(413d) Computational Docking and Design of Fixed-Backbone Binding Protein Scaffolds for Target Epitopes

Binding proteins play a major role in several biological processes such as immunity, signaling, transportation etc. Antibodies are the most widely studied binding proteins with therapeutic and biomarker applications. These applications have led to the development of a $100+ billion global market of therapeutic antibodies. Despite their success, smaller sized binding scaffolds have been developed to overcome disadvantages of antibodies such as large size and production costs, instability, post translational modifications etc. Smaller sized scaffolds retain the high affinity and specificity, are more stable, easier to synthesize and have higher tissue penetration. A category of these alternative scaffolds consists of fixed backbone structures with constant hydrophobic core amino acids. The experimental design of such scaffolds is done by creating libraries of binders through the randomization of surface amino acids and screening them for binding with the target antigen. Examples of fixed backbone scaffolds include Affibodies and DARPins. Therapeutics generated using Affibodies and DARPins are in Phase 2 and 3 clinical trials respectively.

To aid the arduous experimental discovery of novel binding proteins for specific epitopes, computational tools have been developed to analyze and design binding proteins. The common methodology for binding protein design, including antibodies, is to first dock the binding scaffold with the target epitope and mutate the scaffold residues to create attractive interactions. Protein-protein docking tools such as Zdock, Hex etc. are commonly used to dock binding scaffolds followed by iterative design cycles with Rosetta to create beneficial mutations. Antibody specific design methods such as OptMAVEn and RAbD also follow this dock-and-mutate protocol. While successful, a drawback of docking a fixed scaffold structure is that the quality of the docked poses is dependent on a single set of surface residues. Docking with a varied set of scaffold surface residues will increase the solution space of docked poses and lead to the identification of better starting poses that could be further improved with mutations. Hence, a method that could perform both docking and mutations simultaneously could lead to the generation of docked poses with better binding metrics.

In this work, we have developed a novel computational method to simultaneously dock and mutate user-selected scaffold towards a target epitope. The method is divided into two steps: scaffold positioning to form hydrogen bonds and variable residue mutations to fix sidechain clashes. In the first step, the method identifies locations for hydrogen bond formation around wildtype scaffold residues and rotamers of other amino acid types for each variable surface residue. Pairwise distance alignment between hydrogen bond regions and compatible epitope atoms is performed to identify groups of hydrogen bonds that can exist simultaneously. For each such solution group, the antigen is positioned to form the predicted hydrogen bonds and the docked poses are checked for steric clashes. In the second step, poses with only clashing variable scaffold residues are mutated to non-clashing amino acid rotamers that also form other types of interactions, hydrogen bond or hydrophobic burying, with the antigen.

The approach was used to dock ten randomly selected antigens with two scaffolds: Affibodies and DARPins. The docking simulations were repeated with Zdock for comparison purposes. We will present on the details of the case study results. We believe that this work can aid the rapid development of novel binders based on fixed backbone scaffolds.