De novo design of protein interactions with learned surface fingerprints.

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.

Exploring "dark-matter" protein folds using deep learning.

De novo protein design explores uncharted sequence and structure space to generate novel proteins not sampled by evolution. A main challenge in de novo design involves crafting "designable" structural templates to guide the sequence searches toward adopting target structures. We present a convolutional variational autoencoder that learns patterns of protein structure, dubbed Genesis. We coupled Genesis with trRosetta to design sequences for a set of protein folds and found that Genesis is capable of reconstructing native-like distance and angle distributions for five native folds and three novel, the so-called "dark-matter" folds as a demonstration of generalizability. We used a high-throughput assay to characterize the stability of the designs through protease resistance, obtaining encouraging success rates for folded proteins. Genesis enables exploration of the protein fold space within minutes, unrestricted by protein topologies. Our approach addresses the backbone designability problem, showing that small neural networks can efficiently learn structural patterns in proteins. A record of this paper's transparent peer review process is included in the supplemental information.

Computational design of novel protein-protein interactions - An overview on methodological approaches and applications.

Protein-protein interactions (PPIs) govern numerous cellular functions in terms of signaling, transport, defense and many others. Designing novel PPIs poses a fundamental challenge to our understanding of molecular interactions. The capability to robustly engineer PPIs has immense potential for the development of novel synthetic biology tools and protein-based therapeutics. Over the last decades, many efforts in this area have relied purely on experimental approaches, but more recently, computational protein design has made important contributions. Template-based approaches utilize known PPIs and transplant the critical residues onto heterologous scaffolds. De novo design instead uses computational methods to generate novel binding motifs, allowing for a broader scope of the sites engaged in protein targets. Here, we review successful design cases, giving an overview of the methodological approaches used for templated and de novo PPI design.

Z-REX uncovers a bifurcation in function of Keap1 paralogs.

Studying electrophile signaling is marred by difficulties in parsing changes in pathway flux attributable to on-target, vis-à-vis off-target, modifications. By combining bolus dosing, knockdown, and Z-REX-a tool investigating on-target/on-pathway electrophile signaling, we document that electrophile labeling of one zebrafish-Keap1-paralog (zKeap1b) stimulates Nrf2- driven antioxidant response (AR) signaling (like the human-ortholog). Conversely, zKeap1a is a dominant-negative regulator of electrophile-promoted Nrf2-signaling, and itself is nonpermissive for electrophile-induced Nrf2-upregulation. This behavior is recapitulated in human cells: (1) zKeap1b-expressing cells are permissive for augmented AR-signaling through reduced zKeap1b-Nrf2 binding following whole-cell electrophile treatment; (2) zKeap1a-expressing cells are non-permissive for AR-upregulation, as zKeap1a-Nrf2 binding capacity remains unaltered upon whole-cell electrophile exposure; (3) 1:1 ZKeap1a:zKeap1b-co-expressing cells show no Nrf2-release from the Keap1-complex following whole-cell electrophile administration, rendering these cells unable to upregulate AR. We identified a zKeap1a-specific point-mutation (C273I) responsible for zKeap1a's behavior during electrophilic stress. Human-Keap1(C273I), of known diminished Nrf2-regulatory capacity, dominantly muted electrophile-induced Nrf2-signaling. These studies highlight divergent and interdependent electrophile signaling behaviors, despite conserved electrophile sensing.

The more the merrier: how homo-oligomerization alters the interactome and function of ribonucleotide reductase.

Stereotyped as a nexus of dNTP synthesis, the dual-subunit enzyme - ribonucleotide reductase (RNR) - is coming into view as a paradigm of oligomerization and moonlighting behavior. In the present issue of 'omics', we discuss what makes the larger subunit of this enzyme (RNR-α) so interesting, highlighting its emerging cellular interactome based on its unique oligomeric dynamism that dictates its compartment-specific occupations. Linking the history of the field with the multivariable nature of this exceedingly sophisticated enzyme, we further discuss implications of new data pertaining to DNA-damage response, S-phase checkpoints, and ultimately tumor suppression. We hereby hope to provide ideas for those interested in these fields and exemplify conceptual frameworks and tools that are useful to study RNR's broader roles in biology.

The mRNA-Binding Protein HuR Is a Kinetically-Privileged Electrophile Sensor.

The key mRNA-binding proteins HuR and AUF1 are reported stress sensors in mammals. Intrigued by recent reports of sensitivity of these proteins to the electrophilic lipid prostaglandin A2 and other redox signals, we here examined their sensing abilities to a prototypical redox-linked lipid-derived electrophile, 4-hydroxynonenal (HNE). Leveraging our T-REX electrophile delivery platform, we found that only HuR, and not AUF1, is a kinetically-privileged sensor of HNE in HEK293T cells, and sensing functions through a specific cysteine, C13. Cells depleted of HuR, upon treatment with HNE, manifest unique alterations in cell viability and Nrf2-transcription-factor-driven antioxidant response (AR), which our recent work shows is regulated by HuR at the Nrf2-mRNA level. Mutagenesis studies showed that C13-specific sensing alone is not sufficient to explain HuR-dependent stress responsivities, further highlighting a complex context-dependent layer of Nrf2/AR regulation through HuR.

Single-Protein-Specific Redox Targeting in Live Mammalian Cells and C. elegans.

T-REX (targetable reactive electrophiles and oxidants) enables electrophile targeting in living systems with high spatiotemporal precision and at single-protein-target resolution. T-REX allows functional consequences of individual electrophile signaling events to be directly linked to on-target modifications. T-REX is accomplished by expressing a HaloTagged protein of interest (POI) and introducing a Halo-targetable bioinert photocaged precursor to a reactive electrophilic signal (RES). Light exposure releases the unfettered RES on demand, enabling precision modification of the POI due to proximity. Using alkyne-functionalized 4-hydroxynonenal (HNE) as a representative RES, this protocol delineates optimized strategies to (1) execute T-REX in live human cells and C. elegans, (2) quantitate the POI's RES-sensitivity by either azido-fluorescent-dye conjugation or (3) enrich using biotin-azide/streptavidin pulldown procedure in both model systems, and (4) identify the site of RES-labeling on the POI using proteomics. Built-in T-REX controls that allow users to directly confirm on-target/on-site specificity of RES-sensing are also described. © 2018 by John Wiley & Sons, Inc.