We present a protocol to study the distribution of mitochondria and endoplasmic reticulum in whole cells after genetic modification using correlative light and volume electron microscopy including ascorbate peroxidase 2 and horseradish peroxidase staining, serial sectioning of cells with and without the target gene in the same section, and serial imaging via electron microscopy.
Cellular organelles, such as mitochondria and endoplasmic reticulum (ER), create a network to perform a variety of functions. These highly curved structures are folded into various shapes to form a dynamic network depending on the cellular conditions. Visualization of this network between mitochondria and ER has been attempted using super-resolution fluorescence imaging and light microscopy; however, the limited resolution is insufficient to observe the membranes between the mitochondria and ER in detail. Transmission electron microscopy provides good membrane contrast and nanometer-scale resolution for the observation of cellular organelles; however, it is exceptionally time-consuming when assessing the three-dimensional (3D) structure of highly curved organelles. Therefore, we observed the morphology of mitochondria and ER via correlative light-electron microscopy (CLEM) and volume electron microscopy techniques using enhanced ascorbate peroxidase 2 and horseradish peroxidase staining. An en bloc staining method, ultrathin serial sectioning (array tomography), and volume electron microscopy were applied to observe the 3D structure. In this protocol, we suggest a combination of CLEM and 3D electron microscopy to perform detailed structural studies of mitochondria and ER.
Mitochondria and endoplasmic reticulum (ER) are membrane-bound cellular organelles. Their connection is necessary for their function, and proteins related to their network have been described1. The distance between the mitochondria and ER has been reported as approximately 100 nm using light microscopy2; however, recent super-resolution microscopy3 and electron microscopy (EM)4 studies have revealed it to be considerably smaller, at approximately 10-25 nm. The resolution achieved in super-resolution microscopy is lower than EM, and specific labeling is necessary. EM is a suitable technique to attain a sufficiently high-resolution contrast for structural studies of the connections between mitochondria and ER. However, a disadvantage is the limited z-axis information because the thin sections must be approximately 60 nm or thinner for conventional transmission electron microscopy (TEM). For sufficient EM z-axis imaging, three-dimensional electron microscopy (3DEM) can be used5. However, this involves the preparation of hundreds of thin serial sections of whole cells, which is very tricky work that only a few skilled technologists have mastered. These thin sections are collected on fragile formvar film-coated one-hole TEM grids. If the film breaks on one gird, serial imaging and volume reconstruction is not possible. Serial block-face scanning electron microscopy (SBEM) is a popular technique for 3DEM that uses destructive en bloc sectioning inside the scanning electron microscope (SEM) vacuum chamber with either a diamond knife (Dik-SBEM) or a focused ion beam (FIB-SEM)6. However, because those techniques are not available at all facilities, we suggest array tomography7 using serial sectioning and SEM. In array tomography, serial sections cut using an ultramicrotome are transferred to a glass coverslip instead of a TEM grid and visualized via light microscopy and SEM8. To enhance the signal for backscatter electron (BSE) imaging, we utilized an en bloc EM staining protocol employing osmium tetroxide (OsO4)-fixed cells with osmiophilic thiocarbohydrazide (TCH)9, enabling us to obtain images without post-embedding double staining.
Additionally, the mitochondrial marker SCO1 (cytochrome c oxidase assembly protein 1)– ascorbate peroxidase 2 (APEX2)10 molecular tag was used to visualize mitochondria at the EM level. APEX2 is approximately 28 kDa and is derived from soybean ascorbate peroxidase11. It was developed to show the detailed location of specific proteins at the EM level in the same way that green fluorescent protein-tagged protein is used in light microscopy. APEX2 converts 3,3' -diaminobenzidine (DAB) into an insoluble osmiophilic polymer at the site of the tag in the presence of the cofactor hydrogen peroxide (H2O2). APEX2 can be used as an alternative to traditional antibody labeling in EM, with a protein localization throughout the depth of the entire cell. In other words, the APEX2-tagged protein can be visualized by specific osmication11 without immunogold labeling and permeabilization after ultra-cryosectioning. Horseradish peroxidase (HRP) is also a sensitive tag that catalyzes the H2O2-dependent polymerization of DAB into a localized precipitate, providing EM contrast after treatment with OsO4. The ER target peptide sequence HRP-KDEL (lys-asp-glu-leu)12 was applied to visualize ER within a whole cell. To evaluate our protocol of utilizing genetic tags and en bloc staining with reduced osmium and TCH (rOTO method), using the osmication effect at the same time, we compared the membrane contrast with and without the use of each genetic tag in rOTO en bloc staining. Although 3DEM with array tomography and DAB staining with APEX and HRP have, respectively, been utilized for other purposes, our protocol is unique because we have combined array tomography for 3DEM and DAB staining for mitochondria and ER labeling. Specifically, we showed five cells with and without APEX-tagged genes in the same section, which aided in investigating the effect of the genetic modification on cells.
1. Cell culture with patterned grid culture dish and cell transfection with SCO1-APEX2 and HRP-KDEL plasmid vector
2. Light microscopy of cells growing on patterned culture dishes and DAB staining for APEX2 and HRP
3. Sample preparation for the EM block
4. Serial sectioning and mounting on indium-tin-oxide coated coverslips for SEM imaging
5. Imaging in the SEM and alignment of the SEM image stack
6. Segmentation of mitochondria and ER from serial images
Figure 1 describes the schedule and workflow for this protocol. The protocol requires 7 days; however, depending on the time spent on SEM imaging, this may increase. For cell transfection, the confluency of the cells should be controlled so as not to cover the bottom of the entire grid plate (Figure 1A). A high cell density could prevent the identification of the cell of interest during light microscope and EM observation. We used genetically tagged plasmids that expressed APEX2 and HRP to select efficiently transfected cells among the numerous cultured cells in the culture dish. We cultured HEK293T cells and confirmed the expression of SCO1-linked APEX2 (mitochondrial intermembrane space [IMS]) and HRP-conjugated KDEL (ER) in co-transfected cells. Under light microscopy, APEX2-transfected cells were stained a brown color, whereas cells without transfection remained unstained (Figure 1A and Figure 2). This allowed the identification of the transfected target cells, which were then used for correlative light-electron microscopy (CLEM), using DAB staining in a cultured cell population (Figure 3D-F and Figure 4B). It was helpful to mark the glass bottom (Figure 1B,C) to make it easy to identify the location of the target cell during the flat embedding step (Figure 1E,F). When the HEK293T cells were treated with an enhanced “double osmium” staining protocol (rOTO), whole cells were stained a dark color (Figure 1D). After removing the gridded coverslip from the culture dish, we identified the target location on the block surface under a stereo microscope (Figure 1G). We trimmed the cells in the ROI in a trapezoid shape, and the leading and trailing edges were made parallel (Figure 1H,I). To implement mitochondria and ER network reconstruction, we used a large diamond knife with a large water boat to serially section SCO1-APEX2 and HRP-KDEL-expressing HEK293T cells (Figure 1J-L). Serial 90-nm thick ribbon sections were successfully attached to the ITO-coated glass coverslip (Figure 1M), and the surface was coated with 10-nm thick carbon for observation via SEM (Figure 1N).
SCO1-APEX2 and HRP-KDEL proteins generate highly dense electronic signals derived from DAB conversion that are detectable in TEM (Figure 3) and SEM (Figure 4). The dark stain generated by SCO1-APEX2 was observed exclusively in the IMS and not in the matrix space of mitochondria (Figure 3D). Co-transfected cells (the left cell in Figure 3D,E) with both SCO1-APEX2 and HRP-KDEL plasmids expressed a highly dense electron signal in mitochondrial IMS and ER; however, we observed no ER staining in cells that were transfected only with SCO1-APEX2 (the right cell in Figure 3D,F). For serial images using SEM, first, we created an overview of the whole array image using the BSE detector over a large area (Figure 4A). Second, the ROI was placed in the first section and propagated to all other sections (Figure 4B). Finally, we visualized the ROI containing five target cells with 5-nm image pixels (Figure 4C). Zoomed-in images revealed detailed subcellular structures (Figure 4D) such as Golgi apparatus, mitochondria, nuclei, and ER. The serial images clearly showed that ER-mitochondria contacts were occurring on different z-planes (Supplemental Movie 2 and Supplemental Movie 3).
Figure 1: Sample preparation workflow for SEM and TEM. (A) A culture dish containing gridded coverslips was seeded with cells and stained with DAB. (B,C) After DAB staining, a marker pen was used to mark the bottom of the glass where the target cells were located. (D) After OsO4 staining, the cells become a dark color. (E,F) Polymerase chain reaction (PCR) tubes (or any type of embedding capsules) were used to easily make an EM block that contained the marked positions. (E) Top view. (F) Bottom view. (G) A low-magnification stereo microscopy image of the surface of an EM block. (H) The ROI is in the middle of the flat surface. (I) A higher-magnification stereo microscopy image of the rectangular ROI. (J) ITO-coated glass coverslips (white asterisk) are inserted into the ribbon carrier (blue arrow), and the carrier is clamped by turning the handle clockwise. (K,L) The handle of the holder is carefully pushed, and the carrier is adjusted to position the edge of the ITO-coated glass coverslip close to the knife by sliding. (M) The ribbons are attached to the ITO-coated glass coverslips. (N) The ITO-coated glass coverslips are attached to the SEM stubs, and the residual glass is sealed with sticky carbon tape and coated with a 10-nm carbon thread layer. White asterisks indicate the ITO-coated glass coverslip, and black asterisks indicate the carbon tape. Please click here to view a larger version of this figure.
Figure 2: Inverted phase-contrast microscopy image of cultured HEK293T cells stained with DAB. (A) Staining of cells with DAB only (without OsO4 staining). White arrows indicate the unstained cells, and black arrows indicate DAB-stained cells. (B) Higher-magnification image of the ROI. Please click here to view a larger version of this figure.
Figure 3: TEM imaging of HEK293T cells exhibiting the targeted mitochondrial IMS (SCO1-APEX2) and ER (HRP-KDEL). (A-C) Untransfected HEK293T cells showing the double-membrane of mitochondria (M) and endoplasmic reticulum (ER). (D-F) APEX2 and HRP catalyze the polymerization of DAB into a local precipitate, which is subsequently stained with electron-dense OsO4. A dark contrast is apparent in the mitochondrial IMS (black arrowhead) and ER (black arrow); however, cells that were not transfected with HRP–KDEL exhibit unstained ER (white arrow). Scale bars: 1 µm. Please click here to view a larger version of this figure.
Figure 4: Serial SEM imaging of HEK293T cells exhibiting the targeted mitochondrial IMS (SCO1-APEX2) and ER (HRP-KDEL). (A) Overview of the serial section ribbons observed using the BSE detector. (B) Correlation of low-magnification image (inset) with high-magnification BSE image (white dotted-line box indicates the ROI of DAB-stained cells). (C) High magnification of the ROI target cells with 5-nm image pixels. (D,E) A dark contrast is apparent in the mitochondrial IMS and ER but not the Golgi apparatus. (F–I) ER-mitochondria contacts (white arrow) occur on different z-planes. N, nucleus; M, mitochondria; ER, endoplasmic reticulum; G, Golgi apparatus. Please click here to view a larger version of this figure.
Table 1: Solution recipes.
Supplemental Movie 1: SEM image stack. Fiji13 with TrakEM14 software was used to align 91 images. The original aligned data set is 11 GB. To downsize the stack, resized and cropped image set was used. Please click here to view this video. (Right-click to download.)
Supplemental Movie 2: Cropped image stack. To visualize mitochondria, ER, and their contact sites in detail, images were cropped from original data set (5 nm/pixel). Scale bars: 1 µm. Please click here to view this video. (Right-click to download.)
Supplemental Movie 3: 3D Reconstruction of mitochondria and ER. For 3D visualization, the contour of mitochondria and ER was segmented and visualized using IMOD15 software. Mitochondria were visualized as long tubular structures (red), and the ER networks (green) showed their complicated morphology. Yellow represented a large surface area of contact site between mitochondria and ER in different z-planes. Please click here to view this video. (Right-click to download.)
Determining the cellular localization of specific proteins at a nanometer resolution using EM is crucial to understand the cellular functions of proteins. Generally, there are two techniques to study the localization of a target protein via EM. One is the immunogold technique, which has been used in EM since 1960, and the other is a technique using recently developed genetically encoded tags16. Traditional immunogold techniques have employed antibody-conjugated gold particles or quantum dots to show the location of the labeled protein. However, due to the requirement for high-quality antibodies and the penetration efficiency of antibodies affected by resin and fixative, this technique is significantly limited17. Specifically, because immunogold labeling is predominantly restricted to the surface of an ultrathin section without en bloc metal staining and strong osmium fixation, this technique is not directly applicable to modern 3DEM18. To use recent 3DEM methods, including SBEM and array tomography, with protein labeling, we utilized genetically encoded EM tags in this protocol. Genetically encoded tags do not require permeabilization, technically demanding ultra-cryosectioning, and immunostaining of individual sections because they localize to the site of interest prior to fixation.
The procedures for sample preparation in 3DEM generally include a combination of common chemical fixation and heavy metal staining methods because cells are composed mainly of C, H, O, and N, requiring staining with heavy metals to acquire contrast under EM9. Therefore, we employed reduced osmium fixation and metal staining to enhance contrast and conductivity for serial imaging. The procedure to stain samples before sectioning, known as en bloc staining, has been reported as an essential step for 3DEM methods such as SBEM and FIB-SEM19. We confirmed that the en bloc-stained cells in our protocol demonstrated clear SCO1-APEX2 and HRP-KDEL EM contrast exclusively in the mitochondrial IMS and ER, respectively, in TEM (Figure 3). Furthermore, the SEM images from 90-nm serial sections revealed a clear contrast in both organelles (Figure 4). Notably, the enhanced contrast by en bloc staining was distinctly distinguishable from the DAB signal, and the contrast and conductivity resulted in good-quality serial images (Figure 4). Additionally, this high contrast aids the facilitation of subsequent tasks such as alignment and segmentation with three-dimensional (3D) image software.
In recent years, volume electron microscopy techniques (dik-SBEM, FIB-SEM, and array tomography) have answered biological questions that required the observation of a large field of view and a 3D view. Dik-SBEM and FIB-SEM do not involve the physical handling of sections, so time consuming for alignment of images can be reduced. However, the sample has to be destroyed to obtain serial images, and the field of view is smaller than that of array tomography. Serial SEM imaging using array tomography is employed increasingly as an alternative to TEM serial sectioning, and the major advantage of this technique is its non-destructive manner and large field of view. Unlike other 3DEM techniques such as dik-SBEM and FIB-SEM, sections can be stored on a coverslip, an ITO-coated coverslip, a silicon wafer, or a tape and can be repeatedly imaged7.
APEX2 is easy to use and can give a wide range of staining densities without special equipment, unlike mini singlet oxygen generator20 or fluorescent protein21 techniques, generating DAB precipitation via a photooxidation. Its variable application has been tested in several cellular organelles including nucleus, plasma membrane, mitochondria matrix, mitochondrial cristae, ER, tubulin, and actin in COS 7 and HEK293T cell22. However, there are some limitations and several checkpoints for the use of genetically encoded tagging in electron micrographs. The expression levels of the exogenous genes should be controlled to achieve reasonable staining at EM level because if the expression of the genetic tag is too high, it may induce false-positive signals and perturb the ultrastructure of cells via membrane rupture and subcellular organelle aggregation23. Another possible problem is DAB overstaining, which has been reported to cause blurriness and membrane destruction22. To ensure the appropriate expression of genes, DAB-stained cells were compared with unstained cells using both light microscopy and EM (Figure 2 and Figure 3). Additionally, the fixation level should be well regulated to ensure that the endogenous oxidases are fully inactive. To prevent any artifacts from endogenous oxidases during processing, we fixed a monolayer of cells instead of using detached and pelleted cells24. This also helped to identify the cells of interest when we checked the DAB signal under light microscopy. Thus, our results indicate that the staining and fixation of cell monolayers are useful for CLEM using DAB staining (Figure 2). The serial images with array tomography and 3D model revealed that the ER–mitochondria contact sites occur on different z-planes (Supplemental Movie 2 and Supplemental Movie 3). The image produces a complete 3D visualization of the mitochondrial and ER networks in whole cells (Supplemental Movie 1). It suggests 3D volume analysis is essential for quantitative comparison of ER-mitochondria contacts. When studying the complex networks of intracellular organelles, this is an exceptionally useful technique.
In conclusion, this protocol was an efficient combination of CLEM and 3DEM techniques that allowed whole-cell investigation at EM level. Notably, two different tags at the same time and DAB signals in two different organelles were visible in a whole cell. In addition, labeled cells and unlabeled cells in same section can be compared, because of large scale EM. In this protocol, en bloc staining and DAB signals from genetically encoded tags were useful to investigate the interaction between membranous organelles in whole cells. This could be a suitable application for large-scale EM to investigate other cellular interactions.
The authors have nothing to disclose.
This research was supported by KBRI basic research program through Korea Brain Research Institute funded by Ministry of Science and ICT (19-BR-01-08), and National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT)(No. 2019R1A2C1010634). SCO1-APEX2 and HRP-KDEL plasmids were kindly provided by Hyun-Woo Rhee (Seoul National University). TEM data were acquired at Brain Research Core Facilities in KBRI.
Glutaraldehyde | EMS | 16200 | Use only in fume hood |
Paraformaldehyde | EMS | 19210 | Use only in fume hood |
Sodium cacodylate | EMS | 12300 | |
Osmium tetroxide 4 % aqueous solution | EMS | 16320 | Use only in fume hood |
Epon 812 | EMS | 14120 | EMbed 812- 20 ml/ DDSA- 16 ml/ NMA- 8 ml/ DMP-30 – 0.8 ml |
Ultra-microtome Leica ARTOS 3D | Leica | ARTOS 3D | |
Uranyl acetate | EMS | 22400 | Hazardous chemical |
Lead citrate | EMS | 17900 | |
35mm Gridded coverslip dish | Mattek | P35G-1.5-14-CGRD | |
Glow discharger | Pelco | easiGlow | |
Formvar carbon coated Copper Grid | Ted Pella | 01805-F | |
Hydrochloric acid | SIGMA | 258148 | |
Fugene HD | Promega | E2311 | |
Glycine | SIGMA | G8898 | |
3,3′ -diaminobenzamidine (DAB) | SIGMA | D8001 | Hazardous chemical |
30% Hydrogen peroxide solution | Merck | 107210 | |
Potassium hexacyanoferrate(II) trihydrate | SIGMA | P3289 | |
0.22 um syringe filter | Sartorius | 16534 | |
Thiocarbonyldihydrazide | SIGMA | 223220 | Use only in fume hood |
Potassium hydroxide | Fluka | 10193426 | |
L-aspartic acid | SIGMA | A9256 | |
Ethanol | Merck | 100983 | |
Transmission electron microscopy | FEI | Tecnai G2 | |
Indium tin oxide (ITO) coated glass coverslips | SPI | 06489-AB | fragil glass |
Isopropanol | Fisher Bioreagents | BP2618-1 | |
Diamond knife | Leica | AT-4 | |
Inveted light microscopy | Nikon | ECLipse TS100 | |
Scanning electron microscopy | Zeiss | Auriga |