Researchers produce largest scale map of human protein interactions

Scientists will be better able to trace how genetic changes give rise to diseases ranging from cancer to Huntington’s disease with a new map of protein-protein interactions within human cells produced by researchers at the Center for Cancer Systems Biology (CCSB) at Dana-Farber Cancer Institute and associates around the world.

The expanded map, published online today by the journal Cell, is approximately 30 percent larger than the combination of all small-scale studies published in the scientific literature, and suggests that scientists may have been too narrowly focused in tracking the mechanisms responsible for disease. Whereas interaction maps have traditionally centered on genes and proteins known to be involved in specific diseases or important processes, the new map reveals that potential connections to disease exist beyond the confines of earlier studies.

“The search for genetic connections to disease has been compared to an individual looking for his car keys beneath a streetlight at night,” said Thomas Rolland, PhD, a co-first author of the study and researcher at Dana-Farber. “When a police officer asks why he’s looking there, he explains that that’s where the light is. By analogy, the new map casts a wider beam over the area to be searched, where the car keys would lie.”

Since the completion of a reference map of the human genome in 2004, scientists have associated more than 100,000 genetic changes to a wide array of human diseases. But genes form, in a sense, only the backdrop of cell life. Cells transact their business – both internally and externally – primarily through proteins, which are assembled according to instructions stored in the genes. It is the interactions between proteins, occurring constantly, that sustain cells and control their behavior.

“Even though the genetic mutations that underlie many human diseases are known, it’s often unclear how a particular mutation affects a gene or associated protein in such a way as to produce disease,” said Marc Vidal, PhD, director of the CCSB and the study’s co-senior author. “To make those kinds of connections, we need to track which proteins interact, and how mutations disrupt those interactions.”

Mapping the full set of protein interactions in human cells – the “interactome” – would require pairing up proteins on a grand scale. If each gene gives rise to one protein (though many can express more than one), there would be about 20,000 different proteins in human cells. A complete map of the human interactome would therefore entail testing about 200 million distinct pairs of proteins to see if they interact. Such breadth is well beyond current technical capabilities, so researchers usually conduct pairings of smaller numbers of proteins.

In the new study, scientists first combed the scientific literature to identify all the protein pairs that have been tested so far. Of that group, they found 11,000 high-quality interactions. Though that figure might seem large, it represents less than 10 percent of all the protein-protein interactions thought to be in the human interactome.

In combination, these studies made it appear that less than half of human proteins are densely interconnected while the rest did not participate in any interaction, the researchers found. The pattern suggested one of two possibilities: Either those proteins interact only rarely with other proteins; or, because previous studies have focused on proteins known to be involved in disease, interactions involving other, less notorious proteins have been overlooked.

To determine which conclusion was correct, the researchers generated a new map by systematically looking for interactions among some 13,000 proteins, covering almost half the space to be searched. Unlike smaller-scale studies, which were built around known disease proteins, the new study took an “unbiased” approach: Researchers tested proteins ‘across the board,’ regardless of whether they had previously been implicated in disease.

The result was a high-quality map of almost 14,000 interactions among 4,300 distinct proteins. It is the largest experimentally-determined binary protein-protein interaction map yet produced, more than doubling the number of high-quality interactions identified in the last two decades by small-scale studies. Based on the new map, the researchers found that instead of well-known proteins being densely interconnected, their interactions were evenly distributed across the proteome, suggesting that the human interactome is broader than previously appreciated.

“The map will be especially useful in cancer research,” Rolland remarked. “It revealed numerous connections between known and suspected cancer proteins, broadening the landscape for understanding the mechanics behind the disease.

“Combining our map with the results of small-scale studies, we now know of about 30,000 interacting pairs of proteins in human cells,” he continued. “This is invaluable information for researchers seeking to connect genetic changes to the disease process itself.”

The study was primarily funded by grants from the National Human Genome Research Institute at the National Institutes of Health, with additional support from the National Heart, Blood and Lung Institute, National Cancer Institute, National Institute of Child Health and Human Development and National Institute of Mental Health at the National Institutes of Health, the National Science Foundation, the Natural Sciences and Engineering Research Council of Canada, the Canadian Institute for Advanced Research, the Canada Excellence Research Chair, the Krembil Foundation, the Ontario Research Fund, the Avon Foundation, the Junta de Castilla y Leon, the Ministerio de Economia y Competitividad, the Consejo Superior de Investigaciones Científicas, the Spanish Ministerio de Ciencia e Innovación, the European Commission, the Multidisciplinary Research Partnerships of Ghent University, the Fund for Scientific Research-Flanders, the European Molecular Biology Organization, and the Dana-Farber Cancer Institute Strategic Initiative.

Joining Rolland as lead authors of the study are Murat Tasan, PhD, of the University of Toronto; Benoit Charloteaux, PhD, Nidhi Sahni, PhD, and Song Yi, PhD, of Dana-Farber; Samuel Pevzner of Dana-Farber and Boston University; and Quan Zhong, PhD, of Dana-Farber and Wright State University, Dayton, Ohio. Joining Vidal as co-senior authors are Michael A. Calderwood, PhD, David E. Hill, PhD, and Tong Hao of Dana-Farber, and Frederick P. Roth, PhD, of the University of Toronto. The co-authors are Atanas Kamburov, PhD, Xinping Yang, PhD, Lila Ghamsari, PhD, Dawit Balcha, Bridget Begg, Pascal Braun, PhD, Marc Brehme, PhD, Martin Broly, Anne-Ruxandra Carvunis, PhD, Dan Convery-Zupan, Elizabeth Dann, Matija Dreze, PhD, Amelie Dricot, Changyu Fan, Fana Gebreab, Bryan Gutierrez, Madeleine Hardy, Mike Jin, Ruth Kiros, Katja Luck, PhD, Andrew MacWilliams, Ryan Murray, Alexandre Palagi, Matthew Poulin, John Rasla, Patrick Reichert, Viviana Romero, Julie Sahalie, Annemarie Scholz, Akash Shah, Yun Shen, Kerstin Spirohn, Stanley Tam, Alexander Tejeda, Shelly Trigg, Kerwin Vega, Jennifer Walsh, and Michael Cusick, PhD, of Dana-Farber; Jasmin Coulombe-Huntington, PhD, Eric Franzosa, PhD, and Yu Xia, PhD, of McGill University; Susan Ghiassian, Jorg Menche, PhD, Amitabh Sharma, PhD, and Albert-Laszlo Barabasi, PhD, of Northeastern University; Xavier Rambout, PhD, and Jean-Claude Twizere, PhD, of the University of Liege, Liege, Belgium; Irma Lemmens, PhD, Elien Ruyssinck, and Jan Tavernier, PhD, of VIB in Ghent, Belgium; Celia Fontanillo, PhD, and Javier De Las Rivas, PhD, of the University of Salamanca, Salamanca, Spain; Roberto Mosca, PhD, and Patrick Aloy, PhD, of the Institute for Research in Biomedicine, Barcelona, Spain; Roser Corominas, PhD, Shuli Kang, PhD, Guan Ning Lin, PhD, and Lilia Iakoucheva, PhD, of the University of California, San Diego.