Autism is a devastating developmental disorder that can leave children with profound behavioral problems. One of the barriers to understanding the disease is the lack of a valid animal model to test hypotheses for the causes of the disorder or to test new treatments.

Now researchers have created mice that exhibit some of the key behavioral and neurological deficits in the disease. They hope the mouse model of the disease will help them pinpoint specific brain areas linked to the social problems that characterize autism.

People with autism show several typical behavioral problems: poor social interaction, poor communication, and repetitive behaviors, such as rocking. But creating a mouse model for autism has been a challenge. To create animal models of diseases, scientists usually start with a gene that has been implicated in a disorder, then knock out that gene and look for symptoms of the disease in the animal.

While autism has a strong genetic component, "human studies haven't come up with an obvious gene that causes autism," says Richard Paylor, a behavioral geneticist at Baylor College of Medicine in Houston, TX (who was not involved in the research). "There have been lots of reasonable candidates, but no home run."

In a paper published today in the journal Neuron, researchers from the University of Texas Southwestern Medical Center in Dallas, TX, describe a new mouse model in which a gene involved in cell signaling, called PTEN, was knocked out in mature neurons. People with mutations in this gene sometimes have autism. And the PTEN mutant mice show profound deficits in social interactions. Normal mice, for example, are eager to investigate a new mouse in their cage, spending more time with a new visitor than with a familiar mouse or a new inanimate object. Mutant mice, however, spent less time in social interactions and were equally interested in inanimate objects and other mice.

The gene was knocked out only in a subset of neurons in the brain, which allowed researchers to examine exactly how that knockout affects the nerve cells. "We saw very aberrant neurons that resemble what is described in some instances of autism," says Luis Parada, a developmental biologist who led the study. The cells had more synapses than normal and made connections in abnormal places.

"I think this model will be useful for evaluating therapeutic interventions for some of the social problems found in autism," says Paylor.

It's not yet clear if the model will be relevant to the most common forms of autism or to a subset of unusual cases. "Whether this is a bona fide model for autism remains to be seen, but it's one of the most promising starts," says Andy Shih, director of research at the National Alliance for Autism Research in Princeton, NJ.