Design of protein-binding proteins from the target structure alone

Abstract

The design of proteins that bind to a specific site on the surface of a target protein using no information other than the three-dimensional structure of the target remains a challenge1–5. Here we describe a general solution to this problem that starts with a broad exploration of the vast space of possible binding modes to a selected region of a protein surface, and then intensifies the search in the vicinity of the most promising binding modes. We demonstrate the broad applicability of this approach through the de novo design of binding proteins to 12 diverse protein targets with different shapes and surface properties. Biophysical characterization shows that the binders, which are all smaller than 65 amino acids, are hyperstable and, following experimental optimization, bind their targets with nanomolar to picomolar affinities. We succeeded in solving crystal structures of five of the binder–target complexes, and all five closely match the corresponding computational design models. Experimental data on nearly half a million computational designs and hundreds of thousands of point mutants provide detailed feedback on the strengths and limitations of the method and of our current understanding of protein–protein interactions, and should guide improvements of both. Our approach enables the targeted design of binders to sites of interest on a wide variety of proteins for therapeutic and diagnostic applications.

AlphaFold2 can predict single-mutation effects

“…Using AF2, we can bridge the gap between sequence and function in deep-mutational scan experiments, guide directed evolution studies, 22 and design drugs in silico . 23 On a smaller scale, AF2 can be used to screen potential mutants in costly experiments where the number of mutations is limited. Overall, it appears that AF2 provides a step change in our ability to study evolution.…”

Tertiary motifs as building blocks for the design of protein‐binding peptides

“…Despite numerous advances in computational methodology, it remains difficult to design binders of a protein target de novo. The majority of successful studies have approached this task using repurposed protein scaffolds, 5 , 14 , 16 effectively prioritizing the stability of the binding protein over the formation of an optimal interface with the target. An alternative function‐guided approach is to design the binder to be complementary to the desired protein target binding site.…”

“… 14 , 15 This approach has a consistently low success rate and must be coupled with high‐throughput experimental screening to identify leads and optimize affinity. 5 , 14 , 16 While it is difficult to diagnose all of the factors that contribute to the low success rates, it is likely that designed binders do not form sufficiently complementary interfaces. Indeed, analyses of designed binders have found that unsuccessful designs generally make fewer contacts with the target protein surface, compared to successful designs.…”

One-sided design of protein-protein interaction motifs using deep learning

“…The remainder of the interface was then redesigned as part of this focused search. While this approach resulted in the successful design of new binding proteins against 12 different targets, it required both massive computing resources and experimental screening of 715,000 designs 7 . This highlights that the problem is tremendously challenging, but solvable.…”

“…This search had several shortcomings; it was limited to only certain backbone conformations; sampling of hotspot residues positioning was poorly sampled and many redocked residues had none-productive backbone orientations which increased the challenge to identify protein backbones that could harbor these residues. In a more recent advancement 7 , an exhaustive, hierarchical docking search of single, disembodied residues was performed using a simplified forcefield, and scaffolds were then placed onto these to host these contacts, regardless of their binding energies. In a second iterations, backbones fragments presenting some of the pre-docked residues were extracted and re-grafted into more scaffold proteins.…”