Acta Crystallographica. Section D, Structural Biology最新文献

Despite the parallels between problems in computer vision and cryo-electron microscopy (cryo-EM), many state-of-the-art approaches from computer vision have yet to be adapted for cryo-EM. Within the computer-vision research community, implicits such as neural radiance fields (NeRFs) have enabled the detailed reconstruction of 3D objects from few images at different camera-viewing angles. While other neural implicits, specifically density fields, have been used to map conformational heterogeneity from noisy cryo-EM projection images, most approaches represent volume with an implicit function in Fourier space, which has disadvantages compared with solving the problem in real space, complicating, for instance, masking, constraining physics or geometry, and assessing local resolution. In this work, we build on a recent development in neural implicits, a multi-resolution hash-encoding framework called instant-NGP, that we use to represent the scalar volume directly in real space and apply it to the cryo-EM density-map reconstruction problem (InstaMap). We demonstrate that for both synthetic and real data, InstaMap for homogeneous reconstruction achieves higher resolution at shorter training stages than five other real-spaced representations. We propose a solution to noise overfitting, demonstrate that InstaMap is both lightweight and fast to train, implement masking from a user-provided input mask and extend it to molecular-shape heterogeneity via bending space using a per-image vector field.

A global analysis of protein crystal structures in the Protein Data Bank (PDB) using a newly developed computational approach reveals many pairs with (nearly) identical main-chain coordinates. Such cases are identified and analyzed, showing that duplication is possible since the PDB does not currently have tools or mechanisms that would detect potentially duplicate submissions. Some duplicated entries represent modeling efforts of ligand binding that masquerade as experimentally determined structures. We propose that duplicate entries should either be obsoleted by the PDB or, as a minimum, marked with a clear `CAVEAT' record that would alert potential users to the presence of such problems. We also suggest that using a tool for verifying the uniqueness of the deposited structure, such as that presented in this work, should become part of the routine validation procedure for new depositions.

Green-to-red photoconvertible fluorescent proteins (PCFPs) serve as key players in single-molecule localization super-resolution imaging. As an early engineered variant, mEos3.2 has limited applications, mostly due to its slow maturation rate. The recent advent of a novel variant, pcStar, obtained by the simple mutation of only three amino acids (D28E/L93M/N166G) in mEos3.2, exhibits significantly accelerated maturation and enhanced fluorescent brightness. This improvement represents an important advance in the field of biofluorescence by enabling early detection with reliable signals, essential for labelling dynamic biological processes. However, the mechanism underlying the significant improvement in fluorescent performance from mEos3.2 to pcStar remains elusive, preventing the rational design of more robust variants through mutagenesis. In this study, we determined the crystal structures of mEos3.2 and pcStar in their green states at atomic resolution and performed molecular-dynamics simulations to reveal significant divergences between the two proteins. Our structural and computational analyses revealed crucial features that are distinctively present in pcStar, including the presence of an extra solvent molecule, high conformational stability and enhanced interactions of the chromophore with its surroundings, tighter tertiary-structure packing and dynamic central-helical deformation. Resulting from the triple mutations, all of these structural features are likely to establish a mechanistic link to the greatly improved fluorescent performance of pcStar. The data described here not only provide a good example illustrating how distant amino-acid substitutions can affect the structure and bioactivity of a protein, but also give rise to strategic considerations for the future engineering of more widely applicable PCFPs.

The contractile machinery of muscle, especially that of skeletal muscle, has a very regular array of contractile protein filaments, and gives rise to a complex and informative diffraction pattern when irradiated with X-rays. However, analyzing these diffraction patterns is often challenging because (i) only rotationally averaged diffraction patterns can be obtained, resulting in a substantial loss of information, and (ii) the contractile machinery contains two different sets of protein filaments (actin and myosin) with different helical symmetries. The reflections originating from them often overlap. These problems may be solved if the real-space 3D structure of the contractile machinery is directly calculated from the diffraction pattern. Here, we demonstrate that by using the conventional phase-retrieval algorithm (hybrid input-output), the real-space 3D structure of the contractile machinery can be effectively restored from a single rotationally averaged 2D diffraction pattern. In this calculation, we used an in silico model of insect flight muscle, which is known for its highly regular structure. We also extended this technique to an experimentally recorded muscle diffraction pattern.

With the advent of next-generation modelling methods, such as AlphaFold2, structural biologists are increasingly using predicted structures to obtain structure solutions via molecular replacement (MR) or model fitting in single-particle cryogenic sample electron microscopy (cryoEM). Differences between the domain-domain orientations represented in a predicted model and a crystal structure are often a key limitation when using predicted models. Slice'N'Dice is a software package designed to address this issue by first slicing models into distinct structural units and then automatically placing the slices using either Phaser, MOLREP or PowerFit. The slicing step can use the AlphaFold predicted aligned error (PAE) or can operate via a variety of C^α-atom-based clustering algorithms, extending the applicability to structures of any origin. The number of splits can either be selected by the user or determined automatically. Slice'N'Dice is available for both MR and automated map fitting in the CCP4 and CCP-EM software suites.

Resolving continuous conformational heterogeneity in single-particle cryo-electron microscopy (cryo-EM) is a field in which new methods are now emerging regularly. Methods range from traditional statistical techniques to state-of-the-art neural network approaches. Such ongoing efforts continue to enhance the ability to explore and understand the continuous conformational variations in cryo-EM data. One of the first methods was the manifold embedding approach or ManifoldEM. However, comparing it with more recent methods has been challenging due to software availability and usability issues. In this work, we introduce a modern Python implementation that is user-friendly, orders of magnitude faster than its previous versions and designed with a developer-ready environment. This implementation allows a more thorough evaluation of the strengths and limitations of methods addressing continuous conformational heterogeneity in cryo-EM, paving the way for further community-driven improvements.

Microorganisms are known to secrete copious amounts of extracellular polymeric substances (EPS) that form complex matrices around the cells to shield them against external stresses, to maintain structural integrity and to influence their environment. Many microorganisms also secrete enzymes that are capable of remodelling or degrading EPS in response to various environmental cues. One key enzyme class is the poly-β-1,6-linked N-acetyl-D-glucosamine (PNAG)-degrading glycoside hydrolases, of which the canonical member is dispersin B (DspB) from CAZy family GH20. We sought to test the hypothesis that PNAG-degrading enzymes would be present across family GH20, resulting in expansion of the sequence and structural space and thus the availability of PNAGases. Phylogenetic analysis revealed that several microorganisms contain potential DspB-like enzymes. Six of these were expressed and characterized, and four crystal structures were determined (two of which were in complex with the established GH20 inhibitor 6-acetamido-6-deoxy-castanospermine and one with a bespoke disaccharide β-1,6-linked thiazoline inhibitor). One enzyme expressed rather poorly, which restricted crystal screening and did not allow activity measurements. Using synthetic PNAG oligomers and MALDI-TOF analysis, two of the five enzymes tested showed preferential endo hydrolytic activity. Their sequences, having only 26% identity to the pioneer enzyme DspB, highlight the considerable array of previously unconsidered dispersins in nature, greatly expanding the range of potential dispersin backbones available for societal application and engineering.

The structure of the N-terminal actin-binding domain of human dystrophin was determined at 1.94 Å resolution. Each chain in the asymmetric unit exists in a `closed' conformation, with the first and second calponin homology (CH) domains directly interacting via a 2500.6 Å<sup>2</sup> interface. The positioning of the individual CH domains is comparable to the domain-swapped dimer seen in previous human dystrophin and utrophin actin-binding domain 1 structures. The CH1 domain is highly similar to the actin-bound utrophin structure and structural homology suggests that the `closed' single-chain conformation opens during actin binding to mitigate steric clashes between CH2 and actin.

Predicting the 3D structure of RNA is a significant challenge despite ongoing advancements in the field. Although AlphaFold has successfully addressed this problem for proteins, RNA structure prediction raises difficulties due to the fundamental differences between proteins and RNA, which hinder its direct adaptation. The latest release of AlphaFold, AlphaFold3, has broadened its scope to include multiple different molecules such as DNA, ligands and RNA. While the AlphaFold3 article discussed the results for the last CASP-RNA data set, the scope of its performance and the limitations for RNA are unclear. In this article, we provide a comprehensive analysis of the performance of AlphaFold3 in the prediction of 3D structures of RNA. Through an extensive benchmark over five different test sets, we discuss the performance and limitations of AlphaFold3. We also compare its performance with ten existing state-of-the-art ab initio, template-based and deep-learning approaches. Our results are freely available on the EvryRNA platform at https://evryrna.ibisc.univ-evry.fr/evryrna/alphafold3/.

Stephen Harrop is remembered.