HARVARD GAZETTE ARCHIVES

Marc Vidal checks the data he and his colleagues collected on protein interactions. The team found 424 interactions that included at least one protein involved in cancer and other diseases. (Staff photo Kris Snibbe/Harvard News Office)

Of course, there's a big problem to address first, or someone would have done this already. Experts estimate that humans have about 100,000 proteins in each of their cells. In the daily course of living, many of those proteins interact with each other. So researchers faced the gargantuan task of cataloging an incredible number of interactions - millions, if not billions.

"It's a terrific challenge, but also a terrific opportunity," says Marc Vidal, who heads a team that has begun to put together a Who's Who in human proteins at the Harvard Medical School and the Center for Cancer Systems Biology at Dana-Farber Cancer Institute in Boston. The team has looked at 66 million combinations between 8,100 proteins and identified 2,800 interactions among them. The search revealed more than 300 new connections to more than 100 proteins associated with human disease.

"It's a start," says Vidal, an associate professor of genetics at Harvard Medical School. "We could not have done this even five years ago. Now, because of the Human Genome Project, we have a first-draft list of our genes, which carry the blueprints for all proteins." Protein numbers, about 100,000, exceed those of genes, about 25,000, because some genes make more than one protein.

"Protein interactions are cumbersome to detect directly in a cell, so we work through their genes," Vidal explains. "That way, we produced 8,100 proteins whose interactions we tested. These represent only a few percent of all the interactions that occur in human cells. But we have built a scaffold to which other interactions can now be attached."

Tracking protein traffic

Since Vidal and most of his group work in cancer biology, they have a prime interest in finding the causes and new treatments for these malignancies. "Of course, things are a lot more complex than one gene and its protein gone bad," Vidal cautions. "We need to know how mutated genes change the proteins and how all the proteins involved interact with each other. We now have a means for doing that in humans." He calls the system an "interactome," in contrast to "genome," which covers only the mapping of genes.

Up until now, Vidal notes, researchers have been studying protein interactions one at a time. How, for instance, is one reaction involved in the complex process of metabolism. Now, he believes, they have a tool to view the entire network of protein interactions involved in biological processes ranging from digestion to thinking.

Vidal uses the analogy of driving to work or home in busy traffic. Often all you can see is the lines of cars ahead. Another lane seems to be moving faster, so you switch. Immediately, traffic in that lane slows by an incredible factor.

What if there are other ways to get to work or go home, and you could compare them before you leave? You would know all the streets, lights, intersections, and density of traffic at a particular commuting hour, so you would stand a good chance of picking a better route. To switch from cars to cancer, Vidal says, "Looking at one or two genes, as we do now, is like looking at only one intersection or one stretch of a highway. In about ten years, we should be in a position of seeing a whole traffic or protein network at one time. We would be in a better position to predict when a disease might occur and how to treat it."

Not rocket science

Asked how his team built the first interactome, Vidal admits, "It's not rocket science, to tell the truth." The researchers started with the 8,100 proteins that can be manipulated using the current draft of the human genome. They then tested whether these proteins can interact in all possible paired combinations. The team has published the results of this work in the Sept. 28 online edition of the science journal Nature.

"We found 424 interactions that included at least one protein known to be involved in cancer and other diseases," Vidal notes. "Such proteins will be the target of further research into how those mutations lead to disease and how they might be inhibited by drugs or other treatments."

"Now that we have proof that this system works, we can ramp up our production line substantially," Vidal says, that is, look at more interactions more quickly. In the near future, the researchers look forward to publishing expanded editions of Who's Who in Proteins

"In the next five to ten years, we'll have worked out a good chunk of the protein interactions involved in cancers," Vidal promises. "That is likely to change drastically how well we predict how and when diseases occur."