Cut the motor nerve to a muscle and the muscle dies. Not immediately, and not all at once, but the atrophy is relentless; without the signal from the brain telling it to contract, the tissue simply wastes away. This is roughly what happens to the bladder after a spinal cord injury, or to the intestines in some forms of Crohn’s disease, and it’s the reason why restoring function to paralyzed organs is so much harder than restoring it to paralyzed limbs. The organ doesn’t just stop moving. It forgets how.
What a team at MIT has done, in work published this week in Nature Communications, is find a way around that forgetting. Not by building a machine to replace the muscle, and not by growing new tissue in the lab, but by rewiring existing muscle so that it takes its orders from a computer instead of the brain.
The result is something called a myoneural actuator, or MNA. It is, in principle, a living motor: a piece of your own skeletal muscle, reprogrammed at the level of its nerve supply, capable of squeezing a paralyzed intestine back into peristaltic rhythm or stretching a residual limb tendon to restore a sense of proprioception to an amputee. The team has demonstrated both, in a rodent model, and the implications are strange enough that it takes a moment to absorb what is actually being proposed. “We engineered existing muscles to become an actuator, or motor, that reinstates motion in organs,” says Hyungeun Song, one of the study’s co-leads.
It is a piece of skeletal muscle that has been rewired at the level of its nerve supply. Researchers remove the motor nerve that normally connects the muscle to the brain, then reroute a sensory nerve into the muscle in its place. The muscle can then be controlled by electrical stimulation from a computer, bypassing the brain’s voluntary control entirely. Because it uses the patient’s own tissue, there is no risk of immune rejection.
Motor nerves connect directly to the brain, meaning a muscle driven by motor nerve stimulation would still be under partial voluntary control. Sensory nerves carry signals toward the brain, not away from it, making them easier to commandeer for computer control without the brain interfering. There is also a practical advantage: sensory nerve axons are more uniform in diameter than motor nerve axons, which means electrical stimulation recruits muscle fibers more evenly and the muscle fatigues much more slowly under continuous use.
The MIT team demonstrated it on the small intestine and on a residual limb muscle designed to mimic amputation. In principle the system could be applied to the bladder in spinal cord injury, the gut in Crohn’s disease or diabetic enteropathy, the diaphragm in respiratory insufficiency, or the heart as a mechanical assist device. The researchers also suggest it could restore tactile feedback for prosthesis users and potentially allow people to feel sensations in virtual reality environments.
Considerably further than the enthusiasm in press releases tends to suggest. The current results are in rodents, and the path to humans requires larger animal trials, long-term safety data, and regulatory clearance. The surgical technique draws on procedures already used in nerve reconstruction, which helps, but the closed-loop control system also needs to be miniaturized and made wireless before clinical deployment is realistic. The researchers describe it as a long-term prospect rather than an imminent one.
The key insight is a distinction that most people outside neuroscience would never think to make: the difference between motor neurons and sensory neurons. Motor neurons run from the brain down to the muscles and carry the commands that make muscles contract; sensory neurons run in the opposite direction, carrying information about touch, pain, temperature, and body position back up toward the brain. Sensory neurons, in other words, are receivers, not transmitters of motor commands. But the MIT team noticed something useful about that asymmetry. “Sensory neurons, however, are wired to receive, not to command,” Song explains. “We thought we could leverage this dynamic and reroute motor signals through sensory fibers, making a computer, rather than the brain, the muscle’s new command center.”
The engineering challenge was substantial. To make the MNA work, the researchers had to first cut the existing motor nerve supply to a piece of skeletal muscle, which would normally doom the muscle to atrophy. Then they rerouted a sensory nerve into the same muscle, coaxing it to form functional connections with the muscle fibers. Whether sensory nerves could even form proper synapses with skeletal muscle was, as Song’s colleague Guillermo Herrera-Arcos acknowledged, genuinely unknown. “You don’t want the brain to consciously control the muscle actuator because you want the actuator to automatically control an organ, like the heart,” Herrera-Arcos explains. The brain had to be taken out of the loop entirely, replaced by a computer sending electrical stimulation pulses through the sensory nerve instead.
In rodents, it worked. The sensory nerves grew into the muscle, established chemical synapses at the neuromuscular junction (something previously undemonstrated in the literature), and the muscle retained its mass and force output. More than that, the reinnervated muscle actually became harder to fatigue than it had been before, which addresses one of the central failure modes of functional electrical stimulation in existing bionic systems. Under continuous stimulation, the MNA maintained meaningful force output for around 260% longer than native muscle. The reason comes down to fiber diameter. Motor nerves contain axons of highly variable sizes, and when you electrically stimulate them, the biggest axons fire first, burning through the fast-twitch fibers that fatigue most quickly. Sensory nerve axons are much more uniform in size, so the recruitment across muscle fibers is more even. The muscle doesn’t exhaust its reserves in a single burst.
The team demonstrated two applications. First, they wrapped an MNA around a segment of small intestine in a rodent and showed it could generate the squeezing motion that normally propels food through the gut. Second, they designed what they call a Proprioceptive Mechanoneural Interface, a biohybrid neuroprosthetic setup in which an MNA serially coupled to a residual limb muscle could stretch that muscle in precise, computer-controlled increments, triggering the sensory receptors inside and sending proprioceptive signals back toward the brain. The signal wasn’t synthetic or simulated. It was real afferent feedback, generated by a real muscle, conveyed through a real nerve.
That bidirectionality matters. Earlier approaches to bionic integration tended to focus on either efferent control (getting signals out to a prosthetic device) or afferent feedback (getting signals back from one), but not both simultaneously and through the same biological substrate. Song suggests the system could eventually allow a paralyzed stomach to relay hunger signals, or allow skin grafts to transmit tactile information that prosthesis users currently lack entirely. There is also, somewhat unexpectedly, a virtual reality application: the same MNA-to-skin coupling could in principle let a person feel what their avatar is touching in a digital environment.
None of this is close to a clinic. The rodent model is a proof of concept, and the path to human use involves larger animal trials, regulatory approval, and the kind of longitudinal safety data that takes years to accumulate. Implanting an MNA would also require the sort of nerve reconstruction surgery that is already performed in brachial plexus repair and facial nerve procedures, which perhaps makes it less exotic than it sounds, but is still surgery. The closed-loop control system also needs to become wireless and miniaturized before any of this is wearable rather than lab-bench-scale.
Still, the architecture is already quite different from anything currently implanted in patients. Pacemakers are titanium boxes with wires. Cochlear implants are electrodes threaded through bone. The MNA, if it ever reaches humans, would be part of the patient’s own body, grown from their own tissue, sustained by their own blood supply, and incapable of triggering immune rejection because it would be them. “Today’s solutions are mostly synthetic: pacemakers and other mechanical assist devices,” says Herrera-Arcos. “A living muscle actuator implanted alongside a weakened organ would be part of the body itself. That is a category of medicine different from anything seen in clinic.”
What that category eventually looks like is still anyone’s guess. The peripheral nervous system, it turns out, is considerably more plastic than anyone suspected: sensory nerves building cholinergic synapses with skeletal muscle, embryonic mechanisms apparently re-activated by the reinnervation process, muscle fibers responding to a nerve type they were never originally designed to receive. The biology is doing things the textbooks said it couldn’t. That tends to be where medicine gets interesting.