High-affinity binding to the SARS-CoV-2 spike trimer by a nanostructured, trivalent protein-DNA synthetic antibody.

Multivalency enables nanostructures to bind molecular targets with high affinity. Although antibodies can be generated against a wide range of antigens, their shape and size cannot be tuned to match a given target. DNA nanotechnology provides an attractive approach for designing customized multivalent scaffolds due to the addressability and programmability of the nanostructure shape and size. Here, we design a nanoscale synthetic antibody ("nano-synbody") based on a three-helix bundle DNA nanostructure with one, two, or three identical arms terminating in a mini-binder protein that targets the SARS-CoV-2 spike protein. The nano-synbody was designed to match the valence and distance between the three receptor binding domains (RBDs) in the spike trimer, in order to enhance affinity. The protein-DNA nano-synbody shows tight binding to the wild-type, Delta, and several Omicron variants of the SARS-CoV-2 spike trimer, with affinity increasing as the number of arms increases from one to three. The effectiveness of the nano-synbody was also verified using a pseudovirus neutralization assay, with the three-arm nanostructure inhibiting two Omicron variants against which the structures with only one or two arms are ineffective. The structure of the three-arm nano-synbody bound to the Omicron variant spike trimer was solved by negative-stain transmission electron microscopy reconstruction, and shows the protein-DNA nanostructure with all three arms attached to the RBD domains, confirming the intended trivalent attachment. The ability to tune the size and shape of the nano-synbody, as well as its potential ability to attach two or more different binding ligands, will enable the high-affinity targeting of a range of proteins not possible with traditional antibodies.

Generative and interpretable machine learning for aptamer design and analysis of in vitro sequence selection.

Selection protocols such as SELEX, where molecules are selected over multiple rounds for their ability to bind to a target of interest, are popular methods for obtaining binders for diagnostic and therapeutic purposes. We show that Restricted Boltzmann Machines (RBMs), an unsupervised two-layer neural network architecture, can successfully be trained on sequence ensembles from single rounds of SELEX experiments for thrombin aptamers. RBMs assign scores to sequences that can be directly related to their fitnesses estimated through experimental enrichment ratios. Hence, RBMs trained from sequence data at a given round can be used to predict the effects of selection at later rounds. Moreover, the parameters of the trained RBMs are interpretable and identify functional features contributing most to sequence fitness. To exploit the generative capabilities of RBMs, we introduce two different training protocols: one taking into account sequence counts, capable of identifying the few best binders, and another based on unique sequences only, generating more diverse binders. We then use RBMs model to generate novel aptamers with putative disruptive mutations or good binding properties, and validate the generated sequences with gel shift assay experiments. Finally, we compare the RBM's performance with different supervised learning approaches that include random forests and several deep neural network architectures.

Coarse-Grained Simulations for the Characterization and Optimization of Hybrid Protein-DNA Nanostructures.

We present here the combination of experimental and computational modeling tools for the design and characterization of protein-DNA hybrid nanostructures. Our work incorporates several features in the design of these nanostructures: (1) modeling of the protein-DNA linker identity and length; (2) optimizing the design of protein-DNA cages to account for mechanical stresses; (3) probing the incorporation efficiency of protein-DNA conjugates into DNA nanostructures. The modeling tools were experimentally validated using structural characterization methods like cryo-TEM and AFM. Our method can be used for fitting low-resolution electron density maps when structural insights cannot be deciphered from experiments, as well as enable in-silico validation of nanostructured systems before their experimental realization. These tools will facilitate the design of complex hybrid protein-DNA nanostructures that seamlessly integrate the two different biomolecules.

Design and simulation of DNA, RNA and hybrid protein-nucleic acid nanostructures with oxView.

Molecular simulation has become an integral part of the DNA/RNA nanotechnology research pipeline. In particular, understanding the dynamics of structures and single-molecule events has improved the precision of nanoscaffolds and diagnostic tools. Here we present oxView, a design tool for visualization, design, editing and analysis of simulations of DNA, RNA and nucleic acid-protein nanostructures. oxView provides an accessible software platform for designing novel structures, tweaking existing designs, preparing them for simulation in the oxDNA/RNA molecular simulation engine and creating visualizations of simulation results. In several examples, we present procedures for using the tool, including its advanced features that couple the design capabilities with a coarse-grained simulation engine and scripting interface that can programmatically edit structures and facilitate design of complex structures from multiple substructures. These procedures provide a practical basis from which researchers, including experimentalists with limited computational experience, can integrate simulation and 3D visualization into their existing research programs.

Coarse-grained nucleic acid-protein model for hybrid nanotechnology.

The emerging field of hybrid DNA-protein nanotechnology brings with it the potential for many novel materials which combine the addressability of DNA nanotechnology with the versatility of protein interactions. However, the design and computational study of these hybrid structures is difficult due to the system sizes involved. To aid in the design and in silico analysis process, we introduce here a coarse-grained DNA/RNA-protein model that extends the oxDNA/oxRNA models of DNA/RNA with a coarse-grained model of proteins based on an anisotropic network model representation. Fully equipped with analysis scripts and visualization, our model aims to facilitate hybrid nanomaterial design towards eventual experimental realization, as well as enabling study of biological complexes. We further demonstrate its usage by simulating DNA-protein nanocage, DNA wrapped around histones, and a nascent RNA in polymerase.