De novo design of pH-responsive self-assembling helical protein filaments.

Biological evolution has led to precise and dynamic nanostructures that reconfigure in response to pH and other environmental conditions. However, designing micrometre-scale protein nanostructures that are environmentally responsive remains a challenge. Here we describe the de novo design of pH-responsive protein filaments built from subunits containing six or nine buried histidine residues that assemble into micrometre-scale, well-ordered fibres at neutral pH. The cryogenic electron microscopy structure of an optimized design is nearly identical to the computational design model for both the subunit internal geometry and the subunit packing into the fibre. Electron, fluorescent and atomic force microscopy characterization reveal a sharp and reversible transition from assembled to disassembled fibres over 0.3 pH units, and rapid fibre disassembly in less than 1 s following a drop in pH. The midpoint of the transition can be tuned by modulating buried histidine-containing hydrogen bond networks. Computational protein design thus provides a route to creating unbound nanomaterials that rapidly respond to small pH changes.

Streptomyces umbrella toxin particles block hyphal growth of competing species.

Streptomyces are a genus of ubiquitous soil bacteria from which the majority of clinically utilized antibiotics derive1. The production of these antibacterial molecules reflects the relentless competition Streptomyces engage in with other bacteria, including other Streptomyces species1,2. Here we show that in addition to small-molecule antibiotics, Streptomyces produce and secrete antibacterial protein complexes that feature a large, degenerate repeat-containing polymorphic toxin protein. A cryo-electron microscopy structure of these particles reveals an extended stalk topped by a ringed crown comprising the toxin repeats scaffolding five lectin-tipped spokes, which led us to name them umbrella particles. Streptomyces coelicolor encodes three umbrella particles with distinct toxin and lectin composition. Notably, supernatant containing these toxins specifically and potently inhibits the growth of select Streptomyces species from among a diverse collection of bacteria screened. For one target, Streptomyces griseus, inhibition relies on a single toxin and that intoxication manifests as rapid cessation of vegetative hyphal growth. Our data show that Streptomyces umbrella particles mediate competition among vegetative mycelia of related species, a function distinct from small-molecule antibiotics, which are produced at the onset of reproductive growth and act broadly3,4. Sequence analyses suggest that this role of umbrella particles extends beyond Streptomyces, as we identified umbrella loci in nearly 1,000 species across Actinobacteria.

Generalized biomolecular modeling and design with RoseTTAFold All-Atom.

Deep-learning methods have revolutionized protein structure prediction and design but are presently limited to protein-only systems. We describe RoseTTAFold All-Atom (RFAA), which combines a residue-based representation of amino acids and DNA bases with an atomic representation of all other groups to model assemblies that contain proteins, nucleic acids, small molecules, metals, and covalent modifications, given their sequences and chemical structures. By fine-tuning on denoising tasks, we developed RFdiffusion All-Atom (RFdiffusionAA), which builds protein structures around small molecules. Starting from random distributions of amino acid residues surrounding target small molecules, we designed and experimentally validated, through crystallography and binding measurements, proteins that bind the cardiac disease therapeutic digoxigenin, the enzymatic cofactor heme, and the light-harvesting molecule bilin.

IHMCIF: An Extension of the PDBx/mmCIF Data Standard for Integrative Structure Determination Methods.

IHMCIF (github.com/ihmwg/IHMCIF) is a data information framework that supports archiving and disseminating macromolecular structures determined by integrative or hybrid modeling (IHM), and making them Findable, Accessible, Interoperable, and Reusable (FAIR). IHMCIF is an extension of the Protein Data Bank Exchange/macromolecular Crystallographic Information Framework (PDBx/mmCIF) that serves as the framework for the Protein Data Bank (PDB) to archive experimentally determined atomic structures of biological macromolecules and their complexes with one another and small molecule ligands (e.g., enzyme cofactors and drugs). IHMCIF serves as the foundational data standard for the PDB-Dev prototype system, developed for archiving and disseminating integrative structures. It utilizes a flexible data representation to describe integrative structures that span multiple spatiotemporal scales and structural states with definitions for restraints from a variety of experimental methods contributing to integrative structural biology. The IHMCIF extension was created with the benefit of considerable community input and recommendations gathered by the Worldwide Protein Data Bank (wwPDB) Task Force for Integrative or Hybrid Methods (wwpdb.org/task/hybrid). Herein, we describe the development of IHMCIF to support evolving methodologies and ongoing advancements in integrative structural biology. Ultimately, IHMCIF will facilitate the unification of PDB-Dev data and tools with the PDB archive so that integrative structures can be archived and disseminated through PDB.

Outcomes of the EMDataResource Cryo-EM Ligand Modeling Challenge.

The EMDataResource Ligand Model Challenge aimed to assess the reliability and reproducibility of modeling ligands bound to protein and protein/nucleic-acid complexes in cryogenic electron microscopy (cryo-EM) maps determined at near-atomic (1.9-2.5 Å) resolution. Three published maps were selected as targets: E. coli beta-galactosidase with inhibitor, SARS-CoV-2 RNA-dependent RNA polymerase with covalently bound nucleotide analog, and SARS-CoV-2 ion channel ORF3a with bound lipid. Sixty-one models were submitted from 17 independent research groups, each with supporting workflow details. We found that (1) the quality of submitted ligand models and surrounding atoms varied, as judged by visual inspection and quantification of local map quality, model-to-map fit, geometry, energetics, and contact scores, and (2) a composite rather than a single score was needed to assess macromolecule+ligand model quality. These observations lead us to recommend best practices for assessing cryo-EM structures of liganded macromolecules reported at near-atomic resolution.

Accurate prediction of protein-nucleic acid complexes using RoseTTAFoldNA.

Protein-RNA and protein-DNA complexes play critical roles in biology. Despite considerable recent advances in protein structure prediction, the prediction of the structures of protein-nucleic acid complexes without homology to known complexes is a largely unsolved problem. Here we extend the RoseTTAFold machine learning protein-structure-prediction approach to additionally predict nucleic acid and protein-nucleic acid complexes. We develop a single trained network, RoseTTAFoldNA, that rapidly produces three-dimensional structure models with confidence estimates for protein-DNA and protein-RNA complexes. Here we show that confident predictions have considerably higher accuracy than current state-of-the-art methods. RoseTTAFoldNA should be broadly useful for modeling the structure of naturally occurring protein-nucleic acid complexes, and for designing sequence-specific RNA and DNA-binding proteins.

Computational design of sequence-specific DNA-binding proteins.

Sequence-specific DNA-binding proteins (DBPs) play critical roles in biology and biotechnology, and there has been considerable interest in the engineering of DBPs with new or altered specificities for genome editing and other applications. While there has been some success in reprogramming naturally occurring DBPs using selection methods, the computational design of new DBPs that recognize arbitrary target sites remains an outstanding challenge. We describe a computational method for the design of small DBPs that recognize specific target sequences through interactions with bases in the major groove, and employ this method in conjunction with experimental screening to generate binders for 5 distinct DNA targets. These binders exhibit specificity closely matching the computational models for the target DNA sequences at as many as 6 base positions and affinities as low as 30-100 nM. The crystal structure of a designed DBP-target site complex is in close agreement with the design model, highlighting the accuracy of the design method. The designed DBPs function in both Escherichia coli and mammalian cells to repress and activate transcription of neighboring genes. Our method is a substantial step towards a general route to small and hence readily deliverable sequence-specific DBPs for gene regulation and editing.

De novo design of highly selective miniprotein inhibitors of integrins αvß6 and αvß8.

The RGD (Arg-Gly-Asp)-binding integrins αvß6 and αvß8 are clinically validated cancer and fibrosis targets of considerable therapeutic importance. Compounds that can discriminate between homologous αvß6 and αvß8 and other RGD integrins, stabilize specific conformational states, and have high thermal stability could have considerable therapeutic utility. Existing small molecule and antibody inhibitors do not have all these properties, and hence new approaches are needed. Here we describe a generalized method for computationally designing RGD-containing miniproteins selective for a single RGD integrin heterodimer and conformational state. We design hyperstable, selective αvß6 and αvß8 inhibitors that bind with picomolar affinity. CryoEM structures of the designed inhibitor-integrin complexes are very close to the computational design models, and show that the inhibitors stabilize specific conformational states of the αvß6 and the αvß8 integrins. In a lung fibrosis mouse model, the αvß6 inhibitor potently reduced fibrotic burden and improved overall lung mechanics, demonstrating the therapeutic potential of de novo designed integrin binding proteins with high selectivity.

De novo design of highly selective miniprotein inhibitors of integrins αvß6 and αvß8.

The RGD (Arg-Gly-Asp)-binding integrins αvß6 and αvß8 are clinically validated cancer and fibrosis targets of considerable therapeutic importance. Compounds that can discriminate between the two closely related integrin proteins and other RGD integrins, stabilize specific conformational states, and have sufficient stability enabling tissue restricted administration could have considerable therapeutic utility. Existing small molecules and antibody inhibitors do not have all of these properties, and hence there is a need for new approaches. Here we describe a method for computationally designing hyperstable RGD-containing miniproteins that are highly selective for a single RGD integrin heterodimer and conformational state, and use this strategy to design inhibitors of αvß6 and αvß8 with high selectivity. The αvß6 and αvß8 inhibitors have picomolar affinities for their targets, and >1000-fold selectivity over other RGD integrins. CryoEM structures are within 0.6-0.7Å root-mean-square deviation (RMSD) to the computational design models; the designed αvß6 inhibitor and native ligand stabilize the open conformation in contrast to the therapeutic anti-αvß6 antibody BG00011 that stabilizes the bent-closed conformation and caused on-target toxicity in patients with lung fibrosis, and the αvß8 inhibitor maintains the constitutively fixed extended-closed αvß8 conformation. In a mouse model of bleomycin-induced lung fibrosis, the αvß6 inhibitor potently reduced fibrotic burden and improved overall lung mechanics when delivered via oropharyngeal administration mimicking inhalation, demonstrating the therapeutic potential of de novo designed integrin binding proteins with high selectivity.

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.

Hexasome-INO80 complex reveals structural basis of noncanonical nucleosome remodeling.

Loss of H2A-H2B histone dimers is a hallmark of actively transcribed genes, but how the cellular machinery functions in the context of noncanonical nucleosomal particles remains largely elusive. In this work, we report the structural mechanism for adenosine 5'-triphosphate-dependent chromatin remodeling of hexasomes by the INO80 complex. We show how INO80 recognizes noncanonical DNA and histone features of hexasomes that emerge from the loss of H2A-H2B. A large structural rearrangement switches the catalytic core of INO80 into a distinct, spin-rotated mode of remodeling while its nuclear actin module remains tethered to long stretches of unwrapped linker DNA. Direct sensing of an exposed H3-H4 histone interface activates INO80, independently of the H2A-H2B acidic patch. Our findings reveal how the loss of H2A-H2B grants remodelers access to a different, yet unexplored layer of energy-driven chromatin regulation.

Improving de novo protein binder design with deep learning.

Recently it has become possible to de novo design high affinity protein binding proteins from target structural information alone. There is, however, considerable room for improvement as the overall design success rate is low. Here, we explore the augmentation of energy-based protein binder design using deep learning. We find that using AlphaFold2 or RoseTTAFold to assess the probability that a designed sequence adopts the designed monomer structure, and the probability that this structure binds the target as designed, increases design success rates nearly 10-fold. We find further that sequence design using ProteinMPNN rather than Rosetta considerably increases computational efficiency.

Residue-level error detection in cryoelectron microscopy models.

Building accurate protein models into moderate resolution (3-5 Å) cryoelectron microscopy (cryo-EM) maps is challenging and error prone. We have developed MEDIC (Model Error Detection in Cryo-EM), a robust statistical model that identifies local backbone errors in protein structures built into cryo-EM maps by combining local fit-to-density with deep-learning-derived structural information. MEDIC is validated on a set of 28 structures that were subsequently solved to higher resolutions, where we identify the differences between low- and high-resolution structures with 68% precision and 60% recall. We additionally use this model to fix over 100 errors in 12 deposited structures and to identify errors in 4 refined AlphaFold predictions with 80% precision and 60% recall. As modelers more frequently use deep learning predictions as a starting point for refinement and rebuilding, MEDIC's ability to handle errors in structures derived from hand-building and machine learning methods makes it a powerful tool for structural biologists.

Cyclic peptide structure prediction and design using AlphaFold.

Deep learning networks offer considerable opportunities for accurate structure prediction and design of biomolecules. While cyclic peptides have gained significant traction as a therapeutic modality, developing deep learning methods for designing such peptides has been slow, mostly due to the small number of available structures for molecules in this size range. Here, we report approaches to modify the AlphaFold network for accurate structure prediction and design of cyclic peptides. Our results show this approach can accurately predict the structures of native cyclic peptides from a single sequence, with 36 out of 49 cases predicted with high confidence (pLDDT > 0.85) matching the native structure with root mean squared deviation (RMSD) less than 1.5 Å. Further extending our approach, we describe computational methods for designing sequences of peptide backbones generated by other backbone sampling methods and for de novo design of new macrocyclic peptides. We extensively sampled the structural diversity of cyclic peptides between 7-13 amino acids, and identified around 10,000 unique design candidates predicted to fold into the designed structures with high confidence. X-ray crystal structures for seven sequences with diverse sizes and structures designed by our approach match very closely with the design models (root mean squared deviation < 1.0 Å), highlighting the atomic level accuracy in our approach. The computational methods and scaffolds developed here provide the basis for custom-designing peptides for targeted therapeutic applications.

Automatic and accurate ligand structure determination guided by cryo-electron microscopy maps.

Advances in cryo-electron microscopy (cryoEM) and deep-learning guided protein structure prediction have expedited structural studies of protein complexes. However, methods for accurately determining ligand conformations are lacking. In this manuscript, we develop EMERALD, a tool for automatically determining ligand structures guided by medium-resolution cryoEM density. We show this method is robust at predicting ligands along with surrounding side chains in maps as low as 4.5 Å local resolution. Combining this with a measure of placement confidence and running on all protein/ligand structures in the EMDB, we show that 57% of ligands replicate the deposited model, 16% confidently find alternate conformations, 22% have ambiguous density where multiple conformations might be present, and 5% are incorrectly placed. For five cases where our approach finds an alternate conformation with high confidence, high-resolution crystal structures validate our placement. EMERALD and the resulting analysis should prove critical in using cryoEM to solve protein-ligand complexes.

Structural basis underlying specific biochemical activities of non-muscle tropomyosin isoforms.

The actin cytoskeleton is critical for cell migration, morphogenesis, endocytosis, organelle dynamics, and cytokinesis. To support diverse cellular processes, actin filaments form a variety of structures with specific architectures and dynamic properties. Key proteins specifying actin filaments are tropomyosins. Non-muscle cells express several functionally non-redundant tropomyosin isoforms, which differentially control the interactions of other proteins, including myosins and ADF/cofilin, with actin filaments. However, the underlying molecular mechanisms have remained elusive. By determining the cryogenic electron microscopy structures of actin filaments decorated by two functionally distinct non-muscle tropomyosin isoforms, Tpm1.6 and Tpm3.2, we reveal that actin filament conformation remains unaffected upon binding. However, Tpm1.6 and Tpm3.2 follow different paths along the actin filament major groove, providing an explanation for their incapability to co-polymerize on actin filaments. We also elucidate the molecular basis underlying specific roles of Tpm1.6 and Tpm3.2 in myosin II activation and protecting actin filaments from ADF/cofilin-catalyzed severing.

StarMap: a user-friendly workflow for Rosetta-driven molecular structure refinement.

Cryogenic electron microscopy (cryo-EM) data represent density maps of macromolecular systems at atomic or near-atomic resolution. However, building and refining 3D atomic models by using data from cryo-EM maps is not straightforward and requires significant hands-on experience and manual intervention. We recently developed StarMap, an easy-to-use interface between the popular structural display program ChimeraX and Rosetta, a powerful molecular modeling engine. StarMap offers a general approach for refining structural models of biological macromolecules into cryo-EM density maps by combining Monte Carlo sampling with local density-guided optimization, Rosetta-based all-atom refinement and real-space B-factor calculations in a straightforward workflow. StarMap includes options for structural symmetry, local refinements and independent model validation. The overall quality of the refinement and the structure resolution is then assessed via analytical outputs, such as magnification calibration (pixel size calibration) and Fourier shell correlations. Z-scores reported by StarMap provide an easily interpretable indicator of the goodness of fit for each residue and can be plotted to evaluate structural models and improve local residue refinements, as well as to identify flexible regions and potentially functional sites in large macromolecular complexes. The protocol requires general computer skills, without the need for coding expertise, because most parts of the workflow can be operated by clicking tabs within the ChimeraX graphical user interface. Time requirements for the model refinement depend on the size and quality of the input data; however, this step can typically be completed within 1 d. The analytical parts of the workflow are completed within minutes.

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.

AlphaFold accurately predicts distinct conformations based on the oligomeric state of a de novo designed protein.

Using the molecular modeling program Rosetta, we designed a de novo protein, called SEWN0.1, which binds the heterotrimeric G protein Gαq. The design is helical, well-folded, and primarily monomeric in solution at a concentration of 10 µM. However, when we solved the crystal structure of SEWN0.1 at 1.9 Å, we observed a dimer in a conformation incompatible with binding Gαq . Unintentionally, we had designed a protein that adopts alternate conformations depending on its oligomeric state. Recently, there has been tremendous progress in the field of protein structure prediction as new methods in artificial intelligence have been used to predict structures with high accuracy. We were curious if the structure prediction method AlphaFold could predict the structure of SEWN0.1 and if the prediction depended on oligomeric state. When AlphaFold was used to predict the structure of monomeric SEWN0.1, it produced a model that resembles the Rosetta design model and is compatible with binding Gαq , but when used to predict the structure of a dimer, it predicted a conformation that closely resembles the SEWN0.1 crystal structure. AlphaFold's ability to predict multiple conformations for a single protein sequence should be useful for engineering protein switches.