Confronting heterogeneity in cryogenic electron microscopy data: Innovative strategies and future perspectives with data-driven methods.

The surge in the influx of data from cryogenic electron microscopy (cryo-EM) experiments has intensified the demand for robust algorithms capable of autonomously managing structurally heterogeneous datasets. This presents a wealth of exciting opportunities from a data science viewpoint, inspiring the development of numerous innovative, application-specific methods, many of which leverage contemporary data-driven techniques. However, addressing the challenges posed by heterogeneous datasets remains a paramount yet unresolved issue in the field. Here, we explore the subtleties of this challenge and the array of strategies devised to confront it. We pinpoint the shortcomings of existing methodologies and deliberate on prospective avenues for improvement. Specifically, our discussion focuses on strategies to mitigate model overfitting and manage data noise, as well as the effects of constraints, priors, and invariances on the optimization process.

Ghostbuster: A phase retrieval diffraction tomography algorithm for cryo-EM.

Ewald sphere curvature correction, which extends beyond the projection approximation, stretches the shallow depth of field in cryo-EM reconstructions of thick particles. Here we show that even for previously assumed thin particles, reconstruction artifacts which we refer to as ghosts can appear. By retrieving the lost phases of the electron exitwaves and accounting for the first Born approximation scattering within the particle, we show that these ghosts can be effectively eliminated. Our simulations demonstrate how such ghostbusting can improve reconstructions as compared to existing state-of-the-art software. Like ptychographic cryo-EM, our Ghostbuster algorithm uses phase retrieval to improve reconstructions, but unlike the former, we do not need to modify the existing data acquisition pipelines.

Automated model building and protein identification in cryo-EM maps.

Interpreting electron cryo-microscopy (cryo-EM) maps with atomic models requires high levels of expertise and labour-intensive manual intervention in three-dimensional computer graphics programs1,2. Here we present ModelAngelo, a machine-learning approach for automated atomic model building in cryo-EM maps. By combining information from the cryo-EM map with information from protein sequence and structure in a single graph neural network, ModelAngelo builds atomic models for proteins that are of similar quality to those generated by human experts. For nucleotides, ModelAngelo builds backbones with similar accuracy to those built by humans. By using its predicted amino acid probabilities for each residue in hidden Markov model sequence searches, ModelAngelo outperforms human experts in the identification of proteins with unknown sequences. ModelAngelo will therefore remove bottlenecks and increase objectivity in cryo-EM structure determination.

Disease-specific tau filaments assemble via polymorphic intermediates.

Intermediate species in the assembly of amyloid filaments are believed to play a central role in neurodegenerative diseases and may constitute important targets for therapeutic intervention1,2. However, structural information about intermediate species has been scarce and the molecular mechanisms by which amyloids assemble remain largely unknown. Here we use time-resolved cryogenic electron microscopy to study the in vitro assembly of recombinant truncated tau (amino acid residues 297-391) into paired helical filaments of Alzheimer's disease or into filaments of chronic traumatic encephalopathy3. We report the formation of a shared first intermediate amyloid filament, with an ordered core comprising residues 302-316. Nuclear magnetic resonance indicates that the same residues adopt rigid, ß-strand-like conformations in monomeric tau. At later time points, the first intermediate amyloid disappears and we observe many different intermediate amyloid filaments, with structures that depend on the reaction conditions. At the end of both assembly reactions, most intermediate amyloids disappear and filaments with the same ordered cores as those from human brains remain. Our results provide structural insights into the processes of primary and secondary nucleation of amyloid assembly, with implications for the design of new therapies.

Automated model building and protein identification in cryo-EM maps.

Interpreting electron cryo-microscopy (cryo-EM) maps with atomic models requires high levels of expertise and labour-intensive manual intervention. We present ModelAngelo, a machine-learning approach for automated atomic model building in cryo-EM maps. By combining information from the cryo-EM map with information from protein sequence and structure in a single graph neural network, ModelAngelo builds atomic models for proteins that are of similar quality as those generated by human experts. For nucleotides, ModelAngelo builds backbones with similar accuracy as humans. By using its predicted amino acid probabilities for each residue in hidden Markov model sequence searches, ModelAngelo outperforms human experts in the identification of proteins with unknown sequences. ModelAngelo will thus remove bottlenecks and increase objectivity in cryo-EM structure determination.

New tools for automated cryo-EM single-particle analysis in RELION-4.0.

We describe new tools for the processing of electron cryo-microscopy (cryo-EM) images in the fourth major release of the RELION software. In particular, we introduce VDAM, a variable-metric gradient descent algorithm with adaptive moments estimation, for image refinement; a convolutional neural network for unsupervised selection of 2D classes; and a flexible framework for the design and execution of multiple jobs in pre-defined workflows. In addition, we present a stand-alone utility called MDCatch that links the execution of jobs within this framework with metadata gathering during microscope data acquisition. The new tools are aimed at providing fast and robust procedures for unsupervised cryo-EM structure determination, with potential applications for on-the-fly processing and the development of flexible, high-throughput structure determination pipelines. We illustrate their potential on 12 publicly available cryo-EM data sets.

Exploiting prior knowledge about biological macromolecules in cryo-EM structure determination.

Three-dimensional reconstruction of the electron-scattering potential of biological macromolecules from electron cryo-microscopy (cryo-EM) projection images is an ill-posed problem. The most popular cryo-EM software solutions to date rely on a regularization approach that is based on the prior assumption that the scattering potential varies smoothly over three-dimensional space. Although this approach has been hugely successful in recent years, the amount of prior knowledge that it exploits compares unfavorably with the knowledge about biological structures that has been accumulated over decades of research in structural biology. Here, a regularization framework for cryo-EM structure determination is presented that exploits prior knowledge about biological structures through a convolutional neural network that is trained on known macromolecular structures. This neural network is inserted into the iterative cryo-EM structure-determination process through an approach that is inspired by regularization by denoising. It is shown that the new regularization approach yields better reconstructions than the current state of the art for simulated data, and options to extend this work for application to experimental cryo-EM data are discussed.

Structure of the SARS-CoV-2 RNA-dependent RNA polymerase in the presence of favipiravir-RTP.

The RNA polymerase inhibitor favipiravir is currently in clinical trials as a treatment for infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), despite limited information about the molecular basis for its activity. Here we report the structure of favipiravir ribonucleoside triphosphate (favipiravir-RTP) in complex with the SARS-CoV-2 RNA-dependent RNA polymerase (RdRp) bound to a template:primer RNA duplex, determined by electron cryomicroscopy (cryoEM) to a resolution of 2.5 Å. The structure shows clear evidence for the inhibitor at the catalytic site of the enzyme, and resolves the conformation of key side chains and ions surrounding the binding pocket. Polymerase activity assays indicate that the inhibitor is weakly incorporated into the RNA primer strand, and suppresses RNA replication in the presence of natural nucleotides. The structure reveals an unusual, nonproductive binding mode of favipiravir-RTP at the catalytic site of SARS-CoV-2 RdRp, which explains its low rate of incorporation into the RNA primer strand. Together, these findings inform current and future efforts to develop polymerase inhibitors for SARS coronaviruses.

New tools for automated high-resolution cryo-EM structure determination in RELION-3.

Here, we describe the third major release of RELION. CPU-based vector acceleration has been added in addition to GPU support, which provides flexibility in use of resources and avoids memory limitations. Reference-free autopicking with Laplacian-of-Gaussian filtering and execution of jobs from python allows non-interactive processing during acquisition, including 2D-classification, de novo model generation and 3D-classification. Per-particle refinement of CTF parameters and correction of estimated beam tilt provides higher resolution reconstructions when particles are at different heights in the ice, and/or coma-free alignment has not been optimal. Ewald sphere curvature correction improves resolution for large particles. We illustrate these developments with publicly available data sets: together with a Bayesian approach to beam-induced motion correction it leads to resolution improvements of 0.2-0.7 Å compared to previous RELION versions.

Publisher Correction: Structure of the chloroplast ribosome with chl-RRF and hibernation-promoting factor.

In the version of this Article originally published, the name of co-author Annemarie Perez Boerema was coded wrongly, resulting in it being incorrect when exported to citation databases. This has been corrected, though no visible changes will be apparent.

Characterisation of molecular motions in cryo-EM single-particle data by multi-body refinement in RELION.

Macromolecular complexes that exhibit continuous forms of structural flexibility pose a challenge for many existing tools in cryo-EM single-particle analysis. We describe a new tool, called multi-body refinement, which models flexible complexes as a user-defined number of rigid bodies that move independently from each other. Using separate focused refinements with iteratively improved partial signal subtraction, the new tool generates improved reconstructions for each of the defined bodies in a fully automated manner. Moreover, using principal component analysis on the relative orientations of the bodies over all particle images in the data set, we generate movies that describe the most important motions in the data. Our results on two test cases, a cytoplasmic ribosome from Plasmodium falciparum, and the spliceosomal B-complex from yeast, illustrate how multi-body refinement can be useful to gain unique insights into the structure and dynamics of large and flexible macromolecular complexes.

Structure of the chloroplast ribosome with chl-RRF and hibernation-promoting factor.

Oxygenic photosynthesis produces oxygen and builds a variety of organic compounds, changing the chemistry of the air, the sea and fuelling the food chain on our planet. The photochemical reactions underpinning this process in plants take place in the chloroplast. Chloroplasts evolved ~1.2 billion years ago from an engulfed primordial diazotrophic cyanobacterium, and chlororibosomes are responsible for synthesis of the core proteins driving photochemical reactions. Chlororibosomal activity is spatiotemporally coupled to the synthesis and incorporation of functionally essential co-factors, implying the presence of chloroplast-specific regulatory mechanisms and structural adaptation of the chlororibosome1,2. Despite recent structural information3-6, some of these aspects remained elusive. To provide new insights into the structural specialities and evolution, we report a comprehensive analysis of the 2.9-3.1 Å resolution electron cryo-microscopy structure of the spinach chlororibosome in complex with its recycling factor and hibernation-promoting factor. The model reveals a prominent channel extending from the exit tunnel to the chlororibosome exterior, structural re-arrangements that lead to increased surface area for translocon binding, and experimental evidence for parallel and convergent evolution of chloro- and mitoribosomes.

Cryo-EM reconstruction of the chlororibosome to 3.2†Šresolution within 24†h.

The introduction of direct detectors and the automation of data collection in cryo-EM have led to a surge in data, creating new opportunities for advancing computational processing. In particular, on-the-fly workflows that connect data collection with three-dimensional reconstruction would be valuable for more efficient use of cryo-EM and its application as a sample-screening tool. Here, accelerated on-the-fly analysis is reported with optimized organization of the data-processing tools, image acquisition and particle alignment that make it possible to reconstruct the three-dimensional density of the 70S chlororibosome to 3.2 Šresolution within 24 h of tissue harvesting. It is also shown that it is possible to achieve even faster processing at comparable quality by imposing some limits to data use, as illustrated by a 3.7 Šresolution map that was obtained in only 80 min on a desktop computer. These on-the-fly methods can be employed as an assessment of data quality from small samples and extended to high-throughput approaches.

Structures of the human mitochondrial ribosome in native states of assembly.

Mammalian mitochondrial ribosomes (mitoribosomes) have less rRNA content and 36 additional proteins compared with the evolutionarily related bacterial ribosome. These differences make the assembly of mitoribosomes more complex than the assembly of bacterial ribosomes, but the molecular details of mitoribosomal biogenesis remain elusive. Here, we report the structures of two late-stage assembly intermediates of the human mitoribosomal large subunit (mt-LSU) isolated from a native pool within a human cell line and solved by cryo-EM to ∼3-Šresolution. Comparison of the structures reveals insights into the timing of rRNA folding and protein incorporation during the final steps of ribosomal maturation and the evolutionary adaptations that are required to preserve biogenesis after the structural diversification of mitoribosomes. Furthermore, the structures redefine the ribosome silencing factor (RsfS) family as multifunctional biogenesis factors and identify two new assembly factors (L0R8F8 and mt-ACP) not previously implicated in mitoribosomal biogenesis.

Accelerated cryo-EM structure determination with parallelisation using GPUs in RELION-2.

By reaching near-atomic resolution for a wide range of specimens, single-particle cryo-EM structure determination is transforming structural biology. However, the necessary calculations come at large computational costs, which has introduced a bottleneck that is currently limiting throughput and the development of new methods. Here, we present an implementation of the RELION image processing software that uses graphics processors (GPUs) to address the most computationally intensive steps of its cryo-EM structure determination workflow. Both image classification and high-resolution refinement have been accelerated more than an order-of-magnitude, and template-based particle selection has been accelerated well over two orders-of-magnitude on desktop hardware. Memory requirements on GPUs have been reduced to fit widely available hardware, and we show that the use of single precision arithmetic does not adversely affect results. This enables high-resolution cryo-EM structure determination in a matter of days on a single workstation.

SAXS-Guided Metadynamics.

The small-angle X-ray scattering (SAXS) methodology enables structural characterization of biological macromolecules in solution. However, because SAXS provides low-dimensional information, several potential structural configurations can reproduce the experimental scattering profile, which severely complicates the structural refinement process. Here, we present a bias-exchange metadynamics refinement protocol that incorporates SAXS data as collective variables and therefore tags all possible configurations with their corresponding free energies, which allows identification of a unique structural solution. The method has been implemented in PLUMED and combined with the GROMACS simulation package, and as a proof of principle, we explore the Trp-cage protein folding landscape.