Flowing through the Interactome
High-throughput experimental methods are widely used today to identify genes and proteins involved in a particular process, but not all molecules in a pathway can be identified in this manner. To fill the gaps, a new computer program called ResponseNet follows the path of least resistance—like water flowing from sources to sinks in a terrain—to find the most efficient path through the maze of interacting molecules in a cell (the “interactome). The work was published in the March 2009 issue of Nature Genetics.
ResponseNet “is a step toward a much more realistic and mechanistic view of what’s going on in cells that could ultimately do much better in terms of predicting what’s important in diseases,” says co-author Ernest Fraenkel, PhD, a biological engineer at the Massachusetts Institute of Technology (MIT). Indeed, Fraenkel and his colleagues have already produced the first cellular map of the proteins and genes that respond to alpha-synuclein—a key protein linked to Parkinson’s disease.
Two important types of high-throughput experiments are commonly used to identify genes and proteins that are important in a particular condition or disease: mRNA profiling, which measures changes in gene expression under various conditions; and genetic screening, which finds genes that, when deleted or altered, change how cells respond to stimuli. But some components of signaling pathways don’t show up in these experiments. In addition, there’s surprisingly little overlap between the genes identified via these two techniques: Genes found by genetic screening tend to be involved in regulating other genes while genes found by mRNA profiling are often part of metabolic processes. The team hypothesized that the two might be connected; that is, the genes found in genetic screens might be controlling those found by mRNA profiling.
To test their idea, the team turned to the yeast interactome, a massive and complex network of all known yeast protein-gene and protein-protein interactions. “The data are very noisy and incomplete, which means that everything can be connected to everything,” says team member Esti Yeger-Lotem, PhD, an MIT postdoc. Using a flow algorithm—an approach commonly used in the telecommunications industry—they sought the most efficient path from the regulators (genetic screen results) to the differentially expressed genes (mRNA profiling results). “By doing that, ResponseNet identifies intermediary proteins that are predicted to be part of response pathways but are not found by high-throughput methods,” says Laura Riva, PhD, also a postdoc at MIT.
The researchers tested their approach in cells that overexpress alpha-synuclein, a protein that is associated with Parkinson’s disease. “ResponseNet was able to provide the first cellular map of the proteins and genes responding to alpha-synuclein expression,” Riva says.
“Their solution is novel and makes an important step,” comments Aviv Regev, PhD, a computational and systems biologist at MIT and the Broad Institute who was not involved in the work. While the research team hopes to apply this technique to mammalian cells, “the key challenge in applying it to higher organisms is the lack of interaction data to the same scale and coverage as in yeast,” Regev says. For now, though, ResponseNet will make the yeast model a more powerful tool for studying neurodegenerative and other diseases.