De novo design of protein interactions with learned surface fingerprints

Abstract

Physical interactions between proteins are essential for most biological processes governing life1. However, the molecular determinants of such interactions have been challenging to understand, even as genomic, proteomic and structural data increase. This knowledge gap has been a major obstacle for the comprehensive understanding of cellular protein–protein interaction networks and for the de novo design of protein binders that are crucial for synthetic biology and translational applications2–9. Here we use a geometric deep-learning framework operating on protein surfaces that generates fingerprints to describe geometric and chemical features that are critical to drive protein–protein interactions10. We hypothesized that these fingerprints capture the key aspects of molecular recognition that represent a new paradigm in the computational design of novel protein interactions. As a proof of principle, we computationally designed several de novo protein binders to engage four protein targets: SARS-CoV-2 spike, PD-1, PD-L1 and CTLA-4. Several designs were experimentally optimized, whereas others were generated purely in silico, reaching nanomolar affinity with structural and mutational characterization showing highly accurate predictions. Overall, our surface-centric approach captures the physical and chemical determinants of molecular recognition, enabling an approach for the de novo design of protein interactions and, more broadly, of artificial proteins with function.

Computational Approaches to Predict Protein–Protein Interactions in Crowded Cellular Environments

“…They introduced MaSIF (molecular surface interaction fingerprinting), a conceptual framework utilizing geometric deep-learning techniques to capture crucial fingerprints relevant to specific biomolecular interactions, including protein pocket–ligand interactions and protein–protein interaction. Two years later, the same framework was used to design novel protein interactions …”

A Suite of Designed Protein Cages Using Machine Learning Algorithms and Protein Fragment-Based Protocols

“…While a first approach for designing protein cages used genetic fusion 4,29 , subsequent work (beginning with King et al 2012) has increasingly relied on the computational design of de novo non-covalent interfaces between protein subunits. A critical factor therefore in successful material design is the specification of a geometrically precise interface that is accessible and cooperative with the surrounding energy landscape 30 . The interface must encode sufficient information to drive the emergence of quaternary structure 31 , while not corrupting the energetics governing tertiary structure 32 .…”

Nanoparticle-mediated delivery of peptide-based degraders enables targeted protein degradation

“…14,15 As an alternative to small molecule-based approaches, peptidebased ligands possess large protein-protein interaction surfaces, making them suitable for targeting any POI. Coupled with the rapid development of structural biology techniques that provide detailed protein-protein structural information, 16,17 mature directedevolution technologies such as phage and yeast display, [18][19][20] and emerging computational approaches for rapid discovery of synthetic binding peptides, 6,[21][22][23] peptide-based ligands are ideal for extending the scope of PROTACs to "undruggable" proteins. 24,25 Several Peptide-based Proteolysis Targeting Chimeras (PepTACs) targeting oncoproteins and transcription factors make use of a cationic cell-penetrating peptides 26,27 , cyclic peptides, 28,29 or peptide stapling [30][31][32][33] to facilitate cellular uptake.…”