Each time we decide to move, electrical signals cascade from the brain down the spinal cord to instruct our muscles to contract. Damage to the pathways that relay these messages, for example owing to a spinal-cord injury, can result in paralysis, for which there is currently no cure. The nerves of the spinal cord do not heal spontaneously after injury, and efforts to repair them with pharmacological and regenerative techniques have had only limited success. On page 284, Capogrosso et al.1 report that an alternative approach involving a wireless electronic connection between the brain and spinal cord restored movement in two monkeys that were each paralysed in one leg as a result of a spinal-cord injury.

The researchers used an implant in the brain called a neural interface, which decodes information from an array of electrodes that measure the activity of multiple brain cells that normally control leg movements (Fig. 1). They surgically implanted a second electrode array over the lumbar region in the lower part of each monkey's spinal cord, below the level of the injury — a partial lesion of the spinal cord that severed the nerve connections on one side of the cord. These electrodes delivered small electrical currents that generated movements in the paralysed limb. Stimulation was controlled in real time by the decoded brain signals, which were relayed to the lumbar array through a wireless link, bypassing the injured region of the spine. This artificial connection restored communication between the brain and the lumbar spinal cord, enabling the animals to walk again.

Figure 1: Communication bypass.
figure 1

Capogrosso et al.1 established a neural-interface system that enabled communication between the brain and spinal cord in monkeys subjected to a spinal-cord injury that paralysed one leg. An implant in the brain recorded neural activity related to leg movements, and transmitted this information to electrodes at the base of the spinal cord via a wireless link. These electrodes triggered neuronal impulses that generated movement in the leg muscle, allowing the monkeys to walk freely, despite their injury. (Adapted from Fig. 1 of ref. 1.)

Full size image

The idea of using electronic implants to bypass damaged neural pathways dates back to the 1970s (ref. 2), but the twenty-first century has seen remarkable progress in this field. An early-stage clinical trial of a neural-interface technology called BrainGate showed3,4 that brain signals recorded from humans who had been paralysed could facilitate control of a variety of devices, including computers and prosthetic limbs. Similar signals have also been used to control stimulation of the cervical region at the top of the spinal cord, to restore arm and hand movements in paralysed monkeys5. Moreover, lumbar spinal-cord stimulation has yielded promising results in human trials, in which participants with paraplegia have recovered some voluntary movement in their legs6. But at present, these human studies use only open-loop stimulation, which involves continuous trains of neural excitation — this strategy seems to boost the sensitivity of spinal locomotor circuits to weakened inputs from the brain that survived the injury.

The closed-loop stimulation used by Capogrosso et al., in which neural stimulation is controlled in real time by brain signals, could enable more-accurate control of movements and perhaps work even in cases in which all neural inputs to the spinal cord have been severed. Moreover, there is increasing evidence that closed-loop stimulation can drive neuroplasticity7, the mechanism by which the connections between two neurons strengthen if both are active at the same time. Neuroplasticity has a crucial role in rehabilitation following injuries that sever the spinal cord only partially. It is possible that, in such situations, stimulating the spinal cord when appropriate brain activity is detected could help to strengthen surviving motor pathways to promote lasting improvements in movement.

The pace at which neural interfaces are being translated from initial experiments in monkeys to human trials has been rapid. For example, just four years separated the first brain-controlled computer interfaces in monkeys8 and people3. Similarly, a 2012 paper described the first use by a paralysed woman of a brain-controlled robotic arm4 — a technology initially established in non-human primates only four years previously9. Earlier this year, the pattern was again repeated when voluntary hand-grasping was restored in a person with quadriplegia using brain-controlled muscle stimulation10, following monkey experiments11 in 2012. It is therefore not unreasonable to speculate that we could see the first clinical demonstrations of interfaces between the brain and spinal cord by the end of the decade, especially because the implanted components used by Capogrosso et al. have already been approved for human use. This speed of translation from monkeys to humans is particularly impressive, given the technological and surgical complexity of neural interfaces, and it speaks to the close anatomical and physiological similarities in the motor systems of all primates.

Despite this progress, the use of monkeys for neuroscience experiments continues to be questioned in the media, and animal-rights groups are making concerted efforts to ensure that restrictions on such work are tightened in both the United States and Europe. It is notable that, although Capogrosso and colleagues are based in Europe and their research conformed to the current regulations of the European Union, the experiments were conducted in China. Grégoire Courtine, the lead researcher on this study, has in the past described the challenges involved in performing such experiments abroad12, and other scientists might lack the time, energy or resources to pursue their research so far from home. There is thus a real danger that the development of treatments for debilitating neurological conditions will be delayed if high-quality, well-regulated research in monkeys cannot be performed in Europe and America owing to increasingly tight regulations. Equally, as more primate neuroscience moves to Asia, it will be important for researchers to remain committed to refining techniques and improving welfare standards for experimental animals worldwide.

The stakes are high. The World Health Organization estimates13 that, each year, between 250,000 and 500,000 people suffer a spinal-cord injury, resulting in disabilities that can persist for decades. There are still key technological challenges to overcome before neural interfaces can record robust and stable brain signals over these long time periods. Moreover, it remains to be seen whether a brain–spinal-cord interface can restore bipedal walking in humans after injuries that affect both legs, the most common injury in the clinic. Useful locomotion also requires control of balance, steering and obstacle avoidance, which were not addressed in the current paper. Nevertheless, Capogrosso and colleagues' study represents a major step towards restoring lost motor function using neural interfaces.Footnote 1