Oligomerization-driven avidity correlates with SARS-CoV-2 cellular binding and inhibition.

Cellular processes are controlled by the thermodynamics of the underlying biomolecular interactions. Frequently, structural investigations use one monomeric binding partner, while ensemble measurements of binding affinities generally yield one affinity representative of a 1:1 interaction, despite the majority of the proteome consisting of oligomeric proteins. For example, viral entry and inhibition in SARS-CoV-2 involve a trimeric spike surface protein, a dimeric angiotensin-converting enzyme 2 (ACE2) cell-surface receptor and dimeric antibodies. Here, we reveal that cooperativity correlates with infectivity and inhibition as opposed to 1:1 binding strength. We show that ACE2 oligomerizes spike more strongly for more infectious variants, while exhibiting weaker 1:1 affinity. Furthermore, we find that antibodies use induced oligomerization both as a primary inhibition mechanism and to enhance the effects of receptor-site blocking. Our results suggest that naive affinity measurements are poor predictors of potency, and introduce an antibody-based inhibition mechanism for oligomeric targets. More generally, they point toward a much broader role of induced oligomerization in controlling biomolecular interactions.

Biobox: a toolbox for biomolecular modelling.

MOTIVATION: The implementation of biomolecular modelling methods and analyses can be cumbersome, often carried out with in-house software reimplementing common tasks, and requiring the integration of diverse software libraries. RESULTS: We present Biobox, a Python-based toolbox facilitating the implementation of biomolecular modelling methods. AVAILABILITY AND IMPLEMENTATION: Biobox is freely available on https://github.com/degiacom/biobox, along with its API and interactive Jupyter notebook tutorials.

Transmembrane Protein Docking with JabberDock.

Transmembrane proteins act as an intermediary for a broad range of biological process. Making up 20% to 30% of the proteome, their ubiquitous nature has resulted in them comprising 50% of all targets in drug design. Despite their importance, they make up only 4% of all structures in the PDB database, primarily owing to difficulties associated with isolating and characterizing them. Membrane protein docking algorithms could help to fill this knowledge gap, yet only few exist. Moreover, these existing methods achieve success rates lower than the current best soluble proteins docking software. We present and test a pipeline using our software, JabberDock, to dock membrane proteins. JabberDock docks shapes representative of membrane protein structure and dynamics in their biphasic environment. We verify JabberDock's ability to yield accurate predictions by applying it to a benchmark of 20 transmembrane dimers, returning a success rate of 75.0%. This makes our software very competitive among available membrane protein-protein docking tools.

Unravelling the components of diffuse scattering using deep learning.

Many technologically important material properties are underpinned by disorder and short-range structural correlations; therefore, elucidating structure-property relationships in functional materials requires understanding both the average and the local structures. The latter information is contained within diffuse scattering but is challenging to exploit, particularly in single-crystal systems. Separation of the diffuse scattering into its constituent components can greatly simplify analysis and allows for quantitative parameters describing the disorder to be extracted directly. Here, a deep-learning method, DSFU-Net, is presented based on the Pix2Pix generative adversarial network, which takes a plane of diffuse scattering as input and factorizes it into the contributions from the molecular form factor and the chemical short-range order. DSFU-Net was trained on 198 421 samples of simulated diffuse scattering data and performed extremely well on the unseen simulated validation dataset in this work. On a real experimental example, DSFU-Net successfully reproduced the two components with a quality sufficient to distinguish between similar structural models based on the form factor and to refine short-range-order parameters, achieving values comparable to other established methods. This new approach could streamline the analysis of diffuse scattering as it requires minimal prior knowledge of the system, allows access to both components in seconds and is able to compensate for small regions with missing data. DSFU-Net is freely available for use and represents a first step towards an automated workflow for the analysis of single-crystal diffuse scattering.

Ca2+-dependent lipid preferences shape synaptotagmin-1 C2A and C2B dynamics: Insights from experiments and simulations.

Signal transmission between neurons requires exocytosis of neurotransmitters from the lumen of synaptic vesicles into the synaptic cleft. Following an influx of Ca2+, this process is facilitated by the Ca2+ sensor synaptotagmin-1. The underlying mechanisms involve electrostatic and hydrophobic interactions tuning the lipid preferences of the two C2 domains of synaptotagmin-1; however, the details are still controversially discussed. We, therefore, follow a multidisciplinary approach and characterize lipid and membrane binding of the isolated C2A and C2B domains. We first target interactions with individual lipid species, and then study interactions with model membranes of liposomes. Finally, we perform molecular dynamics simulations to unravel differences in membrane binding. We found that both C2 domains, as a response to Ca2+, insert into the lipid membrane; however, C2A adopts a more perpendicular orientation while C2B remains parallel. These findings allow us to propose a mechanism for synaptotagmin-1 during membrane fusion.

Deep learning approaches for conformational flexibility and switching properties in protein design.

Following the hugely successful application of deep learning methods to protein structure prediction, an increasing number of design methods seek to leverage generative models to design proteins with improved functionality over native proteins or novel structure and function. The inherent flexibility of proteins, from side-chain motion to larger conformational reshuffling, poses a challenge to design methods, where the ideal approach must consider both the spatial and temporal evolution of proteins in the context of their functional capacity. In this review, we highlight existing methods for protein design before discussing how methods at the forefront of deep learning-based design accommodate flexibility and where the field could evolve in the future.

Mass Photometry of Membrane Proteins.

Integral membrane proteins (IMPs) are biologically highly significant but challenging to study because they require maintaining a cellular lipid-like environment. Here, we explore the application of mass photometry (MP) to IMPs and membrane-mimetic systems at the single-particle level. We apply MP to amphipathic vehicles, such as detergents and amphipols, as well as to lipid and native nanodiscs, characterizing the particle size, sample purity, and heterogeneity. Using methods established for cryogenic electron microscopy, we eliminate detergent background, enabling high-resolution studies of membrane-protein structure and interactions. We find evidence that, when extracted from native membranes using native styrene-maleic acid nanodiscs, the potassium channel KcsA is present as a dimer of tetramers-in contrast to results obtained using detergent purification. Finally, using lipid nanodiscs, we show that MP can help distinguish between functional and non-functional nanodisc assemblies, as well as determine the critical factors for lipid nanodisc formation.

Protein Docking Using a Single Representation for Protein Surface, Electrostatics, and Local Dynamics.

Predicting the assembly of multiple proteins into specific complexes is critical to understanding their biological function in an organism and thus the design of drugs to address their malfunction. Proteins are flexible molecules, which inherently pose a problem to any protein docking computational method, where even a simple rearrangement of the side chain and backbone atoms at the interface of binding partners complicates the successful determination of the correct docked pose. Herein, we present a means of representing protein surface, electrostatics, and local dynamics within a single volumetric descriptor. We show that our representations can be physically related to the surface-accessible solvent area and mass of the protein. We then demonstrate that the application of this representation into a protein-protein docking scenario bypasses the need to compensate for, and predict, specific side chain packing at the interface of binding partners. This representation is leveraged in our de novo protein docking software, JabberDock, which can accurately and robustly predict difficult target complexes with an average success rate of >54%, which is comparable to or greater than the currently available methods.