Discovery of high-affinity protein binding ligands--backwards

There is a pressing need for high-affinity protein binding ligands for all proteins in the human and other proteomes. Numerous groups are working to develop protein binding ligands but most approaches develop ligands using the same strategy in which a large library of structured ligands is screened against a protein target to identify a high-affinity ligand for the target. While this methodology generates high-affinity ligands for the target, it is generally an iterative process that can be difficult to adapt for the generation of ligands for large numbers of proteins. Methodology/principal findings: We have developed a class of peptide-based protein ligands, called synbodies, which allow this process to be run backwards--i.e. make a synbody and then screen it against a library of proteins to discover the target. By screening a synbody against an array of 8,000 human proteins, we can identify which protein in the library binds the synbody with high affinity. We used this method to develop a high-affinity synbody that specifically binds AKT1 with a K(d)<5 nM. It was found that the peptides that compose the synbody bind AKT1 with low micromolar affinity, implying that the affinity and specificity is a product of the bivalent interaction of the synbody with AKT1. We developed a synbody for another protein, ABL1 using the same method. Conclusions/significance: This method delivered a high-affinity ligand for a target protein in a single discovery step. This is in contrast to other techniques that require subsequent rounds of mutational improvement to yield nanomolar ligands. As this technique is easily scalable, we believe that it could be possible to develop ligands to all the proteins in any proteome using this approach.

This method delivered a high-affinity ligand for a target protein in a single discovery step. This is in contrast to other techniques that require subsequent rounds of mutational improvement to yield nanomolar ligands. As this technique is easily scalable, we believe that it could be possible to develop ligands to all the proteins in any proteome using this approach.