Three drugs will be tested in humans to treat rare, inherited conditions that are often linked to autism: Rett syndrome, fragile X, and tuberous sclerosis complex (TSC). Scientists hope that the new drugs, if successful in the current trials, will eventually help treat more common forms of autism, which affects about 1 in 166 children in the United States. Existing drugs are used to treat symptoms of autism, such as digestive problems and psychosis, rather than the root of the disease.

"We may have our finger on a biochemical pathway that is applicable more generally in autism," said Mark Bear, a neuroscientist at MIT, at the Autism Consortium annual symposium in Boston last month.

One of the most puzzling problems with autism is that the brains of affected children look normal, making it difficult to know where to target new treatments. The discovery in the 1990s of the mutations underlying the three disorders has allowed scientists to create animal models with the same genetic mistakes. These animal models have enabled the search for more subtle molecular processes gone awry in each disease, such as abnormal gene or protein expression that changes the electrical properties of neurons, or the architecture of the synapse--that is, the connections between neurons.

In the past year, several groups have published novel treatments that appear to reverse the damage done in these diseases. "We had thought that in disorders like autism and fragile X, damage was done early, and the best we could do was stop it," said Story Landis, director of the National Institute for Neurological Disorders and Stroke, at the Society for Neuroscience conference, in Washington, DC, last month. But these studies show that you can intervene early and perhaps restore cognitive function, she said. Findings from these studies are now being tested in humans.

People with fragile X, the most common form of heritable mental retardation and a leading cause of autism, have a mutation in the FMRP gene, which normally inhibits protein synthesis stimulated by a receptor called metabotropic glutamate receptor 5, or mGluR5.

Last year, Bear and Gul Dolen, also at MIT, announced that they could correct abnormal brain development and faulty memory and reduce seizures in affected mice by decreasing mGluR5 activity by 50 percent. "The idea that you could reintroduce function is a sea-change event," said Emanuel DiCicco-Bloom, a neuroscientist and physician at the University of Medicine and Dentistry of New Jersey, at the neuroscience conference.

Experimental drugs that target the receptor are already under development, although none have yet been approved by the Food and Drug Administration. Human trials of one such drug is now under way, sponsored by Seaside Therapeutics, a company founded by Bear.