Engineered ascorbate peroxidase as a genetically encoded reporter for electron microscopy

Abstract

Electron microscopy (EM) is the standard method for imaging cellular structures with nanometer resolution, but existing genetic tags are inactive in most cellular compartments or require light and can be difficult to use. Here we report the development of 'APEX', a genetically encodable EM tag that is active in all cellular compartments and does not require light. APEX is a monomeric 28-kDa peroxidase that withstands strong EM fixation to give excellent ultrastructural preservation. We demonstrate the utility of APEX for high-resolution EM imaging of a variety of mammalian organelles and specific proteins using a simple and robust labeling procedure. We also fused APEX to the N or C terminus of the mitochondrial calcium uniporter (MCU), a recently identified channel whose topology is disputed. These fusions give EM contrast exclusively in the mitochondrial matrix, suggesting that both the N and C termini of MCU face the matrix. Because APEX staining is not dependent on light activation, APEX should make EM imaging of any cellular protein straightforward, regardless of the size or thickness of the specimen.

Figure 1. Electron microscopy reporter scheme and characterization of APEX oligomerization state

(A) The APEX reporter, a monomeric and activity-enhanced mutant of pea ascorbate peroxidase (APX), can be genetically fused to any cellular protein of interest (POI). After expression in live cells, the cells are fixed, and a solution of diaminobenzidine (DAB) is overlaid. Upon addition of H₂O₂, APEX, which retains activity in fixative, catalyzes the oxidative polymerization of DAB to generate a cross-linked precipitate. Subsequent staining of the DAB polymer with electron-dense OsO₄ generates EM contrast. (B) Mutations were introduced at the dimer interface of wild-type (wt) APX (from PDB ID 1APX). (C) Mutants of APX were analyzed by gel filtration chromatography. The calculated molecular weight (MW) of wt APX is 28 kDa. As expected, wt APX runs as a dimer (apparent MW 56 kDa). Some mutants also formed higher molecular weight aggregates (MW_app >200 kDa), which may indicate instability. The K14D, E112K double mutant (mAPX, in red) was selected for further characterization. (D) Gel filtration analysis of mAPX, wt APX, and APEX at concentrations ranging from 250 nM to 250 μM. Dimerization of mAPX and APEX is not detected at <10 μM but some dimerization is seen at concentrations >50 μM. For comparison, similar analyses were performed under identical conditions for the fluorescent protein markers mEos2, EYFP, and mApple, as well as miniSOG. Error bars represent the standard deviation of 2–3 independent measurements. For data points with standard deviation values smaller than the height of the marker, no error bars are shown. (E) Imaging wt APX, mAPX, and APEX fusions to connexin43-GFP (C-terminal fusions) in live HEK293T cells. The top row shows GFP fluorescence (not normalized), and the bottom row shows GFP overlay onto the DIC image. Gap junctions could be easily detected for mAPX and APEX fusions (56 and 55% of contact sites between neighboring transfected cells contained GFP-labeled gap junctions, respectively), but not for the wt APX fusion (0.9%), which predominantly displayed fluorescence trapped in the secretory pathway. Scale bars, 10 μm.

Figure 2. Active-site engineering to boost the activity of APEX

(A) Comparison between the active sites of wt HRP and wt APX (from PDB IDs 1H5A and 1V0H, respectively). The heme cofactor is shown in blue. The co-crystallized substrate analogues (benzohydroxamic acid for HRP and salicylhydroxamic acid for APX) are shown in yellow. The HRP active site is lined with aromatic side chains, shown in purple, whereas the APX active site has only a single tryptophan at position 41. In the full HRP structure at left, the four disulfide bonds are rendered in space-filling yellow. Wt APX lacks disulfide bonds. Chemical structures of ascorbate (the natural substrate of APX), DAB (the desired substrate for EM applications), and guaiacol (a model aromatic substrate) are shown below. (B) Kinetic constants of APX mutants engineered to resemble HRP. k_cat and K_M values were measured using a spectrophotometric assay with guaiacol as substrate. (C) DAB polymerization activities of wt APX, mAPX, and APEX expressed in various cellular compartments. HEK293T cells were transfected with the indicated constructs, incubated with or without exogenous heme, fixed, reacted with DAB, then imaged. The top row shows fluorescence of co-transfected nuclear YFP marker. The bottom row shows the brightfield image. DAB polymer appears dark because it absorbs light throughout the visible spectrum. For nuclear-localized constructs, DAB stain blots out the nuclear YFP fluorescence, so bright YFP fluorescence indicates poor APX activity. Scale bars, 25 μm. The graph on the right quantifies the data for cytosolic constructs. For each condition, the mean fraction of transmitted light absorbed was calculated for >60 transfected cells, then averaged together. Wt APX is the most active toward DAB in cells, followed by APEX and then mAPX. For all three constructs, heme addition boosts activity.

Figure 3. Electron microscopy of cellular proteins and organelles with APEX

(A) EM images of the genetic constructs shown at bottom. Additional EM images are shown in Supplementary Figure 6. (1) (left) Low magnification image of a COS-7 cell expressing APEX in the mitochondrial matrix. Compared to neighboring untransfected cells, the APEX-expressing mitochondria give much stronger contrast. (right) High magnification image of a single transfected mitochondrion. (2) COS-7 cell expressing endoplasmic reticulum-targeted APEX-KDEL. (3) COS-7 cells expressing APEX fusion to histone 2B. (two left panels) The zoom shows chromatin detail at the border of the nucleolus. (two right panels) COS-7 cell in metaphase of mitosis. Zoom shows chromosome detail. (4) Vimentin intermediate filaments in a COS-7 cell. APEX enables visualization of individual filaments, and a bead-like pattern with ~20 nm repeat spacing is apparent. (5) (top two panels) Connexin gap junction between two transfected HEK293T cells. The zoom illustrates minimal spread of the DAB reaction product, even in the absence of membrane enclosure (arrow). (bottom four panels) Correlated light and electron microscopy of Cx43-GFP-APEX. Panels show (from left to right): fluorescence image prior to DAB stain, transmitted light image after DAB stain, low magnification EM image, and high magnification EM image. (B) Cartoon presenting the two topology models for the mitochondrial calcium uniporter (MCU) and the predicted EM staining patterns for each model when APEX is fused to either the N- or C-terminus of MCU. (C) EM images showing the DAB staining pattern for MCU-APEX (C-terminal fusion), APEX-MCU (N-terminal fusion), and untransfected mitochondria in COS-7 cells. Both MCU-APEX and APEX-MCU give clear staining in the mitochondrial matrix, while the intermembrane space (IMS) is light. In this experiment, the C32A mutant of APEX was used to eliminate the possibility of disulfide bond formation, but when APEX was used without the C32A mutation, identical results were obtained (data not shown). Additional fields of view for MCU fusion constructs are presented in Supplementary Figure 8. Scale bars, 200 nm. (D) MCU fusions to APEX are functional. Stable HeLa cells with endogenous MCU replaced by the recombinant MCU constructs shown were prepared by lentiviral infection followed by selection in geneticin. MCU-mediated calcium uptake into mitochondria was measured in these cells using Oregon Green Bapta 6F fluorescence. Error bars show the standard deviation from 4–6 independent measurements. shMCU refers to control cells lacking MCU (endogenous or recombinant). shGFP refers to control cells expressing endogenous MCU.

Figure 3. Electron microscopy of cellular proteins and organelles with APEX

(A) EM images of the genetic constructs shown at bottom. Additional EM images are shown in Supplementary Figure 6. (1) (left) Low magnification image of a COS-7 cell expressing APEX in the mitochondrial matrix. Compared to neighboring untransfected cells, the APEX-expressing mitochondria give much stronger contrast. (right) High magnification image of a single transfected mitochondrion. (2) COS-7 cell expressing endoplasmic reticulum-targeted APEX-KDEL. (3) COS-7 cells expressing APEX fusion to histone 2B. (two left panels) The zoom shows chromatin detail at the border of the nucleolus. (two right panels) COS-7 cell in metaphase of mitosis. Zoom shows chromosome detail. (4) Vimentin intermediate filaments in a COS-7 cell. APEX enables visualization of individual filaments, and a bead-like pattern with ~20 nm repeat spacing is apparent. (5) (top two panels) Connexin gap junction between two transfected HEK293T cells. The zoom illustrates minimal spread of the DAB reaction product, even in the absence of membrane enclosure (arrow). (bottom four panels) Correlated light and electron microscopy of Cx43-GFP-APEX. Panels show (from left to right): fluorescence image prior to DAB stain, transmitted light image after DAB stain, low magnification EM image, and high magnification EM image. (B) Cartoon presenting the two topology models for the mitochondrial calcium uniporter (MCU) and the predicted EM staining patterns for each model when APEX is fused to either the N- or C-terminus of MCU. (C) EM images showing the DAB staining pattern for MCU-APEX (C-terminal fusion), APEX-MCU (N-terminal fusion), and untransfected mitochondria in COS-7 cells. Both MCU-APEX and APEX-MCU give clear staining in the mitochondrial matrix, while the intermembrane space (IMS) is light. In this experiment, the C32A mutant of APEX was used to eliminate the possibility of disulfide bond formation, but when APEX was used without the C32A mutation, identical results were obtained (data not shown). Additional fields of view for MCU fusion constructs are presented in Supplementary Figure 8. Scale bars, 200 nm. (D) MCU fusions to APEX are functional. Stable HeLa cells with endogenous MCU replaced by the recombinant MCU constructs shown were prepared by lentiviral infection followed by selection in geneticin. MCU-mediated calcium uptake into mitochondria was measured in these cells using Oregon Green Bapta 6F fluorescence. Error bars show the standard deviation from 4–6 independent measurements. shMCU refers to control cells lacking MCU (endogenous or recombinant). shGFP refers to control cells expressing endogenous MCU.