NEW YORK — Disease-linked mutations, including both somatic mutations found in tumors and germline variants, are more likely to disrupt protein-protein interactions than other types of variants, a new study has found.
While many genetic alterations have been tied to disease, how these changes lead to disease phenotypes has often not been clear. In a study published on Monday in Nature Genetics, researchers from Harvard Medical School and elsewhere examined the effect of disease-associated alteration on protein networks.
An approach like this, they added, begins to close the gap between genomics and clinical medicine by taking a broader view of how a genetic variant can lead to an abnormal phenotype. "This more 'holistic' approach to defining the genetic/genomic basis of disease provides a more nuanced, more precise, and potentially more personalized approach to the genomic basis of clinical disease," senior author Joseph Loscalzo, chair of the department of medicine at Brigham and Women's Hospital, said in an email.
Using data from the Human Gene Mutation Database, The Cancer Genome Atlas, the Exome Aggregation Consortium, and the 1000 Genomes Project, the researchers found not only that disease-linked germline variants are enriched in areas encoding regions of proteins that interact with each other, but also that somatic mutations affecting protein-protein interactions (PPIs) are enriched in tumors. Within tumors, some of these interactions are also tied to patient survival or drug response, the researchers noted.
"We were certainly intrigued to find that interface mutations were so much more common than non-interface mutations," Loscalzo added. "That they are supports the importance of the PPI network as a template for defining disease etiology ... and offers novel strategies for identifying drug targets for cancers as well as germline genetic diseases."
He and his colleagues assembled a human protein-protein interactome network by drawing on protein crystal structures, modeling, and other experimental data. In all, the network included 121,575 protein-protein interactions connecting more than 15,000 unique proteins. Disease-associated mutations from the Human Gene Mutation Database, they found, were enriched in PPI interfaces, as compared to variants from the 1000 Genomes Project and ExAC.
The researchers also applied their approach to TCGA data, examining more than 1.75 million missense somatic mutations from nearly 11,000 tumor exomes that represented 33 cancer types, and also found an enrichment of mutations at PPI interfaces.
They noted, for instance, that the p.Val600Glu BRAF alteration affects the interaction between BRAF and MAP2K1, as previous studies have suggested.
These disrupted PPIs also point to potential treatment approaches. Based on the pharmacogenomic profiles of more than 1,000 cancer cell lines from the Genomics of Drug Sensitivity in Cancer database, the researchers noted that predicted oncoPPI mutations were correlated with sensitivity or resistance to therapeutic agents. Further, in an analysis of about 10,000 patient-derived tumor xenografts, they uncovered 2,808 significant correlations between 49 therapeutics and 1,441 putative oncoPPIs. For example, they noted that amino acid substitutions in the VCL protein where it interacts with FXR1 were correlated with resistance to encorafenib, a melanoma treatment.
They similarly noted a correlation between patient survival and oncoPPIs.
The researchers further validated their findings by examining 13 high-confidence oncoPPIs in systematic binary interaction assays and the effect of two oncoPPIs on tumor cell growth. Within pancreatic cancer cells, they found that a p.Ser427Phe alteration that affects the RXRA-PPARG interaction promotes pancreatic cancer cell growth and indicates sensitivity to PPAR antagonists.
Going forward, Loscalzo said his team plans to examine the functional relevance of some of the mutant PPIs they identified as well as to test therapeutic approaches that may influence their effect. At the same time, they plan to use network-based approaches to identify rational combination therapies.