6 Apr 2022

Deep learning guided optimization of human antibody against SARS-CoV-2 variants with broad neutralization (PNAS)

The ability of viruses to mutate and evade the human immune system and neutralizing antibodies remains an obstacle to antiviral and vaccine development. Many neutralizing antibodies, including some approved for emergency use authorization (EUA), reduced or lost activity against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants. Here, we introduce a geometric deep learning algorithm that efficiently enhances antibody affinity to achieve broader and more potent neutralizing activity against such variants. We demonstrate the utility of our approach on a human antibody P36-5D2, which is effective against SARS-CoV-2 Alpha, Beta, and Gamma but not Delta. We show that our geometric neural network model optimizes this antibody's complementaritydetermining region (CDR) sequences to improve its binding affinity against multiple SARS-CoV-2 variants. Through iterative optimization of the CDR regions and experimental measurements, we enable expanded antibody breadth and improved potency by similar to 10to 600-fold against SARS-CoV-2 variants, including Delta. We have also demonstrated that our approach can identify CDR changes that alleviate the impact of two Omicron mutations on the epitope. These results highlight the power of our deep learning approach in antibody optimization and its potential application to engineering other protein molecules. Our optimized antibodies can potentially be developed into antibody drug candidates for current and emerging SARS-CoV-2 variants.

6 Apr 2022

Ensemble cryo-EM reveals conformational states of the nsp13 helicase in the SARS-CoV-2 helicase replication–transcription complex (Nature Structural and Molecular Biology)

The SARS-CoV-2 nonstructural proteins coordinate genome replication and gene expression. Structural analyses revealed the basis for coupling of the essential nsp13 helicase with the RNA-dependent RNA polymerase (RdRp) where the holo-RdRp and RNA substrate (the replication–transcription complex or RTC) associated with two copies of nsp13 (nsp13₂–RTC). One copy of nsp13 interacts with the template-RNA in an opposing polarity to the RdRp and is envisaged to drive the RdRp backward on the RNA template (backtracking), prompting questions as to how the RdRp can efficiently synthesize RNA in the presence of nsp13. Here we use cryogenic-electron microscopy and molecular dynamics simulations to analyze the nsp13₂–RTC, revealing four distinct conformational states of the helicases. The results indicate a mechanism for the nsp13₂–RTC to turn backtracking on and off, using an allosteric mechanism to switch between RNA synthesis or backtracking in response to stimuli at the RdRp active site.

6 Apr 2022

Structural and biochemical rationale for enhanced spike protein fitness in delta and kappa SARS-CoV-2 variants (Nature Communications)

The Delta and Kappa variants of SARS-CoV-2 co-emerged in India in late 2020, with the Delta variant underlying the resurgence of COVID-19, even in countries with high vaccination rates. In this study, we assess structural and biochemical aspects of viral fitness for these two variants using cryo-electron microscopy (cryo-EM), ACE2-binding and antibody neutralization analyses. Both variants demonstrate escape of antibodies targeting the N-terminal domain, an important immune hotspot for neutralizing epitopes. Compared to wild-type and Kappa lineages, Delta variant spike proteins show modest increase in ACE2 affinity, likely due to enhanced electrostatic complementarity at the RBD-ACE2 interface, which we characterize by cryo-EM. Unexpectedly, Kappa variant spike trimers form a structural head-to-head dimer-of-trimers assembly, which we demonstrate is a result of the E484Q mutation and with unknown biological implications. The combination of increased antibody escape and enhanced ACE2 binding provides an explanation, in part, for the rapid global dominance of the Delta variant.

6 Apr 2022

Structure determination of high-energy states in a dynamic protein ensemble (Nature)

Macromolecular function frequently requires that proteins change conformation into high-energy states. However, methods for solving the structures of these functionally essential, lowly populated states are lacking. Here Dorothee Kern at.al. develop a method for high-resolution structure determination of minorly populated states by coupling NMR spectroscopy-derived pseudocontact shifts (PCSs) with Carr–Purcell–Meiboom–Gill (CPMG) relaxation dispersion (PCS–CPMG). Their approach additionally defines the corresponding kinetics and thermodynamics of high-energy excursions, thereby characterizing the entire free-energy landscape. Using a large set of simulated data for adenylate kinase (Adk), calmodulin and Src kinase, they find that high-energy PCSs accurately determine high-energy structures (with a root mean squared deviation of less than 3.5 angström). Applying their methodology to Adk during catalysis, they find that the high-energy excursion involves surprisingly small openings of the AMP and ATP lids. This previously unresolved high-energy structure solves a longstanding controversy about conformational interconversions that are rate-limiting for catalysis. Primed for either substrate binding or product release, the high-energy structure of Adk suggests a two-step mechanism combining conformational selection to this state, followed by an induced-fit step into a fully closed state for catalysis of the phosphoryl-transfer reaction. Unlike other methods for resolving high-energy states, such as cryo-electron microscopy and X-ray crystallography, their solution PCS–CPMG approach excels in cases involving domain rearrangements of smaller systems (less than 60 kDa) and populations as low as 0.5%, and enables the simultaneous determination of protein structure, kinetics and thermodynamics while proteins perform their function.

6 Jan 2022

Time-resolved structural analysis of an RNA-cleaving DNA catalyst (Nature)

The 10–23 DNAzyme is one of the most prominent catalytically active DNA sequences. Its ability to cleave a wide range of RNA targets with high selectivity entails a substantial therapeutic and biotechnological potential. However, the high expectations have not yet been met, a fact that coincides with the lack of high-resolution and time-resolved information about its mode of action^. Here we provide high-resolution NMR characterization of all apparent states of the prototypic 10–23 DNAzyme and present a comprehensive survey of the kinetics and dynamics of its catalytic function. The determined structure and identified metal-ion-binding sites of the precatalytic DNAzyme–RNA complex reveal that the basis of the DNA-mediated catalysis is an interplay among three factors: an unexpected, yet exciting molecular architecture; distinct conformational plasticity; and dynamic modulation by metal ions. We further identify previously hidden rate-limiting transient intermediate states in the DNA-mediated catalytic process via real-time NMR measurements. Using a rationally selected single-atom replacement, we could considerably enhance the performance of the DNAzyme, demonstrating that the acquired knowledge of the molecular structure, its plasticity and the occurrence of long-lived intermediate states constitutes a valuable starting point for the rational design of next-generation DNAzymes.

6 Jan 2022

De novo protein design by deep network hallucination (Nature)

There has been considerable recent progress in protein structure prediction using deep neural networks to predict inter-residue distances from amino acid sequences^. Here we investigate whether the information captured by such networks is sufficiently rich to generate new folded proteins with sequences unrelated to those of the naturally occurring proteins used in training the models. We generate random amino acid sequences, and input them into the trRosetta structure prediction network to predict starting residue–residue distance maps, which, as expected, are quite featureless. We then carry out Monte Carlo sampling in amino acid sequence space, optimizing the contrast (Kullback–Leibler divergence) between the inter-residue distance distributions predicted by the network and background distributions averaged over all proteins. Optimization from different random starting points resulted in novel proteins spanning a wide range of sequences and predicted structures. We obtained synthetic genes encoding 129 of the network-‘hallucinated’ sequences, and expressed and purified the proteins in Escherichia coli; 27 of the proteins yielded monodisperse species with circular dichroism spectra consistent with the hallucinated structures. We determined the three-dimensional structures of three of the hallucinated proteins, two by X-ray crystallography and one by NMR, and these closely matched the hallucinated models. Thus, deep networks trained to predict native protein structures from their sequences can be inverted to design new proteins, and such networks and methods should contribute alongside traditional physics-based models to the de novo design of proteins with new functions.

6 Jan 2022

Structure-guided bifunctional molecules hit a DEUBAD-lacking hRpn13 species upregulated in multiple myeloma (Nature Communications)

Proteasome substrate receptor hRpn13 is a promising anti-cancer target. By integrated in silico and biophysical screening, we identified a chemical scaffold that binds hRpn13 with non-covalent interactions that mimic the proteasome and a weak electrophile for Michael addition. hRpn13 Pru domain binds proteasomes and ubiquitin whereas its DEUBAD domain binds deubiquitinating enzyme UCHL5. NMR revealed lead compound XL5 to interdigitate into a hydrophobic pocket created by lateral movement of a Pru β-hairpin with an exposed end for Proteolysis Targeting Chimeras (PROTACs). Implementing XL5-PROTACs as chemical probes identified a DEUBAD-lacking hRpn13 species (hRpn13^Pru) present naturally with cell type-dependent abundance. XL5-PROTACs preferentially target hRpn13^Pru, causing its ubiquitination. Gene-editing and rescue experiments established hRpn13 requirement for XL5-PROTAC-triggered apoptosis. These data establish hRpn13 as an anti-cancer target for multiple myeloma and introduce an hRpn13-targeting scaffold that can be optimized for preclinical trials against hRpn13^Pru-producing cancer types.

5 Jan 2022

Asymmetric drug binding in an ATP-loaded inward-facing state of an ABC transporter (Nature Chemical Biology)

Substrate efflux by ATP-binding cassette (ABC) transporters, which play a major role in multidrug resistance, entails the ATP-powered interconversion between transporter intermediates. Despite recent progress in structure elucidation, a number of intermediates have yet to be visualized and mechanistically interpreted. Here, we combine cryogenic-electron microscopy (cryo-EM), double electron–electron resonance spectroscopy and molecular dynamics simulations to profile a previously unobserved intermediate of BmrCD, a heterodimeric multidrug ABC exporter from Bacillus subtilis. In our cryo-EM structure, ATP-bound BmrCD adopts an inward-facing architecture featuring two molecules of the substrate Hoechst-33342 in a striking asymmetric head-to-tail arrangement. Deletion of the extracellular domain capping the substrate-binding chamber or mutation of Hoechst-coordinating residues abrogates cooperative stimulation of ATP hydrolysis. Together, our findings support a mechanistic role for symmetry mismatch between the nucleotide binding and the transmembrane domains in the conformational cycle of ABC transporters and is of notable importance for rational design of molecules for targeted ABC transporter inhibition.

5 Jan 2022

Structural and functional diversity among agonist-bound states of the GLP-1 receptor (Nature Chemical Biology)

Recent advances in G-protein-coupled receptor (GPCR) structural elucidation have strengthened previous hypotheses that multidimensional signal propagation mediated by these receptors depends, in part, on their conformational mobility; however, the relationship between receptor function and static structures is inherently uncertain. Here, we examine the contribution of peptide agonist conformational plasticity to activation of the glucagon-like peptide 1 receptor (GLP-1R), an important clinical target. We use variants of the peptides GLP-1 and exendin-4 (Ex4) to explore the interplay between helical propensity near the agonist N terminus and the ability to bind to and activate the receptor. Cryo-EM analysis of a complex involving an Ex4 analog, the GLP-1R and G_s heterotrimer revealed two receptor conformers with distinct modes of peptide–receptor engagement. Our functional and structural data, along with molecular dynamics (MD) simulations, suggest that receptor conformational dynamics associated with flexibility of the peptide N-terminal activation domain may be a key determinant of agonist efficacy.

5 Jan 2022

Mechanism of siRNA production by a plant Dicer-RNA complex in dicing-competent conformation (Science)

In eukaryotes, small RNAs (sRNAs) play critical roles in multiple biological processes. Dicer endonucleases are a central part of sRNA biogenesis. In plants, DICER-LIKE PROTEIN 3 (DCL3) produces 24-nucleotide (nt) small interfering RNAs (siRNAs) that determine the specificity of the RNA-directed DNA methylation pathway. Here, we determined the structure of a DCL3–pre-siRNA complex in an active dicing-competent state. The 5′-phosphorylated A1 of the guide strand and the 1-nt 3′ overhang of the complementary strand are specifically recognized by a positively charged pocket and an aromatic cap, respectively. The 24-nt siRNA length dependence relies on the separation between the 5′-phosphorylated end of the guide RNA and dual cleavage sites formed by the paired ribonuclease III domains. These structural studies, complemented by functional data, provide insight into the dicing principle for Dicers in general.

10 Nov 2021

Few-fs resolution of a photoactive protein traversing a conical intersection (Nature)

The structural dynamics of a molecule are determined by the underlying potential energy landscape. Conical intersections are funnels connecting otherwise separate potential energy surfaces. Posited almost a century ago, conical intersections remain the subject of intense scientific interest. In biology, they have a pivotal role in vision, photosynthesis and DNA stability. Accurate theoretical methods for examining conical intersections are at present limited to small molecules. Experimental investigations are challenged by the required time resolution and sensitivity. Current structure-dynamical understanding of conical intersections is thus limited to simple molecules with around ten atoms, on timescales of about 100 fs or longer. Spectroscopy can achieve better time resolutions, but provides indirect structural information. Here A. Ourmazd et.al. present few-femtosecond, atomic-resolution videos of photoactive yellow protein, a 2,000-atom protein, passing through a conical intersection. These videos, extracted from experimental data by machine learning, reveal the dynamical trajectories of de-excitation via a conical intersection, yield the key parameters of the conical intersection controlling the de-excitation process and elucidate the topography of the electronic potential energy surfaces involved.