Electron microscopy using the genetically encoded APEX2 tag in cultured mammalian cells

Abstract

Electron microscopy (EM) is the premiere technique for high-resolution imaging of cellular ultrastructure. Unambiguous identification of specific proteins or cellular compartments in electron micrographs, however, remains challenging because of difficulties in delivering electron-dense contrast agents to specific subcellular targets within intact cells. We recently reported enhanced ascorbate peroxidase 2 (APEX2) as a broadly applicable genetic tag that generates EM contrast on a specific protein or subcellular compartment of interest. This protocol provides guidelines for designing and validating APEX2 fusion constructs, along with detailed instructions for cell culture, transfection, fixation, heavy-metal staining, embedding in resin, and EM imaging. Although this protocol focuses on EM in cultured mammalian cells, APEX2 is applicable to many cell types and contexts, including intact tissues and organisms, and is useful for numerous applications beyond EM, including live-cell proteomic mapping. This protocol, which describes procedures for sample preparation from cell monolayers and cell pellets, can be completed in 10 d, including time for APEX2 fusion construct validation, cell growth, and solidification of embedding resins. Notably, the only additional steps required relative to a standard EM sample preparation are cell transfection and a 2- to 45-min staining period with 3,3-diaminobenzidine (DAB) and hydrogen peroxide (H₂O₂).

Figure 1

Overview of targeted EM using APEX2 and its variants in cultured cells. Schematic depiction of the APEX2 methodology. APEX2 is introduced genetically as a fusion to a protein of interest or a targeting peptide. Cells are then fixed chemically, as in a normal EM preparation, followed by staining with DAB and H₂O₂ for 5–45 min. The DAB reaction product can be visualized by bright-field microscopy to conveniently ascertain whether the staining was successful. Samples are then processed for EM via heavy-metal staining, dehydration, embedding, and sectioning. IMS, mitochondrial intermembrane space.

Figure 2

Bright-field and electron microscopy (EM) images of cultured cells stained by APEX2 and its variants. Asterisks indicate the locations of cells lacking APEX2 staining. In some images, earlier versions of APEX2 (such as APEX or dimeric APX^W41F) were used, and these cases are explicitly stated. Except where indicated, bright-field images show cells immediately after DAB staining, but before OsO₄ staining. All cells were fixed using 2% (vol/vol) glutaraldehyde. (a–c) APEX targeted to the mitochondrial matrix via fusion to an N-terminal-targeting peptide in COS-7 cells. Mitochondria in untransfected cells are visible in b, but lack contrast in the matrix. (d–f) APEX targeted to the nucleus via fusion to a nuclear localization sequence (NLS). (d) COS-7 cells; (e,f) HEK293T cells. The arrowhead in f points to the nuclear membrane. (g–i) Dimeric APX^W41F attached to the plasma membrane, facing cytosol, via fusion to the palmitoylation sequence of GAP-43. Expression in cultured rat hippocampal neurons. The neuronal processes in h appear discontinuous because the neuron is not contained within a single thin section. The arrowhead in i points to an APX-stained plasma membrane. (j–l) APEX targeted to the ER lumen via a KDEL localization sequence in COS-7 cells. The EM images show staining from APEX, whereas the bright-field image shows staining from horseradish peroxidase (HRP). This bright-field image of HRP-stained cells was selected because it clearly demonstrates ER morphology. Similar bright-field results can be obtained with APEX2. The arrowhead in l, which is an enlargement of the red box in k, points to DAB-stained ER. The images shown in k and l are from a sample that was published previously. (m–o) APEX fused to the N-terminal end of the mitochondrial calcium uniporter (MCU), a transmembrane protein of the mitochondrial inner membrane expressed in COS-7 cells. APEX faces the mitochondrial matrix. The arrowhead in o, which is an enlargement of the red box in n, points to APEX staining, which is confined to a subset of sites located between cristae. The images shown in n and o are from a sample that was published previously. (p–r) APEX fused to histone H2B, a chromatin protein in COS-7 cells. (p) The three APEX-stained nuclei exhibit a range of staining intensities because of variability in expression levels with transient transfection. The arrowhead in r points to a nuclear pore complex. APEX staining of chromatin is visible in the nucleus on the left in r. (s–u) APEX2 fused to α-tubulin expressed in cultured rat hippocampal neurons by lentiviral infection. APEX2 is fused to the N-terminal end of α-tubulin and faces the hollow core of the microtubule polymer. (s) A bright-field image after OsO₄ staining and embedding in resin; the OsO₄ staining causes APEX2-negative cells to become slightly dark. (t) A neuron with high APEX2 expression and dark staining is visible. (u) In an enlargement of the red box in t, APEX2-stained microtubules are visible in the dendrite at the center (arrowhead), and microtubules lacking contrast from APEX2 are visible at the top (asterisk). The images shown in t and u are from a sample that was published previously. (v–x) APEX2 fused to β-actin, expressed in cultured hippocampal rat neurons by lentiviral infection. (v) Punctate staining on dendritic spines is visible in a bright-field image after OsO₄ staining and embedding in resin; the OsO₄ staining causes APEX2-negative cells to become slightly dark. (w) APEX2-stained spines are visible along the dendrite at the center. (x) An APEX2-stained spine at high magnification (enlargement of the red box in w), with synaptic vesicles (SVs) visible in the axon on the left. The images shown in w and x are from a sample that was published previously. mag, magnification.

Figure 3

Correlated light and electron microscopy using APEX2 and its variants. In some images, earlier versions of APEX2 (such as APEX or dimeric APX^W41F) were used, and these cases are explicitly stated. (a) Connexin43-GFP-APEX in HEK293T cells. Connexin43 is a gap junction protein. (Left) Green fluorescence reveals the location of the construct in fixed cells before DAB staining. Cells were fixed using 2% (vol/vol) glutaraldehyde. (Middle) After DAB staining, a dark reaction product is visible by bright-field microscopy, corresponding to the precise locations of the GFP fluorescence. Arrows 1–3 point to gap junctions at cell–cell contacts, and arrow 4 indicates an internalized gap junction plaque. (Right) EM image of the same region after embedding and cutting of thin sections. The sample depicted in this panel was published previously. (b) APEX staining of mitochondria and the nucleus to mark two distinct cell populations. Separate pools of HEK293T cells were transfected with either mito matrix–dimeric APX^W41F or nuclear-localized APEX (NLS), lifted, and co-plated into the same dish. (Left) A low-magnification bright-field image of a sample that was embedded and trimmed into a pyramidal shape, immediately before sectioning. The region of interest (arrow) is in the middle of the flat, rectangular region at the top of the trimmed block. (Middle) Higher-magnification bright-field image of the region of interest. Cell 1 lacks APEX staining, cells 2 and 3 contain APEX staining in the nucleus, and cell 4 contains mitochondrial APEX staining. The contrast at the intercellular contact sites between cells 2 and 4, and between cells 3 and 4 (indicated with an asterisk), is not caused by APEX, but instead corresponds to staining from split horseradish peroxidase, a separate EM reporter for intercellular protein–protein interactions. (Right) EM image of the region of interest after sectioning. The sample depicted in this panel was published previously. (c) Correlated bright-field and EM images of a COS-7 cell expressing APEX-histone H2B in the metaphase of mitosis. Cells were fixed using 2% (vol/vol) glutaraldehyde.

Figure 4

Illustration of proper ultrastructure preservation and staining from APEX2 and its variants. In some images, earlier versions of APEX2 (such as APEX or dimeric APX^W41F) were used, and these cases are explicitly stated. (a) COS-7 cells lacking APEX2 staining with poor (left) or good (right) ultrastructural preservation. The image on the left exhibits discontinuous cell density and rupturing of the plasma membrane, whereas the image on the right shows continuous density and intact subcellular structures, including mitochondria, microtubules, and ER tubules. (b) EM images of gap junctions stained by APEX fused to connexin43. The image on the left is overstained, resulting in signal saturation. The image on the right shows controlled APEX staining (1-min DAB reaction time), resulting in tight localization of electron density to the gap junction and no saturation of signal. (c) EM images of mitochondria. (Far left) Mitochondrion from a cell lacking APEX. (Second from left) Mitochondria stained by APEX localized to the matrix subcompartment. The intermembrane space is light and unstained by APEX. (Second from right) Cell overstained by dimeric APX^W41F in the mitochondrial matrix, leading to blurriness and poor definition of the mitochondrial membranes. (Far right) A mitochondrion that was badly overstained by dimeric APX^W41F, leading to destruction of cellular ultrastructure and a hole lacking electron density.