Correlative confocal microscopy and ATUM-SEM reveals hallmarks of vascular occlusions and extravasation. (A) Schematic drawing of the ATUM-SEM CLEM workflow. Controlled cortical impact (middle) is followed by systemic injection of 30 nm diameter LNDs (magenta) and 30 nm diameter colloidal gold nanoparticles (black). DyLight 649 lectin (green) was injected 5 min before perfusion (1). After fixation, coronal vibratome sections are generated (2) and positioned onto glass slides with only laying the cover slip on top of the section for confocal imaging (3). The vibratome section is recovered by immersion into the petri dish (4). The previously imaged region of interest is dissected (5) and processed for EM including embedding into resin and contrast enhancement (6). Serial ultramicrotomy and tape collection (7) is followed by wafer mounting and SEM imaging (8). (B) Top: Sum projection of a confocal tile scan (lectin, green; LNDs, magenta; left) and corresponding binocular image (right) of the dissected vibratome section. The lesion area is indicated by a dashed line. Scale bar, 200 µm. Middle: Lesion area identified in the overview tile scan (box in top image) is relocated in the sum projection confocal image and in the serial section low resolution SEM (right). Scale bar, 50 µm. The region of interest (ROI, box) is chosen in the confocal image and re-located in the sum projection SEM image. Bottom: Region of interest confocal image (left) and three single SEM micrographs overlaid with the correlated confocal images (right). Scale bar 5 µm. (C) Scheme of the ATUM-SEM strategy for CLEM. A blood vessel of interest (magenta) in an 80 µm thick vibratome section is relocated by screening serial ultrathin sections at low resolution. The target region is reimaged at high resolution (up to 5x5x50 nm). (D) The correlated region in (B) was segmented and reconstructed from the ultrastructural data. Endothelium (ec, dark green), LNDs and gold particles (Au, magenta), monocyte (mc, orange), neutrophil (np, yellow), platelet (pt, blue), fibrin (fi, grey), erythrocytes (ery, red), pericyte (pc, bright green) are shown. Inset: SEM image of vessel lumen filled with equally sized, putative LNDs (magenta arrowhead) and gold particles (black arrowhead) and bigger aggregates thereof. Scale bar, 100 nm. (E-G) Segmented SEM image (top) and three-dimensional rendering thereof (bottom), scale bars 1 µm, showing (E) the extravasation site from of a neutrophil (yellow), (F) extraluminal gold particles (arrows) next to vessel lumen clotted with platelets (blue) and fibrin (grey) and (G) stalled erythrocytes (red) interspersed with colloidal gold particles. (H) Endothelial morphologies from left to right: normal endothelium (dark green) with tight junction (white arrows) and gold particles (magenta arrows); thinned endothelium covered by a pericyte (bright green), swollen endothelium with mitochondria and tight junction. Immune cells (orange), erythrocytes (red), platelets (blue). Scale bars 1 µm.

with 4 supplements. Principle of reversible section attachment on coated tape. (A) Schematic showing that sections on ccKapton mounted on a silicon wafer cannot be recovered for TEM. (B) Schematics of section removal from cpcKapton or cfcKapton. From left to right: The adhesive tape around the selected section (yellow border) is excised by a razor blade (cutting line: yellow dashed lines) and detached from the wafer. The section is detached from the tape by acetone rinsing using a pipette. The section is collected from the acetone bath onto a TEM slot grid. (C) Detachment workflow of a particular selected section (yellow border). From left to right: SEM overview image of a section (scale bar 200 nm); the imaged section mounted on the wafer; the section is removed together with the underlying adhesive tape (cutting line: yellow dashed lines); the section is floating on a water bath of a diamond knife after acetone treatment; slot grid with the section. (D) SEM images of mouse cortex specimens collected onto different Kapton tapes. From left to right: ccKapton; cfcKapton showing a charging artifact (arrow); cpcKapton red with major charging; cpcKapton black without plasma treatment showing section folds; cpcKapton black after intense plasma treatment. Scale bars 50 µm, (E) Photograph of the coating unit with a permanent pen and halfway labelled ccKapton. (F) Photograph of cpcKapton before discharging (top), after mild (middle) and extensive (bottom) plasma discharging. (G) Ultrastructural quality of a cortical tissue section on ccKapton (left) and cpcKapton after plasma treatment (right) imaged at 10x10 nm resolution. Scale bars, 2 µm. (H) Recovery of an ultrathin section. From left to right: SEM overview image of cortex with corpus callosum (scale bar 100 µm). SEM medium resolution image thereof (scale bar, 10 µm). SEM high-resolution image (10 x 10 nm) of a blood vessel cross section (scale bar, 2 µm). TEM image of the same section showing the selected blood vessel (scale bar, 2 µm). High magnification image of a tight junction (arrowheads), scale bar 200 nm. High magnification images of endothelial vesicles (scale bar, 50 nm). Black boxes indicate location of the image to the right. The online version of this article includes the following figure supplements for figure 3:

Figure supplement 1. Formvar coating for reversible section collection.

Figure supplement 2. Optimizing SEM imaging conditions for section detachment.

Figure supplement 3. Contamination of section surface after detachment.

Figure supplement 4. Estimation of distortion between SEM and TEM images.

Correlated ATUM-Tomo of vascular occlusions. (A) Schematic of the ATUM-Tomo workflow. Semi-thick sections are serially imaged by SEM. Targeted sections (yellow) bearing the site of interest (blood vessel, magenta) are selected for ET. (B) Schematic highlighting how ATUM-Tomo and CLEM-ATUM-Tomo bridge scales in resolution and field of view. TEM-tomography enables high resolution e.g. for vesicular structures (diameter ∼30 nm), ATUM the cellular composition and morphology (e.g. neurovascular unit) (diameter ∼30 nm) and confocal provides a large field of view (e.g. highlighting a lesion site in cortex, 3x3 mm). (C) Confocal microscopy images of the traumatic brain injury lesion. Lectin (green), LNDs (magenta). Numbered boxes indicate positions selected for SEM imaging. Scale bars 200 µm, 20 µm (D) Reconstruction of selected vessels as indicated in (C). Scale bar 10 µm. (E) Regions from the vessels 1 (first two rows) and 2 (third row) in (D) selected for ET. SEM image (first column, scale bar 2 µm first and, third row; 5 µm second row), corresponding ET low (second column, scale bar 500 nm) and high magnification images with segmentation (third column, scale bar 100 nm) and three-dimensional reconstruction thereof (fourth column, scale bar 200 nm). First row: Example of particles in endothelial swellings (red arrow gold particle, magenta arrow LND). Second row: Intact tight junctions (white arrow). Third row: Endothelial disruption and gold particle localization in a platelet. Endothelium (green), pericyte (yellow), platelet (blue), basement membrane (cyan), LNDs (magenta), gold nanoparticles (red), tight junctions (white).

Correlated ATUM-Tomo with serial section ET. (A) Schematic of the ATUM-Tomo workflow. Semi-thick sections are serially imaged by SEM. Targeted sections (yellow) bearing the site of interest (blood vessel, magenta) are selected for serial section ET. Tomograms from consecutive ET volumes can be aligned. (B-D) The boxed region in Fig. 3 D (number 2) has been subjected to ET of nine consecutive sections. (Endothelium (green), pericyte (yellow), platelet (blue), basement membrane (cyan), LNDs (magenta), gold nanoparticles (red). Scale bars 1 µm. (B) Non-smoothed, raw reconstructions of nine consecutive SEM images. (C) Corresponding region of nine detached sections subjected to ET and reconstructed. (D) Side view (left) and longitudinal cut view (right) of the smoothed serial section ET reconstruction.

Formvar coating for reversible section collection. Ultrathin (80 nm thickness) (A) and semi-thick (300 nm thickness) (B) sections imaged by SEM (BSD detector, 8 kV). From left to right: higher resolution and images. (A) Irregular Formvar coating can be detected on the first two images of ultrathin sections while (B) the semi-thick sections do not show any background originating from Formvar. High-resolution images of both ultra- and semi-thick sections are free of background from the coat. Scale bars 100 µm (left), 50 µm (middle), 5 µm (right). (C) Photos of the tape coating unit attached to a spraying unit (Lubrimat L60) for dispersion of Formvar onto the tape. Top: The red spraying unit with Formvar reservoir (Falcon tube) is shown as well as the reel-to-reel system for movement of the ccKapton tape. Bottom: The spray nozzle dispersing the Formvar is surrounded by a Plexiglas cylinder that the tape can transverse.

Optimizing SEM imaging conditions for section detachment. (A) Ultrathin (100 nm thick) and (B) Semi-thick (400 nm thick) sections imaged at 10 nm (left) and 5 nm (right) lateral resolution at 5 kV, 4kV beam deceleration. Only sections imaged at 10 nm lateral resolution could be detached from cpcKapton after imaging, the ones imaged at 5 nm resolution were stuck to the tape. Scale bars 2 µm.

Contamination of section surface after detachment. (A) Ultrathin sections imaged by SEM (left) and TEM (right). The TEM image shows contaminations (arrows). Scale bars 2 µm (left), 5 µm (right). (B) Scheme of the section attached to cpcKapton (permanent marker coat, blue) and after collection onto a slot grid. The section side that faced the coating is contaminated (light blue) up to a thickness of 50 nm while the rest of the section thickness (grey) does not show any dirt originating from the coating. Single images of a tomographic reconstruction at different depths, indicated in the scheme (z1-z4) are shown. The first left image (z1) is superficial and the second is a middle one (z2), both without dirt. The third image at 50 nm distance from the side that faced the coating (z3) shows some dirt (arrow) while one of the images from the very bottom (z4) is most contaminated. High magnification images (bottom) of the boxed regions reveal details of the contamination. Scale bars 1 µm (top), 500 nm (bottom)

Estimation of distortion between SEM and TEM images. Ultrathin section SEM and TEM images were compared after rigid (top) and affine (bottom) alignment to assess image distortion. We picked a clean example (A) and one region with some dirt (B). For both examples and alignment algorithms we calculated the Pearson coefficient between the SEM and TEM image. The images show the spatial distribution of colocalization (left, ImageJ ‘Colocalization Finder’) and mislocalization (middle, ImageJ ‘Colocalization Highlighter’). The right images are calculated from the transformation of rigid to affine alignment (ImageJ ‘Image Registration with SIFT’). Green indicates matching, red diverging points.