75.99 vs. 106.3 ± 19.90% in the Passive condition; 144.0 ± 87.52 vs. 98.1 ± 18.55% in the Active condition).
Typical examples of increase (subject T.T.) and decrease (subject O.O.) in the Sol EMG accompanied by the upper limb movements are, respectively, shown in the top and middle rows of Fig. 4. The Sol EMG activities in both Rest and Passive conditions are superimposed in Fig. 5 A to compare the differences between the two conditions. In subject T.T., the increase in the Sol EMG activity in the Passive condition was observed mainly during the backward-swing phase and the activity during the forward-swing phase was almost silent in both conditions (Fig. 5A, top). On the other hand, in subject O.O., the Sol EMG activity was present even during the forward-swing phase in the Rest condition, but it was considerably attenuated in the Passive condition (Fig. 5A, bottom). At the same time, the Sol EMG activity was also reduced in the backward-swing phase in this subject.
The differences in the Sol EMG activity among three conditions for all subjects are summarized in the bottom panels in Fig. 5B. For the backward-swing phase, the EMG magnitude tended to be larger in the Passive and Active conditions than that in the Rest condition. The result of ANOVA revealed a significant effect of the condition [F(2,13) = 3.777, P < 0.05], although a post hoc test did not identify any significant differences between each condition. For the forward-swing phase, no significant effect of the arm motion was identified by ANOVA [F(2,13) = 2.815, n.s.], but the result clearly demonstrated that three legs that exhibited a large EMG magnitude in the Rest condition (e.g., subject O.O.) showed remarkable reduction of the EMG activity in the Passive and Active conditions. With regard to the modulation index, the result of ANOVA indicated a significant effect of the arm motion [F(2,13) = 9.685, P < 0.01] and the post hoc test identified significant differences between Passive and Rest (P < 0.01) and Active and Rest conditions (P < 0.05). There was no significant difference between Passive and Active conditions.
Leg motions and load on foot sole
Figure 6 A shows the results of the hip, ankle, and shoulder joint kinematics. The bar graphs (left) depict the average range of motion of the joints among the three different experimental conditions, with corresponding waveforms showing the typical joint angles (right). As shown in these figures, both the hip (Rest: 19.3 ± 0.56°; Passive: 19.9 ± 0.59°; Active: 19.9 ± 0.54°) and ankle joint angles (Rest: 16.5 ± 2.45°; Passive: 16.1 ± 2.51°; Active: 16.2 ± 2.47°) moved in a similar manner among the three conditions. In the case of the shoulder joint angle, the angle was kept at a 10° extended position only in the rest condition. It was confirmed that the shoulder range of motion was comparable between passive and active arm motion conditions (Passive vs. Active: 35.4 ± 3.43 vs. 36.0 ± 3.29°).
Figure 6B shows the applied loads to the foot sole during movement. As shown in the right figure, the load was modulated almost sinusoidally with the leg movement cycle. The load was maximal and minimal, respectively, when the hip joint was maximally extended and flexed. The peak-to-peak load was similar among the three conditions (Rest: 489.5 ± 70.50 N; Passive: 490.1 ± 70.54 N; Active: 488.4 ± 73.59 N).
In the present study, we examined the effect of upper limb motion on locomotor-like muscle activity in patients with SCI. The results demonstrate that locomotor-like muscle activity was substantially affected by imposing simultaneous upper limb movements in cervical incomplete SCI subjects—that is, the upper limb movements generally increased the soleus EMG activity during the backward-swing phase corresponding to the stance phase in normal gait, whereas decreasing it during the forward-swing phase corresponded to the swing phase in normal gait. These changes were not observed in thoracic complete SCI subjects. These results indicate that the neural signal induced by the upper limb movements contributes to shape the appropriate locomotive motor output. In the following sections, the neuronal mechanism underlying the present results will be discussed, especially with respect to the contribution of the interlimb neural pathway.
Effect of upper limb motion on the lower limb locomotor-like muscle activity
Ferris and colleagues reported that the lower limb muscle activity during cyclic stepping with a recumbent stepping device was significantly increased for healthy subjects when the active arm movements were performed concurrently (Huang and Ferris 2004). As mentioned in the introduction, although their results support the view that interlimb coupling contributes to the enhancement of muscle activity during human locomotion, the effect of the descending command on the lower limb muscle activity was not ruled out in their study. The “externally driven arms and legs” condition that they investigated (see Figs. 3 and 5 in Huang and Ferris 2004) might have the potential to provide useful information for the discussion of the effect of descending motor commands; however, passive leg motion did not induce reliable phasic EMG activity in the lower limb muscles. This result implies that the EMG activity did not have a sufficient sensitivity to examine the effect of passive arm motion. To increase the sensitivity, they could have examined the effect of passive arm motion on the lower limb muscle activity during voluntary leg motion, but this type of experiment was not performed. It is therefore likely that the enhanced muscle activity was the result of descending commands concurrent with voluntary upper limb movements such as that seen when a Jendrassik maneuver is performed (Gassel and Diamantopoulos 1964). Moreover, Baldissera et al. (2002) reported that the H-reflex modulation of the upper limb muscles during rhythmic ankle movements is generated by corticospinal inputs rather than by segmental inputs. Thus it is necessary to exclude the influence of descending commands to examine the genuine contribution of the interlimb pathway to the locomotor activity.
In this study, we approached this issue by examining the effect of passive upper limb movement on the locomotor-like lower limb muscle activity of SCI subjects. Since the muscle activity is regarded to have originated from the spinal locomotor center (Dietz et al. 1995; Dimitrijvic et al. 1998; Kawashima et al. 2005a), we expected that neural input from the upper limb should affect the lower limb locomotor-like muscle activity through the interlimb neural pathway. As for the exclusion of the influence of descending neural commands, it would be better to do experiments with complete cervical SCI patients. However, it was not easy for them to conduct the experiment because of lower tolerance to the orthostatic hypotension and/or other difficulties associated with the maintenance of passive standing posture (hip joint contracture, shoulder subluxation, etc.). Despite this limitation, we considered that the effect of descending neural commands can be clarified by the comparison of the locomotor-like muscle activity between active and passive arm motion conditions because the extent of involvement of descending neural commands is far less in the passive than that in the active arm motion condition; that is, if passive upper limb motion, which is also an involuntary movement, affects the EMG activity, then the result may be attributed to upper-limb–induced neural inputs mediated by an interlimb pathway. The present results demonstrated that even when the upper limbs were moved passively (Passive condition), the locomotor-like EMG activity in lower limb muscles was substantially modified in the cervical incomplete SCI subjects. As shown in Fig. 5B, both the EMG magnitude and modulation index show no significant difference between Passive and Active conditions. These results indicate that it is not descending commands, but rather sensory neural signals induced by upper limb movements that play a substantial role in modifying the locomotor-like muscle activity through interlimb neural pathways.
The lack of change in muscle activities seen in the complete thoracic SCI subjects during passive arm movements also indicates that such EMG modulation did not arise from artifacts induced by the upper limb movements. Rather, the result strongly suggests that the EMG modulation was attributed to motion-induced neural signals from the arm via interlimb neuronal pathways because the spinal cord injury in these subjects prevented neuronal signals from reaching the spinal segments innervating lower limb muscles. Although all thoracic SCI subjects were judged as clinically motor complete, it is possible that some of the neural connections in the injured spinal cord were spared. However, it should be noted that irrespective of whether the injury was complete did not affect our conclusion: the results certainly show that the spinal cord injury located between the spinal segments innervating upper and lower limb muscles was sufficient to prevent neuronal signal transmission through interlimb neural pathways.
Contribution of upper limb movements on the lower limb locomotive motor output
As shown in Figs. 3 and 4, the changes in magnitude of the EMG activity with the upper limb movement were not consistent but varied from subject to subject; i.e., although more than half of the data demonstrated an enhancement of the EMG magnitude in the lower limb as a result of passive arm motion, three legs showed a remarkable reduction in the EMG magnitude (Fig. 3). However, we do not interpret the latter result as indicating a negative role of the upper limb motions in the locomotor-like activity. For the legs whose Sol EMG magnitude increased, this increase occurred mainly during the backward-swing phase (see Figs. 4 and 5). This enhancement of the EMG activity is consistent with the functional viewpoint because the Sol should be active during the stance phase in ordinary human locomotion (Winter 1991). On the other hand, for the legs whose Sol EMG magnitudes decreased, the reduction of the EMG activity by passive arm motion occurred mainly during the forward-swing phase (Fig. 5). Considering that Sol EMG activity during the forward-swing phase is inappropriate in normal human gait, the decreased EMG activity implies that the upper limb movements contributed to shape the locomotor-like muscle activity more properly by eliminating the inappropriate activity. In these subjects, the timing of the Sol EMG offset was dramatically shortened whereas the EMG onset was similar. It is likely that the sensory signal traveling from the upper limbs affects the timing of EMG termination. Concerning this point, previous animal studies demonstrated that sensory inputs from the forelimb contribute to the termination of hindlimb stance (Akay et al. 2006; Miller et al. 1975). Thus a similar explanation may be adopted for the decreased EMG in the forward-swing phase.
The legs that showed the reduction of muscle activity in the forward-swing phase were right side of O.O., right side of K.T., and left side of Y.H. Interestingly, these subjects have a laterality of their paralysis and the abnormal muscle activity was obtained from the side that has severe motor paralysis. These results suggest that the type of injury and the extent of the paralysis may affect the characteristics of the locomotor-like EMG activity. However, we have limited data on which to base a discussion of this issue. Further studies will be needed to clarify the neural mechanism underlying the abnormal muscle activity and its modulation in accordance with the arm motion.
Although the present results did not show remarkable differences between Passive and Active conditions, we do not interpret that descending motor commands play a minor role in the locomotive motor output. As mentioned in methods, both the upper and lower limb motions were induced by an experimenter during the Active condition so that the subjects conducted gentle arm motion. It is very likely that the lower limb muscle activities would be enhanced in the presence of additional descending commands. With regard to the effect of arm motion on lower limb muscle activity, Visintin and Barbeau (1994) reported that upper limb movement improves lower limb motor patterns in patients with incomplete SCI. Additionally, Behrman and Harkema (2000) mentioned that facilitating arm swing could be beneficial during locomotor training for individuals with spinal cord injury. Our finding can provide clear evidence that the enhanced locomotive motor output is not merely induced by descending motor commands, but certainly involves interlimb coordination.
In previous studies, animal preparations, such as decerebration and spinalization, have been used to examine the neural control of locomotor activity (Barbeau and Rossignol 1987; for review see Duysens and van de Crommert 1998; Pearson 2000). Since the structure and basic nature of the spinal cord are assumed to be the same in human and other mammalian species, the knowledge gained in the animal studies can, at least in part, be transferred to humans; for instance, some researchers pointed out that bipedal and quadrupedal locomotion share common spinal neural control mechanisms (Dietz 2002). On the other hand, unlike other species, humans have acquired an innate neural system for accomplishing bipedal locomotion and upright stance in association with a differentiation of hand movements. Considering such a phyletic evolution process, it is most likely that humans have a different neuronal control of locomotion. The question we addressed here was whether the interlimb neural pathway, which mediates the spinal segments of upper and lower limbs, plays a substantial role in the human locomotive movement. This was partly examined in previous electrophysiological studies (Balter and Zehr 2007; Dietz et al. 2001; Haridas and Zehr 2003; Zehr and Haridas 2003; Zehr and Kido 2001; Zehr et al. 2001), but direct evidence of the contribution of interlimb neural pathways to lower limb locomotor activity has not been provided.
Our hypothesis was that the motion-induced sensory signals from the arm should influence locomotor-like muscle activity through the interlimb neural pathway for the incomplete cervical SCI but not for the complete thoracic SCI subjects. This experimental design might be suitable for examining the genuine contribution of the interlimb neural pathway because the locomotor-like muscle activity is considered not to be contaminated by any descending neuronal input from supraspinal circuits. The results showed that the upper limb movement can properly modify the locomotor-like muscle activity only when neural connections between the spinal segments innervating upper and lower limb muscles are intact. The present results provide a new finding that sensory signals from the upper limb not only enhance, but also shape appropriate locomotive motor outputs. Taking these results together, it is conceivable that the interlimb neural pathway might be involved in the spinal neural circuits generating human bipedal locomotion, as is the case of quadrupedal locomotion of animals. Such interaction of upper and lower limb motions could be one of the neural mechanisms involved in human bipedal locomotion.
This work was supported by a grant from the Japanese Society for the Promotion of Science to N. Kawashima.
The authors thank J. Zariffa and F. Roy for valuable comments and corrections to this manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 by the American Physiological Society