Protein-protein interactions are crucial for all cellular pathways, including signal transduction, DNA replication, transcription/translation, multi-component assemblies, among many others. Disruption of the protein-protein interactions pathways frequently leads to disease. Hence, learning to rationally manipulate protein-protein interactions in cells is of great importance to both basic biology and applied research such as drug design.

Design of inhibitors for matrix metalloproteinases

Matrix metalloproteinases are a family of enzymes whose malfunction has been implicated in variable diseases including arthritis, osteoporosis, ALS, Alzheimer’s disease, and cancer. The function of some MMPs is associated with the most aggressive types of cancer and directly promotes metastasis. However, only some of MMP family members play role in disease progression, while others are essential and/or play therapeutic role. Thus, all non-specific MMP inhibitors proved to be highly toxic and could not be used as drugs. We have developed a pioneering strategy for designing highly specific inhibitors of each MMP type using a combination of computational design and experimental protein engineering. Our MMP inhibitors are based on a small protein, TIMP2, that has been engineered to possess high affinity and high specificity towards each MMP family member. Our engineered proteins possess high potential to become future drugs for cancer and other diseases.

Deep mutational scanning of protein-protein interactions

Protein-protein interactions (PPIs) have evolved to possess binding affinities that best satisfy their functional role. PPIs with different affinities could be structurally very similar, exhibiting similar number and type of intermolecular interactions. To understand how slight structural variations translate into substantial differences in binding affinities, we study binding landscapes, i. e. changes in binding free energy (Gbind) due to all possible mutations. Recently, we developed state-of-the-art methodology that combines protein randomization and affinity sorting coupled to deep sequencing and data normalization that allows us to measure Gbind values for tens of thousands of mutations in a particular PPI. We are using this strategy to map binding landscapes of PPIs with various structures and functions.

Cold-spots in protein-protein interactions

Understanding the energetics and architecture of protein binding interfaces is important for basic research and could potentially facilitate the design of novel binding domains for biotechnological applications. It is well accepted that a few key residues at binding interfaces termed hot-spots of binding are responsible for contributing most of the binding free energy.  We introduced a new concept of binding cold-spots, or interface positions occupied by suboptimal amino acids. Such positions exhibit a potential for affinity enhancement through various mutations. We perform analysis of the whole PDB to study cold-spot frequency, their structural distribution and conservation pattern. Identification of cold-spot positions is crucial for studies of protein evolution and for design of new therapeutical molecules.

Vaccine design

Traditional vaccines are usually based on dead viruses containing only viral particles or virus coat proteins. Such vaccines cause the immune system to create antibodies that bind to certain epitopes on the virus surface and neutralize the virus. Many of traditional vaccines have several disadvantages such as low thermal stability and inability to neutralize different types of the virus. We are developing novel methodology for designing chimeric vaccines, where the viral epitope, most conserved among various types of viruses, is fused to a thermostable protein. Such a chimeric vaccine is highly stable and is likely to be effective in generating neutralizing antibodies against the whole family of viruses. We are testing our new methodology for vaccine