By illuminating key nerve cells that regulate feeding behavior, researchers at the Howard Hughes Medical Institute’s Janelia Farm Research Campus have discovered how to prod mice to eat voraciously—or sparingly—thanks to a radiant new technology.

The scientists created mice whose brain cells can be activated with a beam of light. Hitting one type of neuron drives the animals to their food bowls within minutes, whereas targeting a different type makes them abstain, the researchers found. The findings were published online January 5, 2011, in the journal Nature Neuroscience.

“This would be like a 200-pound adult losing 14 pounds in just one day, simply by activating a specific neuron population in their brain.”
Scott M. Sternson

"Eating and hunger and the motivations associated with food—they're all derived from the activity of neurons in the brain," says lead investigator Scott Sternson, a group leader at Janelia Farm. By unraveling the complex circuits involved in eating, he adds, researchers might better understand why some people eat too much or too little.

Compared with, say, a reflexive shiver, sneeze, or muscle twitch, feeding behaviors are quite complex. They are influenced by a variety of motivations and sensory stimuli—from food smells to body temperature—and require perceptual and motor skills. Unsurprisingly, dozens of brain regions and hundreds of cell types are involved in eating.

A type of nerve cell called agouti-related peptide (AGRP) neurons crop up in the hypothalamus, an evolutionarily ancient brain area that controls a slew of automatic body functions. AGRP cells are so-named because they secrete the AGRP protein. Previous studies showed that when injected into the brains of healthy mice, AGRP triggered the animals to start eating. In contrast, mice engineered to lack these neurons starve.

In the same brain area, pro-opiomelanocortin (POMC) neurons release the POMC protein. It has the opposite effect of AGRP: When injected into the brain, a peptide derived from POMC curbs eating. Mice lacking POMC are obese.

The downside of the older work on these cells is that the experiments could not mimic the natural amounts or precise distribution of AGRP and POMC in the brain. Sternson says his group used a different strategy to overcome those limitations. "We set out to directly test the capacity of these neurons to give rise to feeding behavior," he explains.

Sternson's group took advantage of a recently developed technique that lets researchers manipulate the activity of nerve cells with light. The technique, developed in part by HHMI Early Career Scientist Karl Deisseroth, has helped launch the new field of optogenetics for studying the brain. First, they used a virus to insert a light-sensitive protein called channelrhodopsin-2 (ChR2) into the cell membrane of either AGRP or POMC neurons. The researchers could then trigger the ChR2-expressing cells to fire by exposing them to blue light streaming into a surgically implanted shunt in the skull.

Because the virus doesn't infect every single ARGP cell or POMC cell, the researchers end up making mice in which different numbers of neurons are sensitive to the blue light. Sternson and colleagues found that the more AGRP neurons are stimulated, the more food the animals eat—even in the morning, when mice are typically resting. "The animal wakes up, rubs its face, stretches its legs, walks over to its food cup and all of a sudden starts eating with this remarkably voracious intensity—like it had been starved," Sternson says.

Over a one-hour stimulation period, the animals eat about 20 times more food than do controls. Their intake correlates with the frequency of the light pulses, and as soon as the light goes out, the mice stop eating.

In contrast, over one day of POMC neuron activation, mice lower their food intake by nearly 40 percent and lose 7 percent of their body weight. "This would be like a 200-pound adult losing 14 pounds in just one day, simply by activating a specific neuron population in their brain," Sternson says.

Researchers had hypothesized that AGRP neurons might control food intake by blocking the 'stop eating' signals of POMC neurons. The new study suggests otherwise. Sternson found that AGRP activation leads to overfeeding even in mice that carry genetic glitches that effectively shut off POMC pathways.

Sternson is currently investigating what lies downstream of the AGRP neurons to incite this gluttony. The culprit could be other chemicals produced by the cells, such as neuropeptide Y or the neurotransmitter GABA.

"These neurons' ability to fully orchestrate feeding behavior is just a remarkable thing to see," Sternson says. "It’s really quite unusual that a small population gives rise to such complex behaviors."

The findings raise provocative questions about possibly targeting these cells to spur weight loss in obese people or weight gain in people with anorexia. This sort of mind control is not so far-fetched: a method called deep-brain stimulation, in which surgeons insert metal electrodes into the brain of an awake person, has proven somewhat effective in treating depression and obsessive compulsive disorder, for example.

"Such an approach is conceivable, however it's quite invasive," Sternson says. "My hope is that a better understanding of the mechanism of eating behavior would facilitate the design of less invasive approaches, whether they are pharmacological or even behavioral."