Isotropic reconstruction for electron tomography with deep learning

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.

Subtomogram averaging (STA) is a powerful image processing technique in electron tomography used to determine the 3D structure of macromolecular complexes in their native environments. It is a fast growing technique with increasing importance in structural biology. The computational aspect of STA is very complex and depends on a large number of variables. We noticed a lack of detailed guides for STA processing. Also, current publications in this field often lack a documentation that is practical enough to reproduce the results with reasonable effort, which is necessary for the scientific community to grow. We therefore provide a complete, detailed, and fully reproducible processing protocol that covers all aspects of particle picking and particle alignment in STA. The command line–based workflow is fully based on the popular Dynamo software for STA. Within this workflow, we also demonstrate how large parts of the processing pipeline can be streamlined and automatized for increased throughput. This protocol is aimed at users on all levels. It can be used for training purposes, or it can serve as basis to design user-specific projects by taking advantage of the flexibility of Dynamo by modifying and expanding the given pipeline. The protocol is successfully validated using the Electron Microscopy Public Image Archive (EMPIAR) database entry 10164 from immature HIV-1 virus-like particles (VLPs) that describe a geometry often seen in electron tomography.

Cryo-electron tomography (cryo-ET) can be used to reconstruct three-dimensional (3D) volumes, or tomograms, from a series of tilted two-dimensional images of biological objects in their near-native states in situ or in vitro. 3D subvolumes, or subtomograms, containing particles of interest can be extracted from tomograms, aligned, and averaged in a process called subtomogram averaging (STA). STA overcomes the low signal to noise ratio within the individual subtomograms to generate structures of the particle(s) of interest. In recent years, cryo-ET with STA has increasingly been capable of reaching subnanometer resolution due to improvements in microscope hardware and data processing strategies. There has also been an increase in the number and quality of software packages available to process cryo-ET data with STA. In this review, we describe and assess the data processing strategies available for cryo-ET data and highlight the recent software developments which have enabled the extraction of high-resolution information from cryo-ET datasets.

In the recent years, cryo-electron microscopy (cryo-EM) single-particle analysis (SPA) has greatly benefited from a multitude of technological advances cumulating in breaking the atomic-resolution barrier for biological macromolecules. Since then, structural insights by cryo-EM have led to driving rational design of new vaccines and therapeutics due to high throughput and high-resolution screening and data collection technologies. Additionally, cryo-electron tomography (cryo-ET) has become an indispensable tool for structural cell biology. In this presentation we elaborate on the latest Thermo Fisher Scientific innovations for 300kV cryo electron microscopy for SPA and cryo-ET to advance more workflow efficiency and performance. Our ongoing mission is to further improve the connectivity between the cryo-FIB and cryo-TEM for an even more streamlined tomography workflow that is safer, faster and cleaner. Data acquisition and processing can be a significant cryo-ET bottleneck. Our integrated solution, combining Thermo Scientific Tomography 5 and Tomo Live, automates pre-acquisition tasks, enhances productivity with predictive tracking and enables rapid quality assessment. Robust and optimized motion correction helps produce crisp tomograms and image quality needed for sub-tomogram averaging. Results are displayed with new quality metrics on a web platform, making the workflow user-friendly and facilitating structural biology advancements.

Advances in cryogenic electron microscopy (cryoEM) single particle analysis have revolutionized structural biology by facilitating the in vitro determination of atomic- and near-atomic-resolution structures for fully hydrated macromolecular complexes exhibiting compositional and conformational heterogeneity across a wide range of sizes. Cryogenic electron tomography (cryoET) and subtomogram averaging are rapidly progressing toward delivering similar insights for macromolecular complexes in situ, without requiring tags or harsh biochemical purification. Furthermore, cryoET enables the visualization of cellular and tissue phenotypes directly at molecular, nanometric resolution without chemical fixation or staining artifacts. This forward-looking review covers recent developments in cryoEM/ET and related technologies such as cryogenic focused ion beam milling scanning electron microscopy and correlative light microscopy, increasingly enhanced and supported by artificial intelligence algorithms. Their potential application to emerging concepts is discussed, primarily the prospect of complementing medical histopathology analysis. Machine learning solutions are poised to address current challenges posed by "big data" in cryoET of tissues, cells, and macromolecules, offering the promise of enabling novel, quantitative insights into disease processes, which may translate into the clinic and lead to improved diagnostics and targeted therapeutics.

Cryogenic single-particle photoluminescence (PL) spectroscopy has been used with great success to directly observe the heterogeneous photophysical states present in a population of luminescent particles. Cryogenic electron tomography provides complementary nanometer scale structural information to PL spectroscopy, but the two techniques have not been correlated due to technical challenges. Here, we present a method for correlating single-particle information from these two powerful microscopy modalities. We simultaneously observe PL brightness, emission spectrum, and in-plane excitation dipole orientation of CdSSe/ZnS quantum dots suspended in vitreous ice. Stable and fluctuating emitters were observed, as well as a surprising splitting of the PL spectrum into two bands with an average energy separation of 80 meV. In some cases, the onset of the splitting corresponded to changes in the in-plane excitation dipole orientation. These dynamics were assigned to structures of individual quantum dots and the excitation dipoles were visualized in the context of structural features.

Quantifying resolution limiting factors in subtomogram averaged cryo-electron tomography using simulations.

Human Immunodeficiency virus type 1 (HIV-1) is a lentivirus that infects non-dividing cells of the immune system. In non-dividing cells, nuclear import of the viral genome occurs through the nuclear pore complex (NPC), a large macromolecular assembly that forms a channel of ~40 nm across the nuclear envelope. This process requires interactions between NPC components known as nucleoporins and CA (capsid) proteins of HIV-1 post-fusion complexes. In mature HIV-1 virions, multiple copies of CA assemble into a lattice of hexamers and pentamers, forming a cone-shaped capsid core of ~60 nm in width and ~110 nm in length encasing the viral genome. Because of its large size, it was generally believed that the capsid core entirely or partially disassembled prior to its translocation through the NPC. However, nuclear entry and uncoating are rare events and challenging to characterize and therefore subject to a long-standing debate. Here, cryo-electron tomography (cryo-ET) on cryo-focused ion beam (FIB) milled T-cells was combined with subtomogram averaging (SA) to study the ultrastructure of HIV-1 capsids in the process of nuclear entry. Using a HIV-1 variant arrested at the nuclear pore due to a defect in binding of the CA lattice to the host cell protein CPSF6, snapshots of HIV-1 nuclear entry at multiple stages were captured. Reverse transcription-competent HIV-1 complexes were identified in the cytosol, docking to and within the NPC, and in the nucleoplasm of T cells. Surprisingly, in the cytosol and at the NPC, the viral complexes retained cone-shaped capsids highly resembling in size and geometry the mature capsid cores within intact virions. The density at their interior suggested that they were associated with the viral genome. The cone-shape capsids deeply enter into the NPC central channel with their narrow ends. The hexameric lattices of these capsids were intact or nearly intact. These findings argued against the current models of uncoating in the cytosol or at the NPC and rather supported translocation of intact HIV-1 capsids through the NPC. Instead, HIV-1 capsids uncoat in the nucleus of T cells. Inside the nucleus, tube-shape fragments still containing few lattice elements were observed. The lack of density at their inside suggested that the viral genome was released from these complexes. Uncoating may thus occur through the partial opening and remodeling of the hexameric lattices of HIV-1 capsids, rather than a step-wise disassembly. To address how an intact HIV-1 capsid can enter the central channel of NPC, the human NPC from T cells was structurally analyzed in cellulo by SA. The NPC overall structure was not altered upon infection but rather dilated in comparison to the structure previously obtained from purified envelopes. The diameter of the central channel of NPC in T cells corresponded to ~64 nm, suggesting that the translocation of intact HIV-1 capsid through the NPC is geometrically possible. The dilated conformation of NPCs in T cells is independent of HIV-1 infection, but represents a physiological condition of actively transporting NPCs.

Cryo-electron tomography (cryo-ET) allows the visualization of cellular structures under close-to-life conditions and at molecular resolution. While it is inherently a static approach, yielding structural information about supramolecular organization at a certain time point, it can nevertheless provide insights into function of the structures imaged, in particular, when supplemented by other approaches. Here, we review the use of experimental methods that supplement cryo-ET imaging of whole cells. These include genetic and pharmacological manipulations, as well as correlative light microscopy and cryo-ET. While these methods have mostly been used to detect and identify structures visualized in cryo-ET or to assist the search for a feature of interest, we expect that in the future they will play a more important role in the functional interpretation of cryo-tomograms.

On geometric artifacts in cryo electron tomography

Background Electron cryotomography is a powerful imaging technique allowing for studying functional cellular modules in their native environment with macromolecular resolution. However, it requires access to complex and expensive instrumentation, typically a 300 kV electron cryomicroscope equipped with an energy filter. Simpler and cheaper 200 and 100 kV instruments have been successfully used for single particle cryoEM analyses, which has helped to democratize the technique and broaden access. It has not been systematically studied if 200 kV electron cryomicroscopes can deliver meaningful and interpretable data with respect to electron cryotomography applications. Methods Here, we set out to investigate if a 200 kV electron cryomicroscope without an energy filter can be utilized for in situ structural studies of mammalian cells by electron cryotomography of thin cell edges followed by extensive image analysis including segmentations, subtomogram averaging and molecular sociology studies of lipid droplets. Results We demonstrate that the resulting tomograms of thin edges of U2OS cells are of sufficient quality to annotate the contents of the cell and observe spatial inter-relationships among macromolecules. In particular, we undertook a molecular sociology analysis of lipid droplets and addressed their subcellular distribution and interactions with other organelles. Additionally, we performed subtomogram averaging of purified 70S ribosomes that resulted in ∼15 Å resolution 3D reconstruction. Finally, we examined geographical distribution and scientific output of the two most common electron cryomicroscopy platforms and deduced that 200 kV instruments are heavily underutilized with respect to electron cryotomography applications. Discussion This study demonstrates that 200 kV electron cryomicroscopes can be utilized for structural cell biology studies by electron cryotomography. Given the favorable ratio of their versatility versus costs we foresee that 200 kV electron cryomicroscopes will become workhorses of local electron cryomicroscopy facilities.

Electron cryotomography is a powerful imaging technique allowing for studying functional cellular modules in their native environment with macromolecular resolution. However, it requires access to complex and expensive instrumentation, typically a 300 kV electron cryomicroscope equipped with an energy filter. Simpler and cheaper 200 and 100 kV instruments have been successfully used for single particle cryoEM analyses, which has helped to democratize the technique and broaden access. It has not been systematically studied if 200 kV electron cryomicroscopes can deliver meaningful and interpretable data with respect to electron cryotomography applications. Here, we set out to investigate if a 200 kV electron cryomicroscope without an energy filter can be utilized for in situ structural studies of mammalian cells by electron cryotomography of thin cell edges followed by extensive image analysis including segmentations, subtomogram averaging and molecular sociology studies of lipid droplets. We demonstrate that the resulting tomograms of thin edges of U2OS cells are of sufficient quality to annotate the contents of the cell and observe spatial inter-relationships among macromolecules. In particular, we undertook a molecular sociology analysis of lipid droplets and addressed their subcellular distribution and interactions with other organelles. Additionally, we performed subtomogram averaging of purified 70S ribosomes that resulted in ~ 15 Å resolution 3D reconstruction. Finally, we examined geographical distribution and scientific output of the two most common electron cryomicroscopy platforms and deduced that 200 kV instruments are heavily underutilized with respect to electron cryotomography applications. This study demonstrates that 200 kV electron cryomicroscopes can be utilized for structural cell biology studies by electron cryotomography. Given the favorable ratio of their versatility versus costs we foresee that 200 kV electron cryomicroscopes will become workhorses of local electron cryomicroscopy facilities.

Determining the three-dimensional structure of biomacromolecules at high resolution in their native cellular environment is a major challenge for structural biology. Toward this end, electron cryotomography (ECT) allows large bio-macromolecular assemblies to be imaged directly in their hydrated physiological milieu to ~4nm resolution. Combining ECT with other techniques like fluorescent imaging, immunogold labeling, and genetic manipulation has allowed the in situ investigation of complex biological processes at macromolecular resolution. Furthermore, the advent of cryogenic focused ion beam (FIB) milling has extended the domain of ECT to include regions even deep within thick eukaryotic cells. Anticipating two audiences (scientists who just want to understand the potential and general workflow involved and scientists who are learning how to do the work themselves), here we present both a broad overview of this kind of work and a step-by-step example protocol for ECT and subtomogram averaging using the Legionella pneumophila Dot/Icm type IV secretion system (T4SS) as a case study. While the general workflow is presented in step-by-step detail, we refer to online tutorials, user's manuals, and other training materials for the essential background understanding needed to perform each step.

Electron tomography (ET) offers a unique capacity to image biological structures in situ. However, the resolution of ET reconstructed tomograms is not comparable to that of the single-particle cryo-EM. If many copies of the object of interest are present in the tomograms, their structures can be reconstructed in the tomogram, picked, aligned and averaged to increase the signal-to-noise ratio and improve the resolution, which is known as the subtomogram averaging. To date, the resolution improvement of the subtomogram averaging is still limited because each reconstructed subtomogram is of low reconstruction quality due to the missing wedge issue. In this article, we propose a novel computational model, the constrained reconstruction model (CRM), to better recover the information from the multiple subtomograms and compensate for the missing wedge issue in each of them. CRM is supposed to produce a refined reconstruction in the final turn of subtomogram averaging after alignment, instead of directly taking the average. We first formulate the averaging method and our CRM as linear systems, and prove that the solution space of CRM is no larger, and in practice much smaller, than that of the averaging method. We then propose a sparse Kaczmarz algorithm to solve the formulated CRM, and further extend the solution to the simultaneous algebraic reconstruction technique (SART). Experimental results demonstrate that CRM can significantly alleviate the missing wedge issue and improve the final reconstruction quality. In addition, our model is robust to the number of images in each tilt series, the tilt range and the noise level. The codes of CRM-SIRT and CRM-SART are available at https://github.com/icthrm/CRM. Supplementary data are available at Bioinformatics online.

Structures of macromolecules in their native state provide unique unambiguous insights into their functions. Cryo-electron tomography combined with subtomogram averaging demonstrated the power to solve such structures in situ at resolutions in the range of 3 Angstrom for some macromolecules. In order to be applicable to the structural determination of the majority of macromolecules observable in cells in limited amounts, processing of tomographic data has to be performed in a high-throughput manner. Here we present TomoBEAR—a modular configurable workflow engine for streamlined processing of cryo-electron tomographic data for subtomogram averaging. TomoBEAR combines commonly used cryo-EM packages with reasonable presets to provide a transparent (“white box”) approach for data management and processing. We demonstrate applications of TomoBEAR to two data sets of purified macromolecular targets, to an ion channel RyR1 in a membrane, and the tomograms of plasma FIB-milled lamellae and demonstrate the ability to produce high-resolution structures. TomoBEAR speeds up data processing, minimizes human interventions, and will help accelerate the adoption of in situ structural biology by cryo-ET. The source code and the documentation are freely available.