Controlling behavior, remotely

In the quest to understand how the brain turns sensory input into behavior, Harvard scientists have crossed a major threshold. Using precisely targeted lasers, researchers have been able to take over a tiny animal’s brain, instruct it to turn in any direction they wish, and even implant false sensory information, fooling the animal into thinking food was nearby.

As described in a Sept. 23 paper published in the journal Nature, a team made up of Sharad Ramanathan, an assistant professor of molecular and cellular biology and of applied physics; Askin Kocabas, a postdoctoral fellow in molecular and cellular biology; Ching-Han Shen, a graduate student in molecular and cellular biology; and Zengcai V. Guo, from the Howard Hughes Medical Institute, were able to take control of Caenorhabditis elegans — tiny, transparent worms — by manipulating neurons in the worms’ brain.

The work, Ramanathan said, is important because, by taking control of complex behaviors in a relatively simple animal — C. elegans have just 302 neurons ­— researchers can understand how its nervous system functions.

“If we can understand simple nervous systems to the point of completely controlling them, then it may be a possibility that we can gain a comprehensive understanding of more complex systems,” Ramanathan said. “This gives us a framework to think about neural circuits, how to manipulate them, which circuit to manipulate, and what activity patterns to produce in them.”

“Extremely important work in the literature has focused on ablating neurons, or studying mutants that affect neuronal function, and mapping out the connectivity of the entire nervous system,” he added. “Most of these approaches have discovered neurons necessary for specific behavior by destroying them. The question we were trying to answer was: Instead of breaking the system to understand it, can we essentially hijack the key neurons that are sufficient to control behavior and use these neurons to force the animal to do what we want?”

Before Ramanathan and his team could begin to answer that question, however, they needed to overcome a number of technical challenges.

Using genetic tools, researchers engineered worms whose neurons gave off fluorescent light, allowing them to be tracked during experiments. Researchers also altered genes in the worms that made neurons sensitive to light, meaning they could be activated with pulses of laser light.

The largest challenges, though, came in developing the hardware necessary to track the worms and target the correct neuron in a fraction of a second.

“The goal is to activate only one neuron,” Ramanathan explained. “That’s challenging because the animal is moving, and the neurons are densely packed near its head, so the challenge is to acquire an image of the animal, process that image, identify the neuron, track the animal, position your laser, and shoot the particular neuron — and do it all in 20 milliseconds, or about 50 times a second. The engineering challenges involved seemed insurmountable when we started. Askin Kocabas found ways to overcome these challenges.”

The system that researchers eventually developed uses a movable table to keep the crawling worm centered beneath a camera and laser. The researchers also custom-built computer hardware and software, Ramanathan said, to ensure that the system works at the split-second speeds needed.

The result, he said, was a system capable of controlling not only the worms’ behavior, but their senses as well. In one test described in the paper, researchers were able to use the system to trick a worm’s brain into believing food was nearby, causing it to make a beeline toward the imaginary meal.

Going forward, Ramanathan and his team plan to explore what other behaviors the system can control in C. elegans. Other efforts include designing new cameras and computer hardware, with the goal of speeding up the system from 20 milliseconds to one. The increased speed would allow them to test the system in more complex animals, like zebrafish.

“By manipulating the neural system of this animal, we can make it turn left, we can make it turn right, we can make it go in a loop, we can make it think there is food nearby,” Ramanathan said. “We want to understand the brain of this animal, which has only a few hundred neurons, completely, and essentially turn it into a video game, where we can control all of its behaviors.”

Funding for the research was provided by the Human Frontier Science Program, the National Institutes of Health (NIH) Pioneer Award, and the National Science Foundation.