This video protocol illustrates how to perform scanning transmission electron microscopy tomography of virological specimens. For optimal outcomes, samples are prepared by high-pressure freezing and subsequent freeze substitution.
Method Article
* These authors contributed equally
This video protocol illustrates how to perform scanning transmission electron microscopy tomography of virological specimens. For optimal outcomes, samples are prepared by high-pressure freezing and subsequent freeze substitution.
Electron microscopy (EM) and especially three-dimensional (3D) EM techniques have become established methods in structural virology. Investigation of virus-induced alteration of the cellular ultrastructure, such as Zika virus (ZIKV)-induced replication factories or coronavirus replication organelles, demands 3D imaging. Transmission electron microscopy (TEM) tomography is a widely used method, despite its limitation to samples with a thickness of up to 200 nm. Focused ion beam-scanning electron microscopy (SEM)-tomography can produce 3D data of larger volumes with isotropic, albeit typically lower resolution. Alternative techniques, such as block face-scanning electron microscopy (BF-SEM), SEM array tomography, or TEM imaging of serial sections are used for imaging of larger volumes. However, compared to the previously mentioned techniques, these techniques come at the cost of much lower resolution along the Z-axis of the sample. A technique that provides 3D information of samples up to 1 µm thickness with isotropic resolution of a few nanometers is scanning transmission electron microscopy (STEM) tomography.
Here, we present a protocol for the preparation of high-pressure frozen, freeze-substituted, and resin-embedded virological specimens and their analysis using STEM tomography. This protocol benefits from the advantages of room temperature imaging while preserving the biological ultrastructure in a near-native state. We show two representative examples for questions that can be answered using STEM tomography. First, we apply the protocol for studying virion morphogenesis of a recombinant vesicular stomatitis virus (VSV). The STEM tomograms offer information on recombinant VSV budding that is otherwise not accessible by 2D imaging. Second, we show correlative light and electron microscopy (CLEM) using Foerster resonance energy transfer (FRET) imaging and STEM tomography (FRET-3D-CLEM) of EF-C peptide nanofibrils. This recently published combination gives new insights into the uptake and disassembly of infection-enhancing peptide nanofibrils, especially profiting from the large volume that is accessible by STEM tomography.

Figure 1: Comparison of different 3D microscopy techniques, considering resolution, instrumentation, acquisition time, and effort. The techniques´ maximum feasible volume is between the technically demanding cryo-TEM tomography that offers molecular resolution, and light microscopy, where resolution is limited by the diffraction limit of light. Abbreviations: TEM = transmission electron microscopy; STEM = scanning TEM; SEM = scanning electron microscopy; FIB = focused ion beam. Please click here to view a larger version of this figure.
Virus-related structures and systems span a scale from the sub-angstrom regime (e.g., small molecules for virotherapy, isolated viral proteins) to several meters (e.g., entire infected organisms). Viral particles are in the range of 20-1,000 nm and mainly investigated in their biological context, such as host cells and viral producer cell lines, which are in the µm range. Visualization of specific virus-related structures is thus dependent on choosing a technique that provides a sufficient field of view and suitable resolution. The study of virus-infected cells and tissue pieces relies on microscopy techniques (Figure 1). While overview images of cells and tissues can be acquired using classical light microscopy, intracellular structures involved in virus replication are often studied using electron microscopy (EM). Although nowadays mostly used in basic virological research, EM is also used in diagnostics1,2 and has also found its way into biopharmaceutical applications3.
Often, 2D EM approaches like classical transmission electron microscopy (TEM) cannot sufficiently answer a given biological question. In virology, this is especially the case when studying virus-host-cell interactions such as virion attachment and entry, the formation of viral factories, or virion morphogenesis4,5,6,7,8,9. Therefore, EM tomography is used to analyze the 3D ultrastructure of biological samples. In principle, there are two ways of acquiring a tomogram of a certain sample: 1) stepwise imaging of serial slices that are later combined into a 3D image stack, or 2) acquisition of projections from different viewing angles of the sample followed by computational reconstruction of the information into a virtual image stack.
The techniques of the first type include imaging of serial sections by TEM or SEM (SEM array tomography), serial block face SEM, and focused ion beam (FIB)-SEM tomography. SEM and TEM imaging of serial sections lack the resolution in z, which is limited by section thickness, but provides a large field of view and relies on equipment that is readily available in most EM labs4. FIB-SEM tomography can yield isotropic resolution, but device and maintenance costs can be challenging, and imaging of whole cells at isotropic resolution is time-consuming9,10.
The second type of tomography relies exclusively on the transmission of the electron beam through a sample. This can be done using two different imaging modalities: TEM or scanning transmission electron microscopy (STEM). While TEM tomography is more widespread and easier to implement, the section thickness is limited to ~200 nm. Therefore, structures larger than the section thickness, such as mitochondria, vesicles, or virions, cannot be imaged within a single tomogram. STEM tomography offers the possibility to image samples up to 1 µm thickness11,12. However, it requires a STEM with high acceleration voltage (≥200 kV) and a setup that supports high tilt angles as well as a small semi-convergence angle13. Although STEM imaging is a common technique in material science, it remains a rare technique in biologic EM laboratories.
Here, we present an application-driven approach to use STEM tomography for virological specimens. Due to the high complexity and labor-intensive nature of EM tomography, high-quality sample preparation is crucial. While cryogenic TEM (tomography) undoubtedly is the gold standard for investigations on a molecular level, room-temperature imaging of high-pressure frozen, freeze-substituted cells provides a solid preservation and clarity of cellular ultrastructure7,14. Briefly, adherent cells are grown on a sapphire disk and infected or otherwise treated as desired. The cells are then immobilized by freezing them under high pressure at very high cooling rates, resulting in vitrification of the water within. Subsequently, the water is replaced by a solvent while the samples are contrasted and slowly brought back to room temperature. The samples are then gradually embedded in resin.
There are many protocols for high-pressure freezing and freeze-substitution, which differ in the machinery, carrier system, and composition of the specimen. Detailed protocols for high-pressure freezing have been published, including one by Walther et al.15, and a video protocol by Steyer et al.16, which focuses on tissue preparation for FIB-SEM using microwave-assisted contrast enhancement. For virological research, Read et al.17 and Romero-Brey18 published detailed protocols for the preparation of virus-infected cells by high-pressure freezing with sapphire disks and subsequent analysis using electron tomography. Ultrathin sectioning has also been previously described15,17. Refer to the cited literature for more details regarding these approaches.
Here, we present a workflow that is an adapted and updated version of the previously published protocols15,17 paired for the first time with a detailed video guide. The protocol provides a workflow for sample preparation using high-pressure freezing, freeze substitution, and embedding in epoxy resin. Sections of 800 nm-1,000 nm thickness are cut using an ultramicrotome. A tilt series is acquired with a 200 kV STEM, using the software EM tools. This protocol is optimized for adherent cells and provides a comprehensive workflow while allowing modifications for other specimen types, such as tissue or suspension cells. For STEM tomography under cryogenic conditions, refer to the work of Michael Elbaum's group19,20.
1. Cell culture

Figure 2: High-pressure freezing. (A) Sapphire disks showing carbon coat with "2" (left) or a coordinate system for correlative light and electron microscopy (right). (B) Schematic depiction of an SD-gold spacer-SD sandwich as it is prepared for high-pressure freezing. The cells grown on the disks face each other and are thereby protected in the tight cavity. (C) Holder for high-pressure freezing that contains an SD sandwich. (D) Materials and devices for loading the HPF holder: (1) Heating plate, (2) Stereo microscope, (3) Filter paper, (4) Reaction tube containing 1-hexadecene, (5) Tweezers, (6) Uncoated SDs, (7) Petri dish with gold spacers, (8) HPF holder. (E) Setup of the HPF system: (1) LN-box for transferring frozen samples (see also F), (2) Hair dryer to speed up the warming up of the HPF holder after freezing, (3) Nitrogen dewar to refill the LN box. (F) LN-box for holding frozen samples after high-pressure freezing with a central trough (dotted outline) in which the HPF holder can be opened, and SDs can be safely handled. Materials needed directly after high-pressure freezing: (1) Holder for up to six sample containers, (2) Insulated tweezers, (3) Sample containers with lid. The containers contain small holes and small weights (e.g., a screw) so that they immerse in liquid nitrogen. Abbreviations: SD = sapphire disk; HPF = high-pressure freezer; LN = liquid nitrogen. Please click here to view a larger version of this figure.
2. Sample preparation for electron microscopy

Figure 3: Quality control by TEM-different appearances of cells after embedding, trimming, and ultrathin sectioning. (A) Epoxy resin-embedded cells after the SD was removed. The carbon layer with the scratched '2' remains on the resin-embedded cells and is clearly visible. The almost transparent cells are evenly distributed across the block face. (B) Trimmed block face before microtomy. Part of the scratched 2 is still visible on the surface of the resin block. (C) 70 nm-thin sectionsmounted on a TEM grid. (D) TEM overview of one entire ultrathin section. Note the holes in the section (white arrowheads) that result from incomplete resin infiltration and are a common artifact. (E) Magnified view of a selected cell marked with a white rectangle in (D). Similar to (D), incomplete embedding is visible at the plasma membrane (white arrowheads); however, the rest of the cell shows good preservation and embedding. Scale bars = 100 µm (D), 10 µm (E). Abbreviations: SD = sapphire disk; TEM = transmission electron microscopy. Please click here to view a larger version of this figure.
3. Imaging

Figure 4: Tilt series acquisition. (A) 200 kV STEM. (B) STEM overview of a high-pressure frozen, freeze-substituted cell embedded in epoxy resin. The tilt series shown in (C-E) was acquired at the position marked with the white rectangle. (C-E) Selected images of a tilt series acquired from -72° to +72°. Gold fiducials on both sides of the section are used for computational tomogram reconstruction. Clustered fiducials like the ones marked with the white ellipses cannot be used for reconstruction of the tomogram but are well recognizable at high tilt angles. Those or similar structures outside the region of interest are useful for setting the eucentric height. Scale bar = 1 µm (B). Abbreviation: STEM = scanning transmission electron microscopy. Please click here to view a larger version of this figure.
4. Tomogram reconstruction and segmentation
STEM tomography offers the possibility to image thicker sections than TEM tomography with isotropic resolution in the nanometer range and-as compared to FIB-SEM tomography-provides a relatively time-efficient imaging workflow. In virology, this is especially useful since it allows for imaging of a large number of ultrastructural features for statistical analysis in 3D, or when searching for rare events. Here, we show two representative results that demonstrate how STEM tomography can be used with regard to these two aspects.
Visualization of budding recombinant VSV virions
There is a broad variety of viruses that form particles of different shapes and sizes. To analyze the shape of virions in a virus suspension, often a simple method like negative stain TEM is sufficient. Characterizing the morphology of virions in the cellular context is more difficult. This holds especially true for viruses that form particles with anisotropic morphology, like rhabdoviridae. VSV is a rhabdovirus that forms bullet-shaped virions of ~180 nm in length. In suspension cells, the virions are randomly oriented, making it difficult to always discern the bullet shape in 2D images (Figure 5A-C). To properly define the bullet shape of all recombinant VSV virions in a sample, a 3D method with isotropic resolution is needed-one that can also provide volumetric information over a sufficiently large sample area. A STEM tomogram can include many complete virions in a single dataset (Figure 5D,E). Thereby, it is possible to image a high number of virions for statistical analysis, even if only one tomogram per cell is acquired. In comparison, a TEM tomogram of a 200 nm-thick section would only capture a few complete recombinant VSV virions (as marked in Figure 5E).

Figure 5: STEM tomography of recombinant VSV budding sites. (A-C) Details from a virtual section of a STEM tomogram obtained from an 800 nm thick section show that the bullet shape is not always visible due to (A,B) different orientations of the virions. (C) Only virions that are oriented parallel to the viewing plane present as well-defined bullets. (D) Budding sites of recombinant VSV resemble large clusters of virions with no preferred orientation. (E) The side view of the tomogram demonstrates the ability to capture many complete recombinant VSV virions in one section using STEM tomography. The virtual image stack has a thickness of approximately 600 nm at this position. The dashed lines represent the approximate thickness of a section for TEM tomography (200 nm), where only a few complete bullet-shaped virions would be included. Scale bar = 500 nm (D). Abbreviations: STEM = scanning transmission electron microscopy; VSV = vesicular stomatitis virus. Please click here to view a larger version of this figure.
Degradation of EF-C peptide nanofibrils
In the context of gene therapy, transduction-enhancing peptide nanofibrils have become a topic of interest. In a recently published study24, correlative imaging using Foerster Resonance Energy Transfer (FRET) imaging and STEM tomography (FRET-3D-CLEM) was used to investigate the uptake and disassembly of EF-C peptide nanofibrils. High FRET efficiencies, corresponding to FRET pairs that are close to each other, emit red fluorescence corresponding to assembled fibrils. Low FRET efficiencies, corresponding to FRET pairs that are far from each other, emit green fluorescence, representing disassembled fibrils. Both signals appear in cells when they are in the process of taking up and degrading EF-C peptide nanofibrils. Areas where high and low FRET efficiencies are found appear in yellow (Figure 6A).

Figure 6: Imaging of EF-C peptide nanofibril degradation by correlative FRET imaging and STEM tomography (FRET-3D-CLEM). (A) Confocal microscopy of a cell incubated with EF-C nanofibrils that were labeled with a FRET pair. Green signals correspond to putatively disassembled fibrils. Red corresponds to putatively intact fibrils. Yellow corresponds to areas where a mixture of the red and the green signal is detected. Nu nucleus. (B) STEM overview of an 800 nm-thick section from the same cell. (C) The position of the tomogram is indicated. The image is a composite of two images acquired on different sections, because the cell was partially blocked by the bar grid on one of the sections. The nucleus is labeled with Nu in A, B, and C. (C) Virtual section from the tomogram. Red arrowheads mark regions overlapping with red fluorescence signals. STEM tomography confirms the presence of intact fibrils. The green arrowhead points to a region with disassembled fibrils. (D) Segmentation of the tomogram. An endocytosed filopodium is pinching off the vesicle membrane (white arrow). The large volume that can be investigated with a STEM tomogram (here approximately: 7 µm x 7 µm x 550 nm) makes it easier to observe such a rare event. (E) Virtual sections 19, 24, and 29 of the tomogram. At a pixel size of 6.8 nm, the sections are each 27.2 nm apart. The figure is reproduced from the data published by Rauch-Wirth et al.24 . Scale bars = 5 µm (A), 1 µm (C). Abbreviations: STEM = scanning transmission electron microscopy; FRET = Foerster resonance energy transfer. Please click here to view a larger version of this figure.
Visualization was performed by this correlative approach to confirm that EF-C fibrils are taken up by the cell and subsequently degraded. Light microscopy lets us choose cells for electron microscopy that provide red and green signals, to compare the ultrastructural appearance correlating with the different fluorescence signals. STEM tomograms from the same cell show that the red signal can either originate from fibrils that are still outside the cell or are seemingly freshly internalized by macropinocytosis (Figure 6C). Furthermore, CLEM verified that the green signal indeed originated from degraded fibrils (Figure 6B,C which can be found in lysosomes (Figure 6C, largest vesicle). Due to the large volume accessible by STEM tomography and the 3D visualization of the structures, it is easier to correlate the signal from light microscopy with ultrastructural features and interpret them. Furthermore, rare events, such as pinching-off filopodia inside the lysosome (Figure 6D,E, white arrow) can be detected with this approach.
STEM tomography offers a resolution comparable to TEM tomography combined with the ability to investigate volumes with up to 1 µm thickness. We therefore consider it ideal for imaging virus-induced changes of intracellular structures, like membrane remodeling4,6,25, budding of virions7, or endocytosis5. Depending on the research question, other VolumeEM techniques might be considered. For a technique that provides a larger field of view with a comparable resolution, we recommend FIB-SEM tomography. FIB-SEM tomography will be slower than STEM tomography when imaging a comparable volume with similar resolution, since the imaging time increases when the resolution increases. Often, this is then countermeasured by imaging smaller fields of view. Furthermore, a z-resolution of 6.8 nm (Figure 6) would be technically challenging with ion beam milling. Therefore, FIB-SEM is optimal for imaging larger structures like whole cells and tissue9,10,26,27, albeit usually with (isotropic) pixel sizes of several tens of nanometers. Furthermore, FIB-SEM tomograms show no "missing wedge" effects, which is typical for tomograms reconstructed from a tilt series. Moreover, with the development of plasma FIB machines, alternatives that provide faster and still precise milling are currently being introduced to the field.
If a high resolution in z is not crucial, TEM or SEM imaging of serial sections can be an alternative that allows for the investigation of a very large field of view in a short time28,29, however, usually with anisotropic resolution. In TEM tomography, different tomograms from serial thick sections can be combined30,31. The same is possible in STEM tomography; however, given the thickness of the sections that can be investigated, there is less need to combine a number of tomograms. In virology, TEM tomography is often used to characterize virus-induced membrane remodeling32,33,34,35. While TEM tomography itself provides valuable insights in the replication cycle of viruses, most use-cases would benefit from the capability of STEM tomography to image thicker volumes. This especially holds true for virions with an anisotropic morphology, like rhabdoviridae, where TEM tomography can only capture a few complete virions in one tomogram35 or for imaging several enveloping virions in a replication compartment with a size of the cytoplasmic viral assembly complex (cVAC) of the human cytomegalovirus (HCMV)4.
STEM tomography under cryogenic conditions offers further functional insights19 and a video protocol is available20. Our protocol differs in the sample preparation workflow, which, instead of preserving the sample in the native state, provides near-native ultrastructural preservation. While STEM tomography of high-pressure frozen, freeze-substituted samples cannot achieve the ultrastructural preservation and atomic resolution seen in cryogenic EM, it benefits from the advantages of simpler equipment, more stable specimens, and a higher throughput.
The following steps are critical for the success of the method:
First, a clear goal of the imaging experiment. Due to the labor-intensive nature of the technique, a well-planned experiment with a clear goal for imaging is needed. Detailed knowledge of the model system -- cell line and virus -- should be generated beforehand by using other methods that are less difficult to perform. Therefore, we encourage optimizing infectivity rates in light microscopy-based pre-experiments and adjusting experimental conditions to suit the biological system. Optimization parameters include the number of cells, multiplicity of infection (MOI), and time points based on the virus and the event to be imaged. This will ensure a high proportion of evaluable cells in the sample.
Second, high-quality sample preparation. We emphasize the importance of quality sample preparation. At the present time, we consider vitrification by high-pressure freezing as the gold standard for thick specimens like cells. Freeze substitution, embedding, and subsequent imaging at room temperature is a robust way of acquiring high-quality results when combined with STEM tomography. Therefore, establishing a stable routine for sample preparation is crucial. For this, we would like to highlight the literature about high-pressure freezing14,21. If a less toxic freeze-substitution solution is to be used, we recommend using potassium permanganate as a substitute for osmium tetroxide, as described by Schauflinger et al.36.
Finally, a smooth-running workflow for tilt series acquisition, including a well-aligned microscope, should be established in accordance with the user instructions of the STEM and will differ from microscope to microscope. Establishing a smooth workflow also depends on the scan generator and the image acquisition software.
JGW, RH, EP, JSR, and MD are employees of Boehringer Ingelheim Pharma GmbH & Co.KG, a pharmaceutical company with an interest in developing and producing virus-based products. This had no influence on the design of the paper; therefore, we report no conflicts of interest.
This publication is part of the Global Development Technology Strategy and was supported by the Global Technology Management at Boehringer Ingelheim Development CMC Biologicals. We thank Renate Kunz and Jana Apolloni for excellent technical assistance and helpful discussion concerning sample preparation. We thank Reinhard Weih for excellent assistance with technical issues and concerning device maintenance. We thank Torsten Friedrich for helpful discussion and for aiding in the setup of the microscope. The work of Clarissa Read, Julia LaRoche, and Jan Münch has received funding by a Cooperative Research Centre Grant of the German Research Foundation (SFB1279).
| Name | Company | Catalog Number | Comments |
|---|---|---|---|
| 1-Hexadecene | Merck KGaA | 8220640500 | |
| 200 copper grid with parallel bars | Plano GmbH | G2014C | |
| 37 °C drying chamber | Binder GmbH | 9010-0323 | |
| Acetone | VWR international GmbH | 20066330 | |
| DDSA | Electron Microscopy Sciences | 13710 | |
| Diamond knife | Diatome Ltd | https://www.diatomeknives.com/ultra-35 | |
| DMP-30 | Electron Microscopy Sciences | 13600 | |
| Embed-812 Resin | Electron Microscopy Sciences | 14900 | |
| EM tools | TVIPS GmbH | TVIPS Software: EM-Tools; https://www.tvips.com/imaging-software/em-tools/ | |
| Ethanol 100% | Merck KGaA | 1.00983.2511 | |
| Filter paper | VWR international GmbH | 516-0812 | |
| Flaschen mit Schnappdeckelglas 11 mL | VWR international GmbH | 548-0625 | |
| Formvar | Electron Microscopy Sciences | 15830-25 | |
| Freeze substitution unit Leica EM AFS2 | Leica Microsystems GmbH | https://www.leica-microsystems.com/products/sample-preparation-for-electron-microscopy/p/leica-em-afs2/ | |
| Glod Sol 25 nm | AURION Immuno Gold Reagents & Accessories | 425.011 | |
| Gold pinehole ø 2000 μm | Science Services GmbH | GA2000-Au | |
| Graiticules Buchstaben-Netzchen Kupfer | Plano GmbH | NH6C | |
| Heating plate | MEDAX GmbH & Co. KG | 12501 | |
| High vacuum vaporization system BAF 300 | BAL-TEC AG | ||
| High-pressure freezer Wohlwend HPF Compact 01 | Engineering Office M. Wohlwend GmbH | https://www.wohlwend-hpf.ch/index.html | |
| Liquid nitrogen tank | H.Erben GmbH | 94500 | |
| Nitrogen (fluid) | VWR international GmbH | - | |
| NMA | Electron Microscopy Sciences | 19000 | |
| Osmium tetroxide | ChemPur Feinchemikalien und Forschungsbedarf GmbH | 006051 | |
| Oven 120 °C | Binder GmbH | 9010-0194 | |
| PELCO easi-Glow | Ted Pella, Inc. | 91000 | |
| Poly-L-Lysin 0.1% w/v aq. Solution | Ted Pella, Inc. | 18026 | |
| Safe-Lock Tubes 0.5 mL | Eppendorf SE | 0030 121.023 | |
| Sample containers (Beem capsules) | Plano GmbH | G360-1 | |
| Sapphire disc dia. 3 x 0.16 mm | Engineering Office M. Wohlwend GmbH | 500 | |
| Ultramicrotome Leica EM UC7 | Leica Microsystems GmbH | https://www.leica-microsystems.com/de/produkte/em-probenvorbereitung/p/leica-em-uc7/ | |
| Uranyl acetate | Riedel de Haën AG | 31697 | |
| UV light chamber | Dinies | ELG100S |
Request permission to reuse the text or figures of this JoVE article
Request Permission