De novo design of protein structure and function with RFdiffusion.

There has been considerable recent progress in designing new proteins using deep-learning methods1-9. Despite this progress, a general deep-learning framework for protein design that enables solution of a wide range of design challenges, including de novo binder design and design of higher-order symmetric architectures, has yet to be described. Diffusion models10,11 have had considerable success in image and language generative modelling but limited success when applied to protein modelling, probably due to the complexity of protein backbone geometry and sequence-structure relationships. Here we show that by fine-tuning the RoseTTAFold structure prediction network on protein structure denoising tasks, we obtain a generative model of protein backbones that achieves outstanding performance on unconditional and topology-constrained protein monomer design, protein binder design, symmetric oligomer design, enzyme active site scaffolding and symmetric motif scaffolding for therapeutic and metal-binding protein design. We demonstrate the power and generality of the method, called RoseTTAFold diffusion (RFdiffusion), by experimentally characterizing the structures and functions of hundreds of designed symmetric assemblies, metal-binding proteins and protein binders. The accuracy of RFdiffusion is confirmed by the cryogenic electron microscopy structure of a designed binder in complex with influenza haemagglutinin that is nearly identical to the design model. In a manner analogous to networks that produce images from user-specified inputs, RFdiffusion enables the design of diverse functional proteins from simple molecular specifications.

A tethered ligand assay to probe SARS-CoV-2:ACE2 interactions.

SignificanceIn the dynamic environment of the airways, where SARS-CoV-2 infections are initiated by binding to human host receptor ACE2, mechanical stability of the viral attachment is a crucial fitness advantage. Using single-molecule force spectroscopy techniques, we mimic the effect of coughing and sneezing, thereby testing the force stability of SARS-CoV-2 RBD:ACE2 interaction under physiological conditions. Our results reveal a higher force stability of SARS-CoV-2 binding to ACE2 compared to SARS-CoV-1, causing a possible fitness advantage. Our assay is sensitive to blocking agents preventing RBD:ACE2 bond formation. It will thus provide a powerful approach to investigate the modes of action of neutralizing antibodies and other agents designed to block RBD binding to ACE2 that are currently developed as potential COVID-19 therapeutics.

Improving Protein Expression, Stability, and Function with ProteinMPNN.

Natural proteins are highly optimized for function but are often difficult to produce at a scale suitable for biotechnological applications due to poor expression in heterologous systems, limited solubility, and sensitivity to temperature. Thus, a general method that improves the physical properties of native proteins while maintaining function could have wide utility for protein-based technologies. Here, we show that the deep neural network ProteinMPNN, together with evolutionary and structural information, provides a route to increasing protein expression, stability, and function. For both myoglobin and tobacco etch virus (TEV) protease, we generated designs with improved expression, elevated melting temperatures, and improved function. For TEV protease, we identified multiple designs with improved catalytic activity as compared to the parent sequence and previously reported TEV variants. Our approach should be broadly useful for improving the expression, stability, and function of biotechnologically important proteins.

Different Vinculin Binding Sites Use the Same Mechanism to Regulate Directional Force Transduction.

Vinculin is a universal adaptor protein that transiently reinforces the mechanical stability of adhesion complexes. It stabilizes mechanical connections that cells establish between the actomyosin cytoskeleton and the extracellular matrix via integrins or to neighboring cells via cadherins, yet little is known regarding its mechanical design. Vinculin binding sites (VBSs) from different nonhomologous actin-binding proteins use conserved helical motifs to associate with the vinculin head domain. We studied the mechanical stability of such complexes by pulling VBS peptides derived from talin, α-actinin, and Shigella IpaA out of the vinculin head domain. Experimental data from atomic force microscopy single-molecule force spectroscopy and steered molecular dynamics (SMD) simulations both revealed greater mechanical stability of the complex for shear-like than for zipper-like pulling configurations. This suggests that reinforcement occurs along preferential force directions, thus stabilizing those cytoskeletal filament architectures that result in shear-like pulling geometries. Large force-induced conformational changes in the vinculin head domain, as well as protein-specific fine-tuning of the VBS sequence, including sequence inversion, allow for an even more nuanced force response.

From genes to protein mechanics on a chip.

Single-molecule force spectroscopy enables mechanical testing of individual proteins, but low experimental throughput limits the ability to screen constructs in parallel. We describe a microfluidic platform for on-chip expression, covalent surface attachment and measurement of single-molecule protein mechanical properties. A dockerin tag on each protein molecule allowed us to perform thousands of pulling cycles using a single cohesin-modified cantilever. The ability to synthesize and mechanically probe protein libraries enables high-throughput mechanical phenotyping.

Energy profile of nanobody-GFP complex under force.

Nanobodies (Nbs)-the smallest known fully functional and naturally occuring antigen-binding fragments-have attracted a lot of attention throughout the last two decades. Exploring their potential beyond the current use requires more detailed characterization of their binding forces as those cannot be directly derived from the binding affinities. Here we used atomic force microscope to measure rupture force of the Nb-green fluorescent protein (GFP) complex in various pulling geometries and derived the energy profile characterizing the interaction along the direction of the pulling force. We found that-despite identical epitopes-the Nb binds stronger (41-56 pN) to enhanced GFP than to wild-type GFP (28-45 pN). Measured forces make the Nb-GFP pair a potent reference for investigating molecular forces in living systems both in and ex vivo.

C-5 Propynyl Modifications Enhance the Mechanical Stability of DNA.

Increased thermal or mechanical stability of DNA duplexes is desired for many applications in nanotechnology or -medicine where DNA is used as a programmable building block. Modifications of pyrimidine bases are known to enhance thermal stability and have the advantage of standard base-pairing and easy integration during chemical DNA synthesis. Through single-molecule force spectroscopy experiments with atomic force microscopy and the molecular force assay we investigated the effect of pyrimidines harboring C-5 propynyl modifications on the mechanical stability of double-stranded DNA. Utilizing these complementary techniques, we show that propynyl bases significantly increase the mechanical stability if the DNA is annealed at high temperature. In contrast, modified DNA complexes formed at room temperature and short incubation times display the same stability as non-modified DNA duplexes.

Single-molecule force stability of the SARS-CoV-2-ACE2 interface in variants-of-concern.

Mutations in SARS-CoV-2 have shown effective evasion of population immunity and increased affinity to the cellular receptor angiotensin-converting enzyme 2 (ACE2). However, in the dynamic environment of the respiratory tract, forces act on the binding partners, which raises the question of whether not only affinity but also force stability of the SARS-CoV-2-ACE2 interaction might be a selection factor for mutations. Using magnetic tweezers, we investigate the impact of amino acid substitutions in variants of concern (Alpha, Beta, Gamma and Delta) and on force-stability and bond kinetic of the receptor-binding domain-ACE2 interface at a single-molecule resolution. We find a higher affinity for all of the variants of concern (>fivefold) compared with the wild type. In contrast, Alpha is the only variant of concern that shows higher force stability (by 17%) compared with the wild type. Using molecular dynamics simulations, we rationalize the mechanistic molecular origins of this increase in force stability. Our study emphasizes the diversity of contributions to the transmissibility of variants and establishes force stability as one of the several factors for fitness. Understanding fitness advantages opens the possibility for the prediction of probable mutations, allowing a rapid adjustment of therapeutics, vaccines and intervention measures.

De novo design of knotted tandem repeat proteins.

De novo protein design methods can create proteins with folds not yet seen in nature. These methods largely focus on optimizing the compatibility between the designed sequence and the intended conformation, without explicit consideration of protein folding pathways. Deeply knotted proteins, whose topologies may introduce substantial barriers to folding, thus represent an interesting test case for protein design. Here we report our attempts to design proteins with trefoil (31) and pentafoil (51) knotted topologies. We extended previously described algorithms for tandem repeat protein design in order to construct deeply knotted backbones and matching designed repeat sequences (N = 3 repeats for the trefoil and N = 5 for the pentafoil). We confirmed the intended conformation for the trefoil design by X ray crystallography, and we report here on this protein's structure, stability, and folding behaviour. The pentafoil design misfolded into an asymmetric structure (despite a 5-fold symmetric sequence); two of the four repeat-repeat units matched the designed backbone while the other two diverged to form local contacts, leading to a trefoil rather than pentafoil knotted topology. Our results also provide insights into the folding of knotted proteins.

Hallucination of closed repeat proteins containing central pockets.

In pseudocyclic proteins, such as TIM barrels, ß barrels, and some helical transmembrane channels, a single subunit is repeated in a cyclic pattern, giving rise to a central cavity that can serve as a pocket for ligand binding or enzymatic activity. Inspired by these proteins, we devised a deep-learning-based approach to broadly exploring the space of closed repeat proteins starting from only a specification of the repeat number and length. Biophysical data for 38 structurally diverse pseudocyclic designs produced in Escherichia coli are consistent with the design models, and the three crystal structures we were able to obtain are very close to the designed structures. Docking studies suggest the diversity of folds and central pockets provide effective starting points for designing small-molecule binders and enzymes.

Scaffolding protein functional sites using deep learning.

The binding and catalytic functions of proteins are generally mediated by a small number of functional residues held in place by the overall protein structure. Here, we describe deep learning approaches for scaffolding such functional sites without needing to prespecify the fold or secondary structure of the scaffold. The first approach, "constrained hallucination," optimizes sequences such that their predicted structures contain the desired functional site. The second approach, "inpainting," starts from the functional site and fills in additional sequence and structure to create a viable protein scaffold in a single forward pass through a specifically trained RoseTTAFold network. We use these two methods to design candidate immunogens, receptor traps, metalloproteins, enzymes, and protein-binding proteins and validate the designs using a combination of in silico and experimental tests.

Extreme mechanical stability in protein complexes.

Recently, non-covalent protein complexes and folds with extreme mechanical stabilities have been discovered. Various extracellular adhesin proteins of gram-positive bacteria exhibit complex rupture forces ranging from 800pN in the case of cellulolytic bacteria to over 2000pN withstood by pathogens adhering to their hosts. Here, we review and assess the mechanics of such systems, and discuss progress, as well as open questions regarding their biological function, and underlying molecular mechanisms - in particular the role of increased interaction lifetimes under mechanical load. These unexpected extreme strengths open an unchartered range of protein mechanics that can now be routinely probed by atomic force microscopy-based single-molecule force spectroscopy.

DNA-free directed assembly in single-molecule cut-and-paste.

Single-molecule cut-and-paste facilitates bottom-up directed assembly of nanoscale biomolecular networks in defined geometries and enables analysis with spatio-temporal resolution. However, arrangement of diverse molecules of interest requires versatile handling systems. The novel DNA-free, genetically encodable scheme described here utilises an orthogonal handling strategy to promote arrangement of enzymes and enzyme networks.

The free energy landscape of retroviral integration.

Retroviral integration, the process of covalently inserting viral DNA into the host genome, is a point of no return in the replication cycle. Yet, strand transfer is intrinsically iso-energetic and it is not clear how efficient integration can be achieved. Here we investigate the dynamics of strand transfer and demonstrate that consecutive nucleoprotein intermediates interacting with a supercoiled target are increasingly stable, resulting in a net forward rate. Multivalent target interactions at discrete auxiliary interfaces render target capture irreversible, while allowing dynamic site selection. Active site binding is transient but rapidly results in strand transfer, which in turn rearranges and stabilizes the intasome in an allosteric manner. We find the resulting strand transfer complex to be mechanically stable and extremely long-lived, suggesting that a resolving agent is required in vivo.

Calcium stabilizes the strongest protein fold.

Staphylococcal pathogens adhere to their human targets with exceptional resilience to mechanical stress, some propagating force to the bacterium via small, Ig-like folds called B domains. We examine the mechanical stability of these folds using atomic force microscopy-based single-molecule force spectroscopy. The force required to unfold a single B domain is larger than 2 nN - the highest mechanostability of a protein to date by a large margin. B domains coordinate three calcium ions, which we identify as crucial for their extreme mechanical strength. When calcium is removed through chelation, unfolding forces drop by a factor of four. Through systematic mutations in the calcium coordination sites we can tune the unfolding forces from over 2 nN to 0.15 nN, and dissect the contribution of each ion to B domain mechanostability. Their extraordinary strength, rapid refolding and calcium-tunable force response make B domains interesting protein design targets.

Molecular mechanism of extreme mechanostability in a pathogen adhesin.

High resilience to mechanical stress is key when pathogens adhere to their target and initiate infection. Using atomic force microscopy-based single-molecule force spectroscopy, we explored the mechanical stability of the prototypical staphylococcal adhesin SdrG, which targets a short peptide from human fibrinogen ß. Steered molecular dynamics simulations revealed, and single-molecule force spectroscopy experiments confirmed, the mechanism by which this complex withstands forces of over 2 nanonewtons, a regime previously associated with the strength of a covalent bond. The target peptide, confined in a screwlike manner in the binding pocket of SdrG, distributes forces mainly toward the peptide backbone through an intricate hydrogen bond network. Thus, these adhesins can attach to their target with exceptionally resilient mechanostability, virtually independent of peptide side chains.

Mechanical Stability of a High-Affinity Toxin Anchor from the Pathogen Clostridium perfringens.

The opportunistic pathogen Clostridium perfringens assembles its toxins and carbohydrate-active enzymes by the high-affinity cohesin-dockerin (Coh-Doc) interaction. Coh-Doc interactions characterized previously have shown considerable resilience toward mechanical stress. Here, we aimed to determine the mechanics of this interaction from C. perfringens in the context of a pathogen. Using atomic force microscopy based single-molecule force spectroscopy (AFM-SMFS) we probed the mechanical properties of the interaction of a dockerin from the µ-toxin with the GH84C X82 cohesin domain of C. perfringens. Most probable complex rupture forces were found to be approximately 60 pN and an estimate of the binding potential width was performed. The dockerin was expressed with its adjacent FIVAR (found in various architectures) domain, whose mechanostability we determined to be very similar to the complex. Additionally, fast refolding of this domain was observed. The Coh-Doc interaction from C. perfringens is the mechanically weakest observed to date. Our results establish the relevant force range of toxin assembly mechanics in pathogenic Clostridia.

Monodisperse measurement of the biotin-streptavidin interaction strength in a well-defined pulling geometry.

The widely used interaction of the homotetramer streptavidin with the small molecule biotin has been intensively studied by force spectroscopy and has become a model system for receptor ligand interaction. However, streptavidin's tetravalency results in diverse force propagation pathways through the different binding interfaces. This multiplicity gives rise to polydisperse force spectroscopy data. Here, we present an engineered monovalent streptavidin tetramer with a single cysteine in its functional subunit that allows for site-specific immobilization of the molecule, orthogonal to biotin binding. Functionality of streptavidin and its binding properties for biotin remain unaffected. We thus created a stable and reliable molecular anchor with a unique high-affinity binding site for biotinylated molecules or nanoparticles, which we expect to be useful for many single-molecule applications. To characterize the mechanical properties of the bond between biotin and our monovalent streptavidin, we performed force spectroscopy experiments using an atomic force microscope. We were able to conduct measurements at the single-molecule level with 1:1-stoichiometry and a well-defined geometry, in which force exclusively propagates through a single subunit of the streptavidin tetramer. For different force loading rates, we obtained narrow force distributions of the bond rupture forces ranging from 200 pN at 1,500 pN/s to 230 pN at 110,000 pN/s. The data are in very good agreement with the standard Bell-Evans model with a single potential barrier at Δx0 = 0.38 nm and a zero-force off-rate koff,0 in the 10-6 s-1 range.

Elastin-like Polypeptide Linkers for Single-Molecule Force Spectroscopy.

Single-molecule force spectroscopy (SMFS) is by now well established as a standard technique in biophysics and mechanobiology. In recent years, the technique has benefitted greatly from new approaches to bioconjugation of proteins to surfaces. Indeed, optimized immobilization strategies for biomolecules and refined purification schemes are being steadily adapted and improved, which in turn has enhanced data quality. In many previously reported SMFS studies, poly(ethylene glycol) (PEG) was used to anchor molecules of interest to surfaces and/or cantilever tips. The limitation, however, is that PEG exhibits a well-known trans-trans-gauche to all-trans transition, which results in marked deviation from standard polymer elasticity models such as the worm-like chain, particularly at elevated forces. As a result, the assignment of unfolding events to protein domains based on their corresponding amino acid chain lengths is significantly obscured. Here, we provide a solution to this problem by implementing unstructured elastin-like polypeptides as linkers to replace PEG. We investigate the suitability of tailored elastin-like polypeptides linkers and perform direct comparisons to PEG, focusing on attributes that are critical for single-molecule force experiments such as linker length, monodispersity, and bioorthogonal conjugation tags. Our results demonstrate that by avoiding the ambiguous elastic response of mixed PEG/peptide systems and instead building the molecular mechanical systems with only a single bond type with uniform elastic properties, we improve data quality and facilitate data analysis and interpretation in force spectroscopy experiments. The use of all-peptide linkers allows alternative approaches for precisely defining elastic properties of proteins linked to surfaces.

Monovalent Strep-Tactin for strong and site-specific tethering in nanospectroscopy.

Strep-Tactin, an engineered form of streptavidin, binds avidly to the genetically encoded peptide Strep-tag II in a manner comparable to streptavidin binding to biotin. These interactions have been used in protein purification and detection applications. However, in single-molecule studies, for example using atomic force microscopy-based single-molecule force spectroscopy (AFM-SMFS), the tetravalency of these systems impedes the measurement of monodispersed data. Here, we introduce a monovalent form of Strep-Tactin that harbours a unique binding site for Strep-tag II and a single cysteine that allows Strep-Tactin to specifically attach to the atomic force microscope cantilever and form a consistent pulling geometry to obtain homogeneous rupture data. Using AFM-SMFS, the mechanical properties of the interaction between Strep-tag II and monovalent Strep-Tactin were characterized. Rupture forces comparable to biotin:streptavidin unbinding were observed. Using titin kinase and green fluorescent protein, we show that monovalent Strep-Tactin is generally applicable to protein unfolding experiments. We expect monovalent Strep-Tactin to be a reliable anchoring tool for a range of single-molecule studies.