Disease begins at the cellular level. Deviant proteins often wreak havoc in a dying cell long before a larger ill such as cancer or infection spreads throughout the body. Efforts to understand these proteins and how they interact with one another are at the very core of basic biomedical research, but progress has been slow.
“It’s pretty hard to look under the hood of a cell,” says Frederick Roth, a biophysicist at the University of Toronto.
To jump-start progress, scientists like Roth have been chipping away for years at the human interactome—a complete map of every protein interaction in humans. Last week Roth and his team published a paper in the journal Cell detailing one of the most comprehensive protein maps to date. It covers 14,000 interactions between protein pairs and may help scientists identify some of the genes involved in cancer.
“We found that cancer proteins stick together and tend to interact with each other,” Roth says. “That’s immediately useful. We can say, here’s a protein that we don’t know much about, but a bunch of its partners are cancer proteins, so maybe we should be studying that more carefully.”
A Manual for the Human Cell
Identifying protein interactions could help researchers understand how cells function on a molecular level. Roth draws an analogy to a mechanic who has a list of car parts but no idea how they interconnect. “This takes us from a rough draft of a list of parts, in no particular order, to a list of pairs of parts,” Roth says. “Now we can begin to understand how they fit together.”
Roth estimates that the new map captures between 5 and 10 percent of all the protein interactions in human cells. That may not sound like a lot, but the last big advance for the human interactome was almost a decade ago, when Roth released his first map of only about 3,000 protein interactions.
This new map has taken a long time to come together, but Roth says that it will prove to haven been worth the wait. “We have taken a big step toward a manual for the human cell,” he says.
Pairing Up Proteins
Roth and his team identified protein interactions using what is known as a yeast two-hybrid assay (here is a cool video of exactly how that works). To test whether or not two human proteins interact, the researchers embedded both proteins in a single yeast cell. Each protein contained half of an essential activator switch. The yeast could only survive if the human proteins paired up, joining their halves of the switch together.
“Baker’s yeast is not only great for bread, wine and beer,” Roth says. “You can also use to it read out which pairs of human proteins are sticking together.”
Yeast assays are not the only way to track protein interactions. Nevan Krogan, a professor of cellular and molecular pharmacology at the University of California, San Francisco, who was not involved in the study, is taking what he calls a more biochemically relevant approach to the interactome.
Krogan and his team place affinity tags on different proteins so that they can purify them and then analyze them with mass spectrometry. His method puts less of an emphasis on interactions between individual proteins and instead focuses on how groups of proteins interact with other protein complexes, “so you can identify the molecular machine that exists in the cell,” Krogan says.
Krogan and Roth’s approaches each have advantages and disadvantages, but Krogan believes there is much to be gained by combining results from each method. “Our two approaches are very complementary,” he says. “I think [Roth’s] data will become particularly powerful when it is combined with other kinds of information.”
Cancer by Association
Roth’s new map of the interactome has some immediate applications for cancer research. As noted above, scientists have long suspected that cancer proteins tend to stick together. Roth’s study not only confirms this, it also points researchers toward proteins that might be guilty by association with known cancer proteins.
For example, studies have linked the gene MAPK1IP1L to tumor formation in mice, but it has not been studied extensively and the protein that it produces is not currently recognized as a cancer protein in humans. Roth’s study found that MAPK1IP1L interacts with at least three known cancer proteins. That does not necessarily mean MAPK1IP1L is a cancer gene, but it does suggest an avenue for future research. At the very least, Roth says, “it’s probably worth a closer look.”
Krogan agrees that assembling the human interactome could have broad implications. “This type of information will be absolutely crucial to really understand how mutations result in disease states,” he says. “This study, and studies like this, go a long way in that particular direction.”