MATERNAL MEDIATION

Environmental agents under scrutiny in autism research include drugs, vaccines, viruses, and poisonous substances such as lead and mercury. "There certainly isn't a shortage of environmental suspects that may play a role in autism," notes Andy Shih, chief science officer of the National Alliance for Autism Research in Princeton, N.J. "But these are not all necessarily artificial or manmade and may have to do with influences in the womb."

Zimmerman, a pediatric neurologist, is one of many specialists who think that environmental influences in utero may contribute to autism by disrupting normal early development. To a fetus, any effect from outside the womb-hormones triggered by a mother's stress, for example-is environmental. "It's anything that affects pregnancy," says Zimmerman.

Isaac Pessah, director of the Center for Children's Environmental Health and Disease Prevention Research at the University of California, Davis, agrees. He also points out that newborns and infants are especially vulnerable to the damaging effects of toxic exposures because the human nervous and immune systems "undergo considerable remodeling in the first 2 years of life."

Some scientists suspect that maternal viral infections are one of the principal noninherited causes of autism. Epidemiological studies have found a significantly increased risk of autism in the offspring of mothers exposed to the rubella, or German measles, virus early in pregnancy.

In the Jan. 1, 2003, Journal of Neuroscience, scientists led by Paul H. Patterson of the California Institute of Technology in Pasadena reported that when pregnant mice were infected with a modified human-flu virus, they produced offspring that, as adults, behaved in ways similar to those of many autistic children. Compared with a control group, the affected mice interacted less and were unusually anxious under mildly stressful situations and around unfamiliar objects.

The scientists also found unusually low numbers of critical signaling components, called Purkinje cells, in brain tissue of the affected mice. Autopsies of people with autism have revealed fewer than normal of these cells.

In an upcoming International Journal of Developmental Neuroscience, Patterson's group reports that altered brain development in the mice doesn't appear to occur as a direct result of viral infection in the fetus. Instead, "there's evidence it's related to a natural immune response in the mother, [but the] mechanism is something we're still working on," says Patterson.

Some of the molecules that the mother uses to fight the virus may be crossing the placenta and affecting brain development in the fetus, he explains.

If so, the problem wouldn't be specific to the flu virus. "Lots of kinds of infection could lead to the same effects," Patterson says.