Simulating the cellular context in synthetic datasets for cryo-electron tomography.

Cryo-electron tomography (cryo-ET) allows to visualize the cellular context at macromolecular level. To date, the impossibility of obtaining a reliable ground truth is limiting the application of deep learning-based image processing algorithms in this field. As a consequence, there is a growing demand of realistic synthetic datasets for training deep learning algorithms. In addition, besides assisting the acquisition and interpretation of experimental data, synthetic tomograms are used as reference models for cellular organization analysis from cellular tomograms. Current simulators in cryo-ET focus on reproducing distortions from image acquisition and tomogram reconstruction, however, they can not generate many of the low order features present in cellular tomograms. Here we propose several geometric and organization models to simulate low order cellular structures imaged by cryo-ET. Specifically, clusters of any known cytosolic or membrane bound macromolecules, membranes with different geometries as well as different filamentous structures such as microtubules or actin-like networks. Moreover, we use parametrizable stochastic models to generate a high diversity of geometries and organizations to simulate representative and generalized datasets, including very crowded environments like those observed in native cells. These models have been implemented in a multiplatform open-source Python package, including scripts to generate cryo-tomograms with adjustable sizes and resolutions. In addition, these scripts provide also distortion-free density maps besides the ground truth in different file formats for efficient access and advanced visualization. We show that such a realistic synthetic dataset can be readily used to train generalizable deep learning algorithms.

Atg18 oligomer organization in assembled tubes and on lipid membrane scaffolds.

Autophagy-related protein 18 (Atg18) participates in the elongation of early autophagosomal structures in concert with Atg2 and Atg9 complexes. How Atg18 contributes to the structural coordination of Atg2 and Atg9 at the isolation membrane remains to be understood. Here, we determined the cryo-EM structures of Atg18 organized in helical tubes, Atg18 oligomers in solution as well as on lipid membrane scaffolds. The helical assembly is composed of Atg18 tetramers forming a lozenge cylindrical lattice with remarkable structural similarity to the COPII outer coat. When reconstituted with lipid membranes, using subtomogram averaging we determined tilted Atg18 dimer structures bridging two juxtaposed lipid membranes spaced apart by 80 Å. Moreover, lipid reconstitution experiments further delineate the contributions of Atg18's FRRG motif and the amphipathic helical extension in membrane interaction. The observed structural plasticity of Atg18's oligomeric organization and membrane binding properties provide a molecular framework for the positioning of downstream components of the autophagy machinery.

Direct Cryo-ET observation of platelet deformation induced by SARS-CoV-2 spike protein.

SARS-CoV-2 is a novel coronavirus responsible for the COVID-19 pandemic. Its high pathogenicity is due to SARS-CoV-2 spike protein (S protein) contacting host-cell receptors. A critical hallmark of COVID-19 is the occurrence of coagulopathies. Here, we report the direct observation of the interactions between S protein and platelets. Live imaging shows that the S protein triggers platelets to deform dynamically, in some cases, leading to their irreversible activation. Cellular cryo-electron tomography reveals dense decorations of S protein on the platelet surface, inducing filopodia formation. Hypothesizing that S protein binds to filopodia-inducing integrin receptors, we tested the binding to RGD motif-recognizing platelet integrins and find that S protein recognizes integrin αvß3. Our results infer that the stochastic activation of platelets is due to weak interactions of S protein with integrin, which can attribute to the pathogenesis of COVID-19 and the occurrence of rare but severe coagulopathies.

Direct Cryo-ET observation of platelet deformation induced by SARS-CoV-2 Spike protein.

SARS-CoV-2 is a novel coronavirus responsible for the COVID-19 pandemic. Its high pathogenicity is due to SARS-CoV-2 spike protein (S protein) contacting host-cell receptors. A critical hallmark of COVID-19 is the occurrence of coagulopathies. Here, we report the direct observation of the interactions between S protein and platelets. Live imaging showed that the S protein triggers platelets to deform dynamically, in some cases, leading to their irreversible activation. Strikingly, cellular cryo-electron tomography revealed dense decorations of S protein on the platelet surface, inducing filopodia formation. Hypothesizing that S protein binds to filopodia-inducing integrin receptors, we tested the binding to RGD motif-recognizing platelet integrins and found that S protein recognizes integrin α v ß 3 . Our results infer that the stochastic activation of platelets is due to weak interactions of S protein with integrin, which can attribute to the pathogenesis of COVID-19 and the occurrence of rare but severe coagulopathies.

MemBrain: A deep learning-aided pipeline for detection of membrane proteins in Cryo-electron tomograms.

BACKGROUND AND OBJECTIVE: Cryo-electron tomography (cryo-ET) is an imaging technique that enables 3D visualization of the native cellular environment at sub-nanometer resolution, providing unpreceded insights into the molecular organization of cells. However, cryo-electron tomograms suffer from low signal-to-noise ratios and anisotropic resolution, which makes subsequent image analysis challenging. In particular, the efficient detection of membrane-embedded proteins is a problem still lacking satisfactory solutions. METHODS: We present MemBrain - a new deep learning-aided pipeline that automatically detects membrane-bound protein complexes in cryo-electron tomograms. After subvolumes are sampled along a segmented membrane, each subvolume is assigned a score using a convolutional neural network (CNN), and protein positions are extracted by a clustering algorithm. Incorporating rotational subvolume normalization and using a tiny receptive field simplify the task of protein detection and thus facilitate the network training. RESULTS: MemBrain requires only a small quantity of training labels and achieves excellent performance with only a single annotated membrane (F1 score: 0.88). A detailed evaluation shows that our fully trained pipeline outperforms existing classical computer vision-based and CNN-based approaches by a large margin (F1 score: 0.92 vs. max. 0.63). Furthermore, in addition to protein center positions, MemBrain can determine protein orientations, which has not been implemented by any existing CNN-based method to date. We also show that a pre-trained MemBrain program generalizes to tomograms acquired using different cryo-ET methods and depicting different types of cells. CONCLUSIONS: MemBrain is a powerful and annotation-efficient tool for the detection of membrane protein complexes in cryo-ET data, with the potential to be used in a wide range of biological studies. It is generalizable to various kinds of tomograms, making it possible to use pretrained models for different tasks. Its efficiency in terms of required annotations also allows rapid training and fine-tuning of models. The corresponding code, pretrained models, and instructions for operating the MemBrain program can be found at: https://github.com/CellArchLab/MemBrain.

Statistical spatial analysis for cryo-electron tomography.

Cryo-electron tomography (cryo-ET) is uniquely suited to precisely localize macromolecular complexes in situ, that is in a close-to-native state within their cellular compartments, in three-dimensions at high resolution. Point pattern analysis (PPA) allows quantitative characterization of the spatial organization of particles. However, current implementations of PPA functions are not suitable for applications to cryo-ET data because they do not consider the real, typically irregular 3D shape of cellular compartments and molecular complexes. Here, we designed and implemented first and the second-order, uni- and bivariate PPA functions in a Python package for statistical spatial analysis of particles located in three dimensional regions of arbitrary shape, such as those encountered in cellular cryo-ET imaging (PyOrg). To validate the implemented functions, we applied them to specially designed synthetic datasets. This allowed us to find the algorithmic solutions that provide the best accuracy and computational performance, and to evaluate the precision of the implemented functions. Applications to experimental data showed that despite the higher computational demand, the use of the second-order functions is advantageous to the first-order ones, because they allow characterization of the particle organization and statistical inference over a range of distance scales, as well as the comparative analysis between experimental groups comprising multiple tomograms. Altogether, PyOrg is a versatile, precise, and efficient open-source software for reliable quantitative characterization of macromolecular organization within cellular compartments imaged in situ by cryo-ET, as well as to other 3D imaging systems where real-size particles are located within regions possessing complex geometry.

Deep learning improves macromolecule identification in 3D cellular cryo-electron tomograms.

Cryogenic electron tomography (cryo-ET) visualizes the 3D spatial distribution of macromolecules at nanometer resolution inside native cells. However, automated identification of macromolecules inside cellular tomograms is challenged by noise and reconstruction artifacts, as well as the presence of many molecular species in the crowded volumes. Here, we present DeepFinder, a computational procedure that uses artificial neural networks to simultaneously localize multiple classes of macromolecules. Once trained, the inference stage of DeepFinder is faster than template matching and performs better than other competitive deep learning methods at identifying macromolecules of various sizes in both synthetic and experimental datasets. On cellular cryo-ET data, DeepFinder localized membrane-bound and cytosolic ribosomes (roughly 3.2 MDa), ribulose 1,5-bisphosphate carboxylase-oxygenase (roughly 560 kDa soluble complex) and photosystem II (roughly 550 kDa membrane complex) with an accuracy comparable to expert-supervised ground truth annotations. DeepFinder is therefore a promising algorithm for the semiautomated analysis of a wide range of molecular targets in cellular tomograms.

In situ architecture of neuronal α-Synuclein inclusions.

The molecular architecture of α-Synuclein (α-Syn) inclusions, pathognomonic of various neurodegenerative disorders, remains unclear. α-Syn inclusions were long thought to consist mainly of α-Syn fibrils, but recent reports pointed to intracellular membranes as the major inclusion component. Here, we use cryo-electron tomography (cryo-ET) to image neuronal α-Syn inclusions in situ at molecular resolution. We show that inclusions seeded by α-Syn aggregates produced recombinantly or purified from patient brain consist of α-Syn fibrils crisscrossing a variety of cellular organelles. Using gold-labeled seeds, we find that aggregate seeding is predominantly mediated by small α-Syn fibrils, from which cytoplasmic fibrils grow unidirectionally. Detailed analysis of membrane interactions revealed that α-Syn fibrils do not contact membranes directly, and that α-Syn does not drive membrane clustering. Altogether, we conclusively demonstrate that neuronal α-Syn inclusions consist of α-Syn fibrils intermixed with membranous organelles, and illuminate the mechanism of aggregate seeding and cellular interaction.

Trans-synaptic assemblies link synaptic vesicles and neuroreceptors.

Synaptic transmission is characterized by fast, tightly coupled processes and complex signaling pathways that require a precise protein organization, such as the previously reported nanodomain colocalization of pre- and postsynaptic proteins. Here, we used cryo-electron tomography to visualize synaptic complexes together with their native environment comprising interacting proteins and lipids on a 2- to 4-nm scale. Using template-free detection and classification, we showed that tripartite trans-synaptic assemblies (subcolumns) link synaptic vesicles to postsynaptic receptors and established that a particular displacement between directly interacting complexes characterizes subcolumns. Furthermore, we obtained de novo average structures of ionotropic glutamate receptors in their physiological composition, embedded in plasma membrane. These data support the hypothesis that synaptic function is carried by precisely organized trans-synaptic units. It provides a framework for further exploration of synaptic and other large molecular assemblies that link different cells or cellular regions and may require weak or transient interactions to exert their function.

The structural basis of Rubisco phase separation in the pyrenoid.

Approximately one-third of global CO2 fixation occurs in a phase-separated algal organelle called the pyrenoid. The existing data suggest that the pyrenoid forms by the phase separation of the CO2-fixing enzyme Rubisco with a linker protein; however, the molecular interactions underlying this phase separation remain unknown. Here we present the structural basis of the interactions between Rubisco and its intrinsically disordered linker protein Essential Pyrenoid Component 1 (EPYC1) in the model alga Chlamydomonas reinhardtii. We find that EPYC1 consists of five evenly spaced Rubisco-binding regions that share sequence similarity. Single-particle cryo-electron microscopy of these regions in complex with Rubisco indicates that each Rubisco holoenzyme has eight binding sites for EPYC1, one on each Rubisco small subunit. Interface mutations disrupt binding, phase separation and pyrenoid formation. Cryo-electron tomography supports a model in which EPYC1 and Rubisco form a codependent multivalent network of specific low-affinity bonds, giving the matrix liquid-like properties. Our results advance the structural and functional understanding of the phase separation underlying the pyrenoid, an organelle that plays a fundamental role in the global carbon cycle.

Reliable estimation of membrane curvature for cryo-electron tomography.

Curvature is a fundamental morphological descriptor of cellular membranes. Cryo-electron tomography (cryo-ET) is particularly well-suited to visualize and analyze membrane morphology in a close-to-native state and molecular resolution. However, current curvature estimation methods cannot be applied directly to membrane segmentations in cryo-ET, as these methods cannot cope with some of the artifacts introduced during image acquisition and membrane segmentation, such as quantization noise and open borders. Here, we developed and implemented a Python package for membrane curvature estimation from tomogram segmentations, which we named PyCurv. From a membrane segmentation, a signed surface (triangle mesh) is first extracted. The triangle mesh is then represented by a graph, which facilitates finding neighboring triangles and the calculation of geodesic distances necessary for local curvature estimation. PyCurv estimates curvature based on tensor voting. Beside curvatures, this algorithm also provides robust estimations of surface normals and principal directions. We tested PyCurv and three well-established methods on benchmark surfaces and biological data. This revealed the superior performance of PyCurv not only for cryo-ET, but also for data generated by other techniques such as light microscopy and magnetic resonance imaging. Altogether, PyCurv is a versatile open-source software to reliably estimate curvature of membranes and other surfaces in a wide variety of applications.

Author Correction: Template-free detection and classification of membrane-bound complexes in cryo-electron tomograms.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Author Correction: Template-free detection and classification of membrane-bound complexes in cryo-electron tomograms.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Template-free detection and classification of membrane-bound complexes in cryo-electron tomograms.

With faithful sample preservation and direct imaging of fully hydrated biological material, cryo-electron tomography provides an accurate representation of molecular architecture of cells. However, detection and precise localization of macromolecular complexes within cellular environments is aggravated by the presence of many molecular species and molecular crowding. We developed a template-free image processing procedure for accurate tracing of complex networks of densities in cryo-electron tomograms, a comprehensive and automated detection of heterogeneous membrane-bound complexes and an unsupervised classification (PySeg). Applications to intact cells and isolated endoplasmic reticulum (ER) allowed us to detect and classify small protein complexes. This classification provided sufficiently homogeneous particle sets and initial references to allow subsequent de novo subtomogram averaging. Spatial distribution analysis showed that ER complexes have different localization patterns forming nanodomains. Therefore, this procedure allows a comprehensive detection and structural analysis of complexes in situ.

Tricalbin-Mediated Contact Sites Control ER Curvature to Maintain Plasma Membrane Integrity.

Membrane contact sites (MCS) between the endoplasmic reticulum (ER) and the plasma membrane (PM) play fundamental roles in all eukaryotic cells. ER-PM MCS are particularly abundant in Saccharomyces cerevisiae, where approximately half of the PM surface is covered by cortical ER (cER). Several proteins, including Ist2, Scs2/22, and Tcb1/2/3 are implicated in cER formation, but the specific roles of these molecules are poorly understood. Here, we use cryo-electron tomography to show that ER-PM tethers are key determinants of cER morphology. Notably, Tcb proteins (tricalbins) form peaks of extreme curvature on the cER membrane facing the PM. Combined modeling and functional assays suggest that Tcb-mediated cER peaks facilitate the transport of lipids between the cER and the PM, which is necessary to maintain PM integrity under heat stress. ER peaks were also present at other MCS, implying that membrane curvature enforcement may be a widespread mechanism to regulate MCS function.

Dynamic instability of clathrin assembly provides proofreading control for endocytosis.

Clathrin-mediated endocytosis depends on the formation of functional clathrin-coated pits that recruit cargos and mediate the uptake of those cargos into the cell. However, it remains unclear whether the cargos in the growing clathrin-coated pits are actively monitored by the coat assembly machinery. Using a cell-free reconstitution system, we report that clathrin coat formation and cargo sorting can be uncoupled, indicating that a checkpoint is required for functional cargo incorporation. We demonstrate that the ATPase Hsc70 and a dynamic exchange of clathrin during assembly are required for this checkpoint. In the absence of Hsc70 function, clathrin assembles into pits but fails to enrich cargo. Using single-molecule imaging, we further show that uncoating takes place throughout the lifetime of the growing clathrin-coated pits. Our results suggest that the dynamic exchange of clathrin, at the cost of the reduced overall assembly rates, primarily serves as a proofreading mechanism for quality control of endocytosis.

The Architecture of Traveling Actin Waves Revealed by Cryo-Electron Tomography.

Actin waves are dynamic supramolecular structures involved in cell migration, cytokinesis, adhesion, and neurogenesis. Although wave-like propagation of actin networks is a widespread phenomenon, the actin architecture underlying wave propagation remained unknown. In situ cryo-electron tomography of Dictyostelium cells unveils the wave architecture and provides evidence for wave progression by de novo actin nucleation. Subtomogram averaging reveals the structure of Arp2/3 complex-mediated branch junctions in their native state, and enables quantitative analysis of the 3D organization of branching within the waves. We find an excess of branches directed toward the substrate-attached membrane, and tent-like structures at sites of branch clustering. Fluorescence imaging shows that Arp2/3 clusters follow accumulation of the elongation factor VASP. We propose that filament growth toward the membrane lifts up the actin network as the wave propagates, until depolymerization of oblique filaments at the back causes the collapse of horizontal filaments into a compact layer.

In Situ Structure of Neuronal C9orf72 Poly-GA Aggregates Reveals Proteasome Recruitment.

Protein aggregation and dysfunction of the ubiquitin-proteasome system are hallmarks of many neurodegenerative diseases. Here, we address the elusive link between these phenomena by employing cryo-electron tomography to dissect the molecular architecture of protein aggregates within intact neurons at high resolution. We focus on the poly-Gly-Ala (poly-GA) aggregates resulting from aberrant translation of an expanded GGGGCC repeat in C9orf72, the most common genetic cause of amyotrophic lateral sclerosis and frontotemporal dementia. We find that poly-GA aggregates consist of densely packed twisted ribbons that recruit numerous 26S proteasome complexes, while other macromolecules are largely excluded. Proximity to poly-GA ribbons stabilizes a transient substrate-processing conformation of the 26S proteasome, suggesting stalled degradation. Thus, poly-GA aggregates may compromise neuronal proteostasis by driving the accumulation and functional impairment of a large fraction of cellular proteasomes.

In Situ Architecture and Cellular Interactions of PolyQ Inclusions.

Expression of many disease-related aggregation-prone proteins results in cytotoxicity and the formation of large intracellular inclusion bodies. To gain insight into the role of inclusions in pathology and the in situ structure of protein aggregates inside cells, we employ advanced cryo-electron tomography methods to analyze the structure of inclusions formed by polyglutamine (polyQ)-expanded huntingtin exon 1 within their intact cellular context. In primary mouse neurons and immortalized human cells, polyQ inclusions consist of amyloid-like fibrils that interact with cellular endomembranes, particularly of the endoplasmic reticulum (ER). Interactions with these fibrils lead to membrane deformation, the local impairment of ER organization, and profound alterations in ER membrane dynamics at the inclusion periphery. These results suggest that aberrant interactions between fibrils and endomembranes contribute to the deleterious cellular effects of protein aggregation. VIDEO ABSTRACT.