Clues to Autism Emerge in Protein Network

Autism is as puzzling for scientists as it is heartbreaking for parents. Some patients function well despite a few behavioral quirks, whereas others are profoundly disabled. The dozens of "suspect genes" are scattered among various types of the disorder and show up in only a handful of patients. Now, by shifting focus from the genes to the proteins they produce, researchers have identified a densely connected network that may help reveal how autism develops. The finding may also lay the framework for developing new treatments, even for very different types of the disorder.

The inspiration to look for a protein network came from the research of geneticist Marc Vidal of the Dana-Farber Cancer Institute in Brookline, Massachusetts. Vidal had mapped the first of these networks—later known as interactomes—in the wake of the Human Genome Project. "The Human Genome Project gave us a list of parts," Vidal explains. "By studying how the proteins interacted, we could see how the parts were assembled."

Proteins working together inside cells sometimes physically touch each other; often, many of them will also link to a few central proteins that play a key role in a particular biological process. A diagram of the resulting relationships can look like a Koosh ball of interconnecting lines radiating from one or more hubs (see picture).

In a 2006 study published in Cell, Vidal and Huda Zoghbi, a neurobiologist at Baylor College of Medicine in Houston, Texas, teamed up to show previously unsuspected interactions among the proteins produced by mutated genes in several of the movement disorders known as ataxias. The fact that the "protein products" of seemingly unrelated genes actually work together hints at a common process through which the different types of ataxias unfold.

In the new study, Zoghbi and colleagues looked for protein interactions in the more complex disorder of autism. The researchers used "bait" proteins from over two dozen known autism genes, fishing in a pool of human DNA for other proteins that would interact with the baits. If the proteins attach to each other through repeated tests, it's a sign that they cooperate in some biological process and aren't just getting stuck together.

Using several sensitive screening methods, the researchers ultimately "caught" over 500 proteins that formed connections with 26 of the bait proteins as well as interacted with each other.

The fact that so many proteins interact with the 26 bait proteins indicates that these original proteins play key roles in a complex process that wasn't apparent on the level of the "suspect" genes. Since these genes are drawn from different types of autism, the process hinted at what may be a common feature.

The emerging pathway is likely a problem that occurs at the synapse, the point of contact between neurons, Zoghbi says. She has long argued that autism and other neurodevelopmental disorders may result from faulty connections at these junctions, but the possibility of a common thread—evidenced by the complex interactions among the autism-related proteins—is encouraging, Zoghbi says. Her team reports its findings online today in Science Translational Medicine.

Two proteins called SHANK and TSC1, which are involved in very different autism-related syndromes that are not thought to be related, proved to be connected by 21 other proteins. Additionally, when the researchers checked their network against the DNA of patients with nonsyndromic, or "stand-alone," autism, they found abnormalities involving three of the network genes. Both findings suggest that different types of autism may share a common pathway even when they occur in distinct syndromes or alone—something that wasn't clear just from looking at the genes.

These common pathways are hopeful targets for drug development, Zoghbi says. "Our interactome is only a first step, but it could lead to a framework to investigate new genes and test new drugs."

"Interactomes like this one make the whole debate of genes versus the environment a lot more sophisticated," adds Vidal, who was not involved in the current study and who mapped a network involved in breast cancer susceptibility in 2007. "In understanding how genes lead to disease, interactomes give us the best of both worlds."