Design of protein-binding proteins from the target structure alone.
Cao L, Coventry B, Goreshnik I, Huang B, Sheffler W, Park JS, Jude KM, Marković I, Kadam RU, Verschueren KHG, Verstraete K, Walsh STR, Bennett N, Phal A, Yang A, Kozodoy L, DeWitt M, Picton L, Miller L, Strauch EM, DeBouver ND, Pires A, Bera AK, Halabiya S, Hammerson B, Yang W, Bernard S, Stewart L, Wilson IA, Ruohola-Baker H, Schlessinger J, Lee S, Savvides SN, Garcia KC, Baker D. · 2022
Read Design of protein-binding proteins from the target structure alone. by Cao L, Coventry B, Goreshnik I, Huang B, Sheffler W, Park JS, Jude KM, Marković I, Kadam RU, Verschueren KHG, Verstraete K, Walsh STR, Bennett N, Phal A, Yang A, Kozodoy L, DeWitt M, Picton L, Miller L, Strauch EM, DeBouver ND, Pires A, Bera AK, Halabiya S, Hammerson B, Yang W, Bernard S, Stewart L, Wilson IA, Ruohola-Baker H, Schlessinger J, Lee S, Savvides SN, Garcia KC, Baker D. (2022) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc9117152.txt.
Introduction
Cao and co-authors set out a general way to design proteins that bind a chosen spot on a target protein. The method starts from the target's three-dimensional shape alone. It first scans a vast range of possible binding modes. It then searches hard near the most promising modes.
The team designed binders for 12 diverse protein targets with unlike shapes and surfaces. All the binders were smaller than 65 amino acids and proved very stable. After lab tuning, they bound their targets at nanomolar to picomolar strength. Crystal structures of five binder-target pairs closely matched the design models. Data from vast design sets also show where the method still falls short. The work opens a route to binders for therapy and diagnosis.
Catalog record
| CREATOR | Cao L, Coventry B, Goreshnik I, Huang B, Sheffler W, Park JS, Jude KM, Marković I, Kadam RU, Verschueren KHG, Verstraete K, Walsh STR, Bennett N, Phal A, Yang A, Kozodoy L, DeWitt M, Picton L, Miller L, Strauch EM, DeBouver ND, Pires A, Bera AK, Halabiya S, Hammerson B, Yang W, Bernard S, Stewart L, Wilson IA, Ruohola-Baker H, Schlessinger J, Lee S, Savvides SN, Garcia KC, Baker D. |
|---|---|
| YEAR | 2022 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 120 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Design of protein-binding proteins from the target structure alone By Longxing Cao, Brian Coventry, Inna Goreshnik, Buwei Huang, William Sheffler, Joon Sung Park, Kevin M. Jude, Iva Marković, Rameshwar U. Kadam, Koen H. G. Verschueren, Kenneth Verstraete, Scott Thomas Russell Walsh Published in: Nature DOI: 10.1038/s41586-022-04654-9 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. MAIN Protein interactions have crucial roles in biology, and general approaches to design proteins that disrupt or modulate these interactions would have great utility. Empirical selection approaches that start from large antibody, designed ankyrin repeat protein or other protein scaffold libraries can generate binders to protein targets. However, it is difficult at the outset to target a specific region on a target protein surface and to sample the entire space of possible binding modes. Computational methods can target specific target surface locations and provide a more principled and a potentially faster approach to generate binders than random library selection methods, as well as insight into the fundamental properties of protein interfaces (which must be understood for design to be successful). Most current computational methods used to design proteins that bind to a target surface utilize information derived from structures of the native complex on specific side-chain interactions or protein backbone placements optimal for binding1–3. Computational docking of antibody scaffolds with varied loop geometries has yielded binders, but the designed binding modes have rarely been validated with high-resolution structures4. Binders have been generated starting from several computationally identified hotspot residues, which were then used to guide the positioning of naturally occurring protein scaffolds5. However, for many target proteins, there are no obvious pockets or clefts on the protein surface into which a small number of privileged side chains can be placed, and guidance by a small number of hotspot residues limits the approach to a small fraction of possible interaction modes. DESIGN METHOD We sought to develop a general approach to the design of high-affinity binders to arbitrary protein targets that addresses two major challenges. First, there are generally no clear side-chain interactions or secondary structure packing arrangements that can mediate strong interactions with the target; instead there are vast numbers of individually very weak possible interactions. Second, the number of ways of choosing which of these numerous weak interactions to incorporate into a single binding protein is combinatorially large, and any given protein backbone is unlikely to be able to simultaneously present side chains that can encompass any preselected subset of these interactions. To illustrate our approach, consider the simple analogy of a difficult climbing wall with only a few suitable footholds or handholds distant from each other. Previous hotspot-based approaches correspond to focusing on routes that involve these footholds and handholds, but this greatly limits possibilities and there may be no way to connect them into a successful route. An alternative is to first identify all the possible handholds and footholds, no matter how poor; second, have thousands of climbers select subsets of these and try to climb the wall; third, identify those routes that showed the most promise, and fourth, have a second group of climbers explore them in detail. Following this analogy, we devised the following multistep approach to overcome the above two challenges: step (1), enumerate a large and comprehensive set of disembodied side-chain interactions with the target surface; step (2), identify from large in silico libraries of protein backbones those that can host many of these side chains without clashing with the target; step (3), identify recurrent backbone motifs in these structures; and step (4), generate and place against the target a second round of scaffolds that contain these interacting motifs (Fig. 1a and Extended Data Fig. 1). Steps (1) and (2) widely search the space, whereas steps (3) and (4) intensify the search in the regions that show the most promise. We describe each step in more detail below.Fig. 1Overview of the de novo protein binder design pipeline.a, Schematic of our two-stage binder design approach. In the global search stage, billions of disembodied amino acids are docked onto the selected region of the target protein surface using RifGen, the favourable interacting amino acids are stored as rifres (step 1), and miniprotein scaffolds are then docked on the target guided by these favourable side-chain interactions (step 2). The interface sequences are then designed to maximize interactions with the target (step 3). In the focused search stage, interface structural motifs are extracted and clustered (steps 4 and 5). These privileged motifs are then used to guide another round of docking and design (steps 6 and 7). Designs are then selected for experimental characterization based on computational metrics (step 8). See Extended Data Fig. 1 for a more detailed flow chart of the de novo binder design pipeline. b, Comparison of the sampling efficiency of PatchDock, RifDock and resampling protocols. Bar graph shows the distribution over the three approaches of the top 1% of binders based on Rosetta ddG and contact molecular surface values after pooling equal-CPU-time dock-and-design trajectories for each of the 13 target sites and averaging per-target distributions (Methods). We began by docking disembodied amino acids against the target protein and storing the backbone coordinates and target binding energies of the typically billions of amino acids that make favourable hydrogen bonding or nonpolar interactions in a six-dimensional spatial hash table for rapid look-up (Fig. 1a and Methods). This rotamer interaction field (RIF) enables rapid approximation of the target interaction energy achievable by a protein scaffold docked against a target based on its backbone coordinates alone (with no need for time-consuming side-chain sampling). For each dock, the target interaction energies of each of the matching amino acids in the hash table are summed. A related approach was used for the design of small-molecule binders6; as protein targets are so much bigger and because nonpolar interactions are the primary driving force for protein–protein interactions, we focused the RIF generation process on nonpolar sites in specific surface regions of interest. For example, for the