Label-free 3D-CLEM Using Endogenous Tissue Landmarks

Abstract

Emerging 3D correlative light and electron microscopy approaches enable studying neuronal structure-function relations at unprecedented depth and precision. However, established protocols for the correlation of light and electron micrographs rely on the introduction of artificial fiducial markers, such as polymer beads or near-infrared brandings, which might obscure or even damage the structure under investigation. Here, we report a general applicable "flat embedding" preparation, enabling high-precision overlay of light and scanning electron micrographs, using exclusively endogenous landmarks in the brain: blood vessels, nuclei, and myelinated axons. Furthermore, we demonstrate feasibility of the workflow by combining in vivo 2-photon microscopy and focused ion beam scanning electron microscopy to dissect the role of astrocytic coverage in the persistence of dendritic spines.

Keywords: Biological Sciences Research Methodologies; Biological Sciences Tools; Neuroscience; Techniques in Neuroscience.

Graphical abstract

Figure 1

Availability and Precision of Endogenous CLEM Landmarks (A–C) Blood vessels (arrowheads), nuclei (magenta) and myelinated axons (yellow; boxed area) can be visualized by DIC (A), CLSM (B) and SCoRe (C) microscopy. (D) Overlay of DIC, CLSM, and SCoRe images. Blood vessels (arrowheads); nuclei (magenta); myelinated axons (yellow; boxed area). (E–H) 3D reconstructions of blood vessels (E), nuclei (F), and myelinated axons (G) with distance traces between target dendrite and closest landmarks (H). (I) Overlay of the 3D reconstructions of landmarks and the target dendrite. (J) Frequency distribution of correlative landmarks, plotted against their respective distance to the target dendrite. (K) FIB/SEM micrograph of cortical mouse brain tissue clearly represents blood vessels (blue), nuclei (magenta), and myelinated axons (yellow) by their typical shape and contrast. Related to Figures S1–S3.

Figure 2

In vivo and Ex Vivo Light Microscopy for CLEM (A) Cranial window implantation gives optical access to the cortex of the mouse brain. (B) Magnified image section of cranial window in (A). The blood vessel pattern enables the retrieval of previously imaged positions (framed areas: pos 1–pos 3). (C) Maximum intensity projection of in vivo 2-photon image stack of pos 3 (framed in B). Framed area designates the target dendrite at the last imaging time point (day 41). (D and E) Reconstruction of the cortex by alignment of vibratome sections (D), based on the blood vessel pattern, facilitates identification of the brain slice (green) containing the target dendrites of three different positions (pos 1– pos 3) (E). (F) Maximum intensity projection of ex vivo CLSM image stack of pos 3 (D and E). Framed area designates the target dendrite [compare (F) with (C)].

Figure 3

Sample Preparation and Retrieval of Landmarks (A) Vibratome sections are mounted onto a glass slide with a spacer and sealed by a coverslip. After LM, the coverslip and the spacer are removed. Post-fixation (glutardialdehyde, reduced osmium-ferrocyanide-thiocarbohydrazide-osmium (rOTO), uranyl acetate [UrAC]), dehydration and infiltration with epoxy resin are performed in vials. After removal of excess resin and polymerization, the slide is trimmed to appropriate size. The specimen is mounted with colloidal silver onto an aluminum stub, conducted with bridges of colloidal silver, and coated with carbon (15–20 nm) by evaporation. (B and C) Comparison of a bright field (BF) micrograph (B) with a scanning electron micrograph (C) of the selected vibratome slice (Figures 2 D and 2E). At low magnification, changes in morphology are easily recognized (B, C, circles). Intersected blood vessels are visible in both images (arrows), serving as the most prominent landmarks. (D–F) Optical sections of surface near nuclei (D; white circles) can be correlated to intersected nuclei, visible in SEM at higher voltages (20 kV) due to both topographic and material contrast (E; white circles). Superimposition of both signals (DRAQ5: magenta, nuclei; eGFP: green, dendrites) with the SEM image (F) serves as a precise map for localizing the target dendrite in top view (boxed area). Related to Figures S1–S3.

Figure 4

3D alignment of LM and FIB/SEM Tomograms (A) Economic trench milling in several steps: successive decrease in ion beam energy with increasing milling depth. FIB/SEM tomography is performed by eucentric tilting of the specimen to 54° into the coincidence point (inset). The target area is coated with approximately 1 μm platinum by ion beam deposition. Thin tracking lines (tl) and autotune lines (atl) serve for controlling the milling/imaging process (section thickness, focus, astigmatism). (B) High-resolution images (white squares) are taken every 15 nm of milling. In addition, key frames are taken in intervals of 1 μm in z-direction, providing micrographs for fast, “on the fly” 3D correlation of natural landmarks: blood vessels (blue), nuclei (magenta), and myelinated axons (yellow). (C) When reaching the final block face the region of interest (ROI; white square) with the target dendrite (green spot) is defined in x/y using the coordinates of the landmarks derived from the 3D LM data. Blood vessels (blue); nuclei (magenta); myelinated axons (yellow). (D) Superimposition of landmarks (blood vessels, nuclei, and axons; transparent) of the LM reconstructions (black box) with the FIB/SEM reconstructions of the corresponding structures (solid). (E) Preliminary fast reconstructions of several potential dendrites (green) in the target volume (white box) by an automatic labeling algorithm (Magic Wand, Amira™). (F) Usage of myelinated axons (yellow) as correlative marker to identify the target dendrite (green). Related to Video S1.

Figure 5

3D-CLEM of Cortical Tripartite Synapses (A) In vivo 2-photon micrographs of eGFP-labeled apical dendrites of layer V pyramidal neurons in the somatosensory cortex imaged before and during enriched environment. Enriched environment exposure started at day 18 and was continued until end of the imaging period. White arrowheads mark spines that formed newly and remained stable for at least two consecutive imaging time points; gained and lost spines are labeled with green and magenta arrowheads, respectively. Scale bar = 5 μm. (B–D) Comparative juxtaposition of the same dendritic segment recorded by in vivo 2-photon (B), ex vivo CLSM (C) and FIB/SEM microscopy (D). White boxes indicate dendritic spines that were detected in high-resolution CLSM microscopy and FIB/SEM; white arrowheads indicate spines that were detected in all imaging modalities. Scale bar = 2 μm. (E) 3D reconstructed FIB/SEM tomogram of a complete tripartite synapse (A, astrocyte; D, dendrite; M, mitochondrion; PSD, postsynaptic density; Sp, spine; SV, synaptic vesicles). Inset shows a single micrograph of the corresponding FIBS/SEM stack depicting a dendritic spine (green) with associated presynapse (yellow) and astrocyte (purple). Scale bar = 1 μm. Related to Figure S4.