People with obsessive-compulsive disorder (OCD) repeat ritualistic actions such as hand washing or checking locks so frequently that it interferes with their everyday lives. But although the symptoms of the disease are obvious, its causes are not. By knocking out one gene, neurobiologist Guoping Feng at Duke University Medical School in Durham, North Carolina, and his colleagues have engineered mice that show OCD-like behaviours.

The work began with Feng's interest in the signal-receiving, or postsynaptic, ends of neurons. “The postsynaptic complex has hundreds of proteins,” he says, “But why are so many needed for an apparatus that is less than a micrometre across?” The researchers engineered mice lacking the gene for one of these proteins, the scaffolding protein SAPAP3, and waited for signs that the mice were different.

“We predicted that they might have severe, abnormal behaviour, but we didn't know what would come up,” says Feng. Patches of bald, irritated skin appeared around the eyes, snouts and necks of the mice. The mutant mice literally rub their own fur and skin away with compulsive grooming. “They can't stop grooming themselves,” says Feng. Videotapes showed the mice grooming themselves when normal mice would be sleeping.

Characterizing these mice took several laboratories six years, says Feng. William Wetsel's team, also at Duke University, looked at behaviour, and found that the mutant mice showed high anxiety (spending more time along walls or in dark areas of experimental chambers, for instance). In such mice, both anxiety and grooming were reduced by antidepressants. A third Duke-based team, led by Nicole Calakos, found that the altered synapses were less able to transmit signals. And Richard Weinberger's laboratory at the University of North Carolina, Chapel Hill, found that mice lacking SAPAP3 had morphological defects in the postsynaptic complexes in the striatal region of the brain.

But to pin the observed symptoms to defects in the striatum, the researchers needed to show that restoring SAPAP3 function in the striatum alone could relieve the OCD-like behaviour. Feng's team injected engineered lentiviruses with working versions of the gene into the striatum of 7-day-old mice. Six months later, five of the eight mice injected with SAPAP3 had no skin lesions from excessive grooming, whereas all of the control mice had lesions. Treated mice showed less anxiety, and their neurons received and transmitted signals more effectively (see page 894).

Most studies of OCD in humans have focused on neurons that either produce or respond to dopamine or serotonin. Feng's work, however, implicates neurons that respond to glutamate.

Once destined to become a physician in his native China, Feng now finds himself back addressing the needs of patients. His team is collaborating with clinicians to identify mutations in SAPAP3 and similar proteins in human families with a history of OCD. The findings also have implications for drug discovery, which so far has not focused on glutamate synapses. Feng says drug companies are interested in using the OCD-like mice for drug screening. Small molecules are unlikely to affect SAPAP3 itself, Feng says, but proteins that function downstream of postsynaptic signalling in the striatum might make suitable drug targets.