Computational Tool Predicts Protein-Protein Interactions

Researchers from Cornell University and the Cleveland Clinic have developed an open-access web database and software to facilitate the discovery of critical protein-protein interactions that can be targeted with medications.

The study, published in Nature Biotechnology, introduces PIONEER (Protein-protein InteractiOn iNtErfacE pRediction), a computational tool designed to streamline this process. The researchers demonstrated PIONEER’s effectiveness by identifying potential therapeutic targets for various cancers and other complex diseases.

Feixiong Cheng, Ph.D., Co-Lead Author and Head of the Genome Center at the Cleveland Clinic, explains that while genomic research plays a crucial role in drug discovery, it is often insufficient by itself. Typically, it takes ten to fifteen years from identifying a disease-causing gene to initiating clinical trials for drugs based on genetic data.

In theory, making new medicines based on genetic data is straightforward: mutated genes make mutated proteins. We try to create molecules that stop these proteins from disrupting critical biological processes by blocking them from interacting with healthy proteins, but in reality, that is much easier said than done.

Feixiong Cheng, Study Co-Lead Author and Head, Genome Center, Cleveland Clinic

In the human body, a single protein can interact in multiple ways with hundreds of other proteins, forming a complex network of interactions known as the interactome. According to Dr. Cheng, each protein’s interactions can branch out to hundreds more, creating a vast, intricate web. The challenge intensifies when disease-causing DNA mutations are introduced. A single altered protein can give rise to multiple interactomes, as some genes may undergo different mutations yet result in the same disease.

Drug developers face tens of thousands of potential interactions that could contribute to disease. To streamline this process, Dr. Cheng collaborated with Dr. Haiyuan Yu, Director of the Cornell University Center for Innovative Proteomics, to create an AI tool that helps genetic and genomic researchers, as well as drug developers, pinpoint the most promising protein-protein interactions. The team integrated large datasets, including:

• Genomic sequences from nearly 100,000 individuals with disease-causing mutations, either congenital or acquired later in life (such as in cancer),
• The three-dimensional structures of over 16,000 human proteins, along with mutation impact data,
• Details on nearly 300,000 known protein-protein interactions.

The database allows researchers to explore the interactome for over 10,500 disorders, from von Willebrand disease to alopecia. When a disease-associated mutation is identified, researchers can input it into PIONEER to receive a prioritized list of potentially drug-targetable protein-protein interactions.

Researchers can also search for a disease by name to access a list of possible protein interactions linked to that illness for further investigation. PIONEER is beneficial for biomedical researchers across a wide range of fields, including autoimmune, cancer, cardiovascular, metabolic, neurological, and pulmonary disorders.

The researchers validated their database predictions in the lab by creating over 3,000 mutations across more than 1,000 proteins and studying their effects on nearly 7,000 protein-protein interaction pairs. Building on these findings, initial studies are now underway to develop and test treatments for endometrial and lung cancers. The team also demonstrated that mutations affecting protein-protein interactions in their model could potentially predict:

• Prognoses and survival rates for various cancers, including sarcoma, a rare but potentially deadly cancer.
• Responses to anti-cancer drugs within large pharmacogenomics databases.

They confirmed experimentally that changes in the interaction between NRF2 and KEAP1 proteins can predict lung cancer tumor growth, highlighting a new target for developing personalized cancer therapies.

The resources needed to conduct interactome studies poses a significant barrier to entry for most genetic researchers. We hope PIONEER can overcome these barriers computationally to lessen the burden and grant more scientists with the ability to advance new therapies.

Feixiong Cheng, Study Co-Lead Author and Head, Genome Center, Cleveland Clinic

Journal Reference:

Xiong, D., et al. (2024) A structurally informed human protein-protein interactome reveals proteome-wide perturbations caused by disease mutations. Nature Biotechnology. doi.org/10.1038/s41587-024-02428-4.