Abstract
Researchers have created a map of 14,000 human protein–protein interactions, the largest of its kind to date. The work confirms that cancer-associated proteins tend to link to each other and serve as hubs of protein interactions, and shows that researchers can use the maps to uncover new cancer-associated proteins.
The largest and most comprehensive map of protein interactions to date shows that cancer-associated proteins connect to each other and suggests how researchers might discover new cancer-promoting genes.
Interacting proteins can work together to foster the growth and spread of tumor cells, but detecting these molecular relationships is laborious. Even if each protein comes in only one form, researchers would have to test 200 million combinations to determine the pairwise interactions for all of the 20,000 proteins coded for by human genes. Past studies have evaluated only a fraction of these proteins and often relied on limited evidence to infer connections.
Marc Vidal, PhD, of Harvard Medical School in Boston, MA, Frederick Roth, PhD, of Mount Sinai Hospital in Toronto, Canada, and colleagues systematically measured the interactions among more than half of all human proteins, a larger number than in any prior study. The researchers started by performing yeast two-hybrid screens, the standard technique for identifying interacting proteins, on all possible pairs of proteins in this group. They confirmed positive results with three additional binary interaction assays, eventually discovering roughly 14,000 interactions.
The scientists then used their map of these interactions to show that cancer-associated proteins mainly interact with each other and are well connected, serving as the focal points of networks. Previous studies had suggested both phenomena, but their conclusions could reflect researchers' tendency to focus on interactions among popular proteins.
“We found that indeed cancer-associated proteins are hubs, and that indeed cancer-associated proteins tend to stick to one another,” says Roth.
That result means that researchers can use interaction maps to find new cancer-promoting proteins that are linked to known ones. As Vidal, Roth, and colleagues reported last month in Cell, they evaluated several potential cancer-associated proteins based on whether their close neighbors in the interaction map were on lists of confirmed cancer-promoting proteins. One protein that stood out in their analysis was STAT3, a transcription factor that helps control cell division and apoptosis. Since the researchers performed their analysis, the STAT3 gene has been added to the Cancer Gene Census, an authoritative list of cancer genes.
Interaction maps could also enable researchers to pinpoint candidate cancer genes suggested by genome-wide association studies and to determine whether specific mutations are likely to promote cancer by testing whether they disrupt particular interactions, the researchers say.