bims-lances Biomed News
on Landscapes from Cryo-EM and Simulations
Issue of 2025–03–09
four papers selected by
James M. Krieger, National Centre for Biotechnology



  1. bioRxiv. 2025 Feb 21. pii: 2025.02.18.638828. [Epub ahead of print]
      Advances in machine learning have transformed structural biology, enabling swift and accurate prediction of protein structure from sequence. However, challenges persist in capturing sidechain packing, condition-dependent conformational dynamics, and biomolecular interactions, primarily due to scarcity of high-quality training data. Emerging techniques, including cryo-electron tomography (cryo-ET) and high-throughput crystallography, promise vast new sources of structural data, but translating raw experimental observations into mechanistically interpretable atomic models remains a key bottleneck. Here, we aim to address these challenges by improving the efficiency of structural analysis through combining experimental measurements with a landmark protein structure prediction method - AlphaFold2. We present an augmentation of AlphaFold2, ROCKET, that refines its predictions using cryo-EM, cryo-ET, and X-ray crystallography data, and demonstrate that this approach captures biologically important structural variation that AlphaFold2 does not. By performing structure optimization in the space of coevolutionary embeddings, rather than Cartesian coordinates, ROCKET automates difficult modeling tasks, such as flips of functional loops and domain rearrangements, that are beyond the scope of current state-of-the-art methods and, in some instances, even manual human modeling. The ability to efficiently sample these barrier-crossing rearrangements unlocks a new horizon for scalable and automated model building. Crucially, ROCKET does not require retraining of AlphaFold2 and is readily adaptable to multimers, ligand-cofolding, and other data modalities. Conversely, our differentiable crystallographic and cryo-EM target functions are capable of augmenting other structure prediction methods. ROCKET thus provides an extensible framework for the integration of experimental observables with biomolecular machine learning.
    DOI:  https://doi.org/10.1101/2025.02.18.638828
  2. Res Sq. 2025 Feb 19. pii: rs.3.rs-5994356. [Epub ahead of print]
      We introduce AlphaFold-NMR, a novel approach to NMR structure determination that reveals previously undetected protein conformational states. Unlike conventional NMR methods which rely on NOE-derived spatial restraints, AlphaFold-NMR combines AI-driven conformational sampling with Bayesian scoring of realistic protein models against NOESY and chemical shift data. This method uncovers alternative conformational states of the enzyme Gaussia luciferase, involving large-scale changes in the lid, binding pockets, and other surface cavities. It also identifies similar yet distinct conformational states of the human tumor suppressor Cyclin-Dependent Kinase 2-Associated Protein 1. These studies demonstrate the potential of AI-based modeling with enhanced sampling to generate diverse structural models, followed by conformer selection and validation with experimental data, as an alternative to traditional restraint-satisfaction protocols for protein NMR structure determination. The AlphaFold-NMR framework enables discovery of conformational heterogeneity and cryptic pockets that conventional NMR analysis methods do not distinguish, providing new insights into protein structure-function relationships.
    DOI:  https://doi.org/10.21203/rs.3.rs-5994356/v1
  3. Biochem Biophys Res Commun. 2025 Feb 27. pii: S0006-291X(25)00272-4. [Epub ahead of print]755 151558
      One of the approaches for treatment of COVID-19 is a use of neutralizing antibodies (nAbs). The study of the mechanisms by which nAbs recognize different strains of SARS-CoV-2 may facilitate the development of new drugs and vaccines against the coronavirus infection. In this work, we present the 3.1 Å resolution cryo-electron microscopy structure of a full-length trimeric spike-protein (S-protein) of the SARS-CoV-2 Alpha (B.1.1.7) variant in complex with the Fab of the REGN10987 nAb. In the complex, two receptor-binding domains (RBDs) of the S-protein were observed in the 'up' state, whereas third RBD was in the 'down' state. This distinguishes the obtained structure from the complex of Delta (B.1.617.2) S-protein with REGN10987-Fab, where only one RBD was in the 'up' state. Probably some of the substituted residues (K478T, A570D, and S982A) located at the interprotomer interfaces are responsible for the greater Alpha S-protein opening upon the REGN10987-Fab binding. The Fab identically binds to the RBD in the both 'up' and 'down' conformations. The RBD-Fab interaction interface was refined to a resolution of 3.6 Å. The antibody binds to the receptor-binding motif (RBM), which prevents the S-protein from the binding to its receptor, angiotensin-converting enzyme 2 (ACE-2). Comparison with the structures of the Wuhan (wild type) and Delta RBD variants in complex with REGN10987-Fab revealed that the N501Y and T478K/L452R mutations presented in the RBM of the Alpha and Delta variants, respectively, do not affect the mode of the RBD-Fab interaction.
    Keywords:  Neutralizing antibodies; RBD; S-Protein; SARS-CoV-2; cryo-EM
    DOI:  https://doi.org/10.1016/j.bbrc.2025.151558
  4. IUCrJ. 2025 Mar 01. 12(Pt 2): 139-140
      Interrogating individual two-dimensional (2D) cryo-EM images for the presence of defined three-dimensional (3D) structures that correspond to previously known (or predicted) macromolecular complexes is very challenging, but offers attractive opportunities for the analysis of large numbers of specimens. The work of Zhang et al. [(2025), IUCrJ, 12, 155-176] represents a significant step forward towards this goal.
    Keywords:  cryo-EM; cryo-electron microscopy; image processing; protein structure; template matching
    DOI:  https://doi.org/10.1107/S2052252525001861