We study protein-protein interactions for a number of reasons: to understand how molecules interact, to appreciate how these interactions drive the function of the cell, to learn how some interactions can lead to disease and to develop ideas about how we could treat diseases that do arise. In a typical cell, hundreds of thousands of protein-protein interactions take place. Being able to detail the structure of these interactions can tell us how significant they might be and how we can modify them should we need to.
Trying to work out how two different 3D structures might interact isn’t so easy. In silico approaches allow researchers to identify and model multiple interactions quickly and inexpensively, and these go hand in hand with experimental studies for validation. To improve the in silico stages a new paper published in PNAS (link is external) presents an algorithm that can model protein-protein interactions ten to a hundred times faster than before with no loss in accuracy. The new technique uses fast Manifold Fourier Transform to handle the calculations, allowing them to whip through a huge number of potential protein-protein conformations simultaneously to determine their interactions in under 15 minutes using a personal laptop.
In terms of experimental investigation of protein-protein investigations Affimer proteins have a good track record in helping to detail binding interfaces and interrupt interactions for therapeutic benefit.
In a study (link is external) carried out at Leeds University Affimer proteins identified to the HIF-1? co-activator p300 were able to inhibit the interaction between HIF-1? and p300, with IC50 values in the 1-5µM range. This was a huge improvement in binding compared to an alternative commercial protein library trialled by the researchers. Modelling the interaction of the Affimer with its p300 binding partner predicted that their interaction occured via an alternative site to that of the natural HIF-1? binding partner with p300. This could form the basis of future therapeutic development targeting this pathway, based upon the pharmacophore of the interacting Affimer surface.
Other examples (link is external) of Affimer proteins being used to disrupt protein-protein interactions include the specific modulation of blood clotting, either to prevent or promote thrombosis by disrupting the interaction between fibrinogen and complement C3, and allosteric inhibitors of Fc?RIIIa- IgG interactions to prevent excessive immune responses to infection.
The use of both computer modelling and biochemical experimental validation for determining protein-protein interactions are key to informing the design of highly potent compounds that can disrupt challenging but pharmaceutically important interactions.
The new in silico modelling resource will soon be available to the scientific community through the publicly available protein-protein docking server ClusPro, which was judged to be the best automated docking server in the latest rounds of the international blind protein docking competition called CAPRI (Critical Assessment of Prediction Interaction). The authors of the study believe the new method, implemented in the ClusPro server, will bring scientists one step closer to modelling protein interactions for the whole cell, a key step in designing drugs to interrupt defective protein interactions that lead to disease.