Technical Comments

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

See allHide authors and affiliations

Science  19 Oct 2012:
Vol. 338, Issue 6105, pp. 328
DOI: 10.1126/science.1226274


Sławińska et al. questioned the involvement of supraspinal centers in restoring locomotion after multisystem neuroprosthetic training in rats with paralyzing spinal cord injury. Here, we clarify misconceptions and present additional results illustrating the robust influence of brain input on electrochemically enabled spinal circuitries. We reassert that our intervention reestablished supraspinal control over hindlimb locomotion in paralyzed rats.

A century ago, Sherrington reported the unexpected ability of isolated spinal circuits to generate rhythmic leg movements in response to sensory input (1). Since this seminal observation, spinal locomotion has been described in various species and conditions (2). We exploited this concept to design an electrochemical neuroprosthesis (3) and training procedures (4, 5) that restored full weight-bearing locomotion in rats with a complete spinal cord injury (SCI). Sławińska et al., as well as many others (6, 7), reported comparable observations in mice (8), rats (9), and cats (10) over the past three decades.

These hindlimb movements, however, remain involuntary. In spinal animals, quadrupedal hindlimb locomotion occurs in the complete absence of supraspinal input and is controlled via sensory ensembles (5, 11, 12). For example, the intact forelimbs initiate locomotion and sustain propulsion during quadrupedal gait, which generates sensory feedback in the hindlimbs due to biomechanical coupling. Like the rear person in a pantomime horse, lumbosacral circuitries can recognize these task-specific sensory states and use this information to coordinate complex movements. In a recent study (11), we demonstrated that, as early as 10 days after a paralyzing SCI, quadrupedally positioned rats produced weight-bearing locomotion, climbed a staircase, and steered in a curve during electrochemically enabled motor states. We did not report these results in our Science paper (13) because lumbosacral circuits control these seemingly complex quadrupedal movements without contribution from supraspinal centers. Therefore, they presented no novelty in the framework of this study. Indeed, rats with the same SCI were incapable of initiating bipedal walking [figure 1C and movie S1 in (13)]. They also failed to sustain robotically initiated locomotion [figure S6 in (13)]. Moreover, when the robot moved nontrained rats toward a staircase, the isolated spinal circuits failed to adjust gait movements, which led to a dramatic collapse against the staircase. Together, these results demonstrate that the absence of supraspinal input led to the inability to initiate, sustain, and modulate bipedal hindlimb locomotion overground. They also reveal that quadrupedal locomotor testing yields inconclusive results and that bipedal walking is a more appropriate paradigm to uncover the ability of supraspinal centers to functionally access lumbosacral circuits.

In (13), we described a type of motor control that is radically different from automated spinal stepping: that is, the recovery of supraspinal control over a range of locomotor behaviors (13). Rats with paralyzing lesions could initiate bipedal locomotion from a resting position on their own will, walk for extended periods of time, pass obstacles, and climb a staircase. To accomplish these tasks, the rats implemented feedforward strategies because they significantly increased their step height before negotiating the staircase [figure S6 in (13)]. In the majority of trials, the hindpaw did not come in contact with the staircase, which excludes the contribution of proprioceptive feedback for triggering these task-specific adjustments (see movie S1, staircase). Thus, active training under highly functional states promoted extensive and ubiquitous remodeling of axonal projections, which allowed the brain to regain adaptive control over electrochemically enabled lumbosacral circuits (13). This plasticity of descending systems did not occur in rats trained on a treadmill. Although these animals showed full weight-bearing stepping on a treadmill and coordinated quadrupedal overground locomotion (11), they failed to produce voluntary bipedal locomotion despite head, shoulder, trunk, and arm movements during their unsuccessful attempts to walk.

Undoubtedly, the biomechanical design of the rodent locomotor system and the remarkable ability of lumbosacral circuits to use sensory input as a source of control for locomotion extensively contributed to the produced movements (5, 13). In fact, the same type of neural operations most likely underlies the production of locomotion in healthy terrestrial mammals, including humans. Whether this neural process evidences voluntary or involuntary motor control is a problem of definition: What is voluntary motor control? Everyone will agree that trained rats initiated locomotion voluntarily, because they triggered their walk from a resting position at the view and/or smell of the reward. They thus have developed strategies to initiate walking and, in the case of climbing a staircase, even to anticipate the need for increased step height and enhanced locomotor output to resist gravity. We deployed a series of complex additional experiments to demonstrate that neural inputs arising from supralesional regions were required for the expression of locomotion overground and that cortical networks were critically involved in generating these movements. All these experiments emphasized the necessity of intraspinal and cortical circuitries for the production of bipedal hindlimb locomotion in trained rats.

To further document the supraspinal modulation of hindlimb locomotion in trained rats with paralyzing SCI, we performed two complementary experiments, which are illustrated in movie S1. In the first sequence, a paralyzed rat shows repetitive whole-limb extension, and even jumps, to reach a reward while continuously walking on the treadmill under electrochemical stimulations. When the reward is withdrawn, the rat willingly blocks hindlimb movements. How could these voluntary actions be triggered? Due to the constraining trunk support system, increase in weight-bearing input via tilting of the trunk cannot explain this modulation. Sławińska et al. (14) postulate that arm movements could elicit these hindlimb movements. Although forelimb movements do contribute to facilitating gait in rats, just like arm movements facilitate leg movements in humans (15), it is difficult to conceive that these inputs would be sufficient to elicit forces substantially larger than the body weight in order to jump. Moreover, Sławińska et al. recognized that forelimb movements were insufficient to initiate hindlimb locomotion after inactivation of the motor cortex. The second sequence shows that trained rats display continuous and coordinated hindlimb swimming despite the lack of useful weight-bearing input to trigger movements in water.

In the absence of experiments that would prove otherwise, we conclude that this series of multifaceted evidence demonstrates that the production of overground locomotor movements in trained rats resulted from powerful interactions between electrochemically enabled spinal circuits and supraspinally originating motor commands.

Finally, it is worth remembering the pioneer recovery of supraspinally mediated movements after several months of electrically enabled training in a paraplegic man (16). Despite three years of chronic paralysis, this individual regained supraspinal control over joint-specific leg movements during stimulation. These results are consistent with our conclusions and suggest that the plasticity of spared neuronal systems observed in rats may also occur in humans with severe SCI. We hope that our work will inspire new thinking to restore motor function for affected individuals in the future.

References and Notes

  1. Acknowledgments: A patent is pending: Apparatus and Method for Restoring Voluntary Control of Locomotion in Neuromotor Disorders (U.S. application no. 61/653,021).
View Abstract


Navigate This Article