Technical Comments

Comment on “Restoring Voluntary Control of Locomotion After Paralyzing Spinal Cord Injury”

Science  19 Oct 2012:
Vol. 338, Issue 6105, pp. 328
DOI: 10.1126/science.1226082


Van den Brand et al. (Reports, 1 June 2012, p. 1182) claim to have restored voluntary control of locomotion after paralyzing spinal cord injury. They have not considered recent findings that their upright posture paradigm contributes to locomotor capability after such injuries. We propose that postural adjustments that activate the locomotor central pattern generator in the upright posture, rather than direct voluntary control of locomotion, account for their results.

Van den Brand et al. claimed to have restored voluntary control of hindlimb locomotion after paralyzing spinal cord injury using an electrochemical neuroprosthesis combined with overground locomotor training in adult rats (1). They used epidural electrical stimulation together with systemically applied drugs and bipedal locomotor training in a robotic postural interface to force rats to walk toward a food reward. Given the impact of such claims for paraplegic patients around the world, it is important to critically examine the evidence provided. In a recent publication, Sławińska and others (2) clearly showed that the upright bipedal posture alone provides sensory feedback that promotes coordinated hindlimb locomotion in the absence of training, drug application, or supraspinal influence. We propose that the bipedal training procedure used by van den Brand et al. (1) initiated locomotion due to forward shifts in posture, thus engaging powerful feedback from load receptors of the hindlimbs that potently facilitates locomotor activity (2, 3). The videos [supplementary materials for (1)] show that the rats move their forelimbs, head, and trunk, thus shifting the center of mass forward. The net effect is an increased loading of the hindlimbs that per se could lead to spinal stepping, especially in the upright posture. Injections of potent excitatory drugs combined with lumbosacral electrical stimulation, as used in their paradigm, likely increases the responsiveness of the spinal locomotor central pattern generator (CPG) to load-bearing inputs. As their figure 4 shows, the onset of overground locomotion is accompanied by an increase in the ground reaction force (GRF), thus facilitating locomotor activity without having to engage direct voluntary control of the locomotor CPG. We propose that the training paradigm resulted in a strategy that shifted the center of mass forward so that the afferent feedback associated with the upright posture, together with the resultant hip extension, engaged the spinal CPG for locomotion and propelled the animal toward the food reward. This interpretation does not require that the brain regain “supraspinal control over the electrochemically enabled lumbosacral circuits,” as the authors propose [supplementary materials for (1)].

If the training is effective in restoring voluntary locomotion, and this is not because the animals are in the upright posture as we propose, then the rats should display marked recovery of overground quadrupedal locomotion, the normal position for progression in rodents. However, a description of the animals in a quadrupedal locomotion task was not included in this paper. There are reports (46) that recovery of quadrupedal locomotion occurs spontaneously in adult rats and mice with staggered hemisection injuries similar to that used by van den Brand et al. The concept that propriospinal neurons can relay a locomotor message is derived from experiments in Schmidt’s laboratory, where it was demonstrated that neonatal rats subjected to simultaneous staggered hemisections can recover locomotion (79). They have recently demonstrated (6) spontaneous recovery of hindlimb locomotion in adult rats subjected to simultaneous staggered hemisections, such as those described in the van den Brand paper (1). Barrière et al. (10, 11) showed that a previous thoracic hemisection facilitates recovery of hindlimb locomotion after a subsequent total transection in cats, and they have confirmed this observation in adult rats (12). Therefore, it seems likely that spontaneous recovery of propriospinal activity plays a role that was not revealed by the analysis used in the van den Brand et al. study. The N-methyl-d-aspartate (NMDA) lesions (1) of the thoracic region between the two hemisections may be effective in stopping locomotion by removing spontaneous propriospinal plasticity rather than by blocking newly developed voluntary control of the locomotor system.

Forelimb movement observed in these experiments likely contributed substantially to propriospinal signals that help drive locomotor activity in the hindlimbs (13). Videos in (1) suggest that the rats learned to flail their forelimbs as part of the strategy to attain forward progression to the food reward. Thus, the results with NMDA injections into the thoracic relay site can be explained as an interruption of the relay of propriospinal messages for control of locomotion from the forelimbs.

The results with muscimol injected into the motor cortex (1) are also consistent with our suggestion that the animals developed a new strategy to facilitate postural adjustments required to initiate locomotion, rather than establishing new connections with the lumbosacral locomotor centers. Indeed, after injecting muscimol, the forward shift of the trunk does not occur [movie 3 in (1)]. Moreover, van den Brand used obstacle avoidance and stair climbing to support their claim of voluntary control of hindlimb locomotion. It must be pointed out that stairs and obstacles do not actually constitute “two conditions requiring voluntarily mediated gait tuning” as stated in the van den Brand et al. paper. The review by Drew et al. (14) cited in support of this claim addresses only visually guided movement, but visual guidance does not appear to be involved in the cases reported here. It is well known that spinal-transected animals can negotiate obstacles after contact, as is the case in the videos accompanying this paper. Thus, the conclusion that these tasks require voluntary control is not warranted, and the data provided do not establish voluntary control as the basis for the rats’ success with these tasks.

Recordings of cortical activity provided in (1) are more supportive of a role for the cortex in controlling postural adjustments than hindlimb locomotor activity, because the neuronal activity recorded from the cortex is associated with an increase in GRF that occurs at the onset of hindlimb locomotion and not with the rhythmic activity during locomotion (their figure 4). It would be of interest for the authors to attempt to correlate the recorded cortical activity with the GRF changes or other signals from the hindlimbs to examine this possibility. They show that postural adjustments occur in the overground-trained group (PC2 in their figure S5), which “exhibited enhanced lateral body movements that alternatively loaded the left and right limbs during locomotion.” This observation is consistent with our interpretation that the development of new strategies to maximize the impact of the sensory feedback associated with the upright posture accounts for the recovery of hindlimb locomotion.

Thus, the conclusion is inescapable that the data presented by van den Brand et al. do not provide convincing evidence for “restoring voluntary control of locomotion after paralyzing spinal cord injury.” Furthermore, the interpretation of experiments in which bipedal locomotion is used to monitor and/or train the recovery of locomotion after injury must include consideration of the direct physiological and biomechanical consequences of placing rats in an upright posture.

References and Notes

  1. Acknowledgments: We thank T. Drew for valuable input regarding the interpretation of the cortical recordings reported in the van den Brand et al. (1) paper.
View Abstract


Navigate This Article