By cataloging protein-protein interactions in cell lines, and combining their results with in vivo studies as well as publicly available data, scientists have defined new interactions that could be used diagnostically, and/or harnessed, for well-studied cancer drivers and more obscure proteins alike.
In a testament to the method's power, the researchers identified several new interaction partners for BRCA1, including one that predicted sensitivity to poly (ADP-ribose) polymerase (PARP) inhibitors.
"We're finding new stable stoichometric interactions in one of the most studied proteins of all time," co-corresponding author Nevan Krogan told BioWorld Science.
Another example is phosphoinositide 3-kinase (PI3K), for which the team identified multiple previously unknown interaction partners in both head and neck cancers and breast cancer.
Krogan is the director of the Quantitative Biosciences Institute in the School of Pharmacy at the University of California, San Francisco, and co-corresponding author of the papers, which appeared in the October 1, 2021, issue of Science.
In the big picture, the work aims to understand the therapeutic relevance of the rare mutations that make a up a large part of any tumor's genetic makeup.
"We are classing the complexity that's come from the genomic data," Krogan said. That genomic data, in Krogan's take of the situation, has led to an ever-increasing level of detail without a way to separate important details from noise as far as therapeutic relevance is concerned. "Every gene in the genome has been linked to cancer in one way or the other," he said.
Krogan, co-corresponding author Trey Ideker, and their colleagues started their experiments by looking at protein complexes in cell lines for head and neck squamous cell carcinoma, a tumor type where "despite a wealth of data detailing the genetic alterations... few targeted therapies are available," they wrote.
The team used that wealth to select 31 proteins for study, including 16 different mutations of the PI3K catalytic subunit PIK3CA. PI3K is the most frequently mutated gene in head and neck cancer, but many of the variants are of unknown significance.
Looking at two cancer cell lines and a noncancerous esophageal cell line, the team uncovered close to 800 protein-protein interactions of both wild-type and mutant proteins, only about 15% of which had already been reported in public databases.
They also showed that the different mutations of PIK3CA had different functional consequences. Follow-up in vivo experiments showed that some mutations predicted the response to HER3 inhibitors.
For breast cancer, the team looked at cell lines modeling estrogen receptor-driven and triple-negative breast cancer, which together make up more than 90% of breast cancer cases. Here, too, they were able to identify new interaction partners for PI3K, as well as for BRCA.
In a commentary published with the paper, researchers from Stanford University wrote that the findings "underscore the possibilities for using specific functional assays to further mine PPI data and the value of generating equivalent studies in other cancers."
Krogan and his team are now using their approach to look at lung cancer, as well as infectious diseases. Neurodegenerative disorders are another area that could benefit from innovative approaches.
"A key point here is that the technology is disease agnostic," he said. He predicted that looking at proteins in different diseases will break down the current categories of those diseases.
We're making connections between scientists working in different disease areas," he said. "Science is so siloed...There's such great overlap between diseases that we just don't know until we do these kinds of studies."