Next generation histology methods for three-dimensional imaging of fresh and archival human brain tissues

Abstract

Modern clearing techniques for the three-dimensional (3D) visualisation of neural tissue microstructure have been very effective when used on rodent brain but very few studies have utilised them on human brain material, mainly due to the inherent difficulties in processing post-mortem tissue. Here we develop a tissue clearing solution, OPTIClear, optimised for fresh and archival human brain tissue, including formalin-fixed paraffin-embedded material. In light of practical challenges with immunostaining in tissue clearing, we adapt the use of cresyl violet for visualisation of neurons in cleared tissue, with the potential for 3D quantification in regions of interest. Furthermore, we use lipophilic tracers for tracing of neuronal processes in post-mortem tissue, enabling the study of the morphology of human dendritic spines in 3D. The development of these different strategies for human tissue clearing has wide applicability and, we hope, will provide a baseline for further technique development.

Fig. 1

Immunofluorescence staining of archival brain tissues. a Upper panel: comparison of formalin-fixed human cerebellar tissue immersed in PBS (left), OPTIClear (middle) and OPTIClear after 5 days of SDS delipidation at 55 °C (right). Lower panel: comparison of 5 mm-thick, formalin-fixed human striatum tissue block after 3.5 months of SDS delipidation at 55 °C, before (left) and (after) OPTIClearing. b Colour depth-coded, Z-stack image of a block of 2 mm-thick cingulate cortex that has been formalin-fixed for 50 years, stained with antibodies against GFAP with an imaging depth of 125.57 µm. The tissue was cleared after SDS treatment for 4 months at 55 °C and cleared in OPTIClear. Scale bar = 50 µm. c Z-stack image of a piece of formalin-fixed, paraffin-embedded midbrain tissue immunostained for GFAP (red) and counterstained with DAPI (blue) after dewaxing and rehydration (Z-stack depth = 20.77 µm)

Fig. 2

Non-immunohistochemical staining with tissue clearing on formalin-fixed brain tissues. a Colour depth-coded image with maximum intensity projection of a piece of 1-mm-thick medulla block stained with cresyl violet and cleared with OPTIClear (z-stack depth = 228.83 µm). Lower image showing a zoomed and tilted view of area outlined by the white box. b A piece of frontal cortical block immunostained for ZO-1 (green) and counterstained with DyLight 649-labelled Lycopersicon esculentum lectin (red). c DiI crystals were inserted into a cerebellar folium to trace the fibres up to the granular layer, followed by tissue clearing with OPTIClear for 6 h at 37 °C. Most of the fibres traced were mossy fibres. The inset shows the gross appearance of the sample with red crystals of DiI inserted within the OPTICleared sample. d Enlarged view of the white boxed region in a. with colour depth-coding (Z-depth 151.09 μm). e Z-stack images of dendritic spine-like projections visualised in areas outlined by white boxes. Scale bars, 1 μm

Fig. 3

Application of 3D immunohistochemistry with next-generation histology. a Maximum intensity projection image of a 9.5 mm-wide, 8.1 mm-tall, 2 mm-thick spinal cord block stained for neurofilament (imaging Z-depth = 462.60 µm). Magnified view of different laminae was shown in lower images (Scale bars = 200 µm). Roman numerals from I to X: laminae of Rexed, where the suffix m and l denotes the medial and lateral divisions, respectively. Aβ, Aδ to III/IV Aβ-type and Aδ-type input fibres from dorsolateral tract to lamina III/IV, Aβ, C to III/IV Aβ-type and C-type input fibres from dorsolateral tract to lamina III and IV of Rexed, df dorsal funiculus, d. root dorsal root, i.c. VIII interconnecting fibres between bilateral laminae VIII of Rexed, IX to v. root: output motor fibres from lamina IX of Rexed to ventral root, lcst lateral corticospinal tract, lf lateral funiculus, lf to VII lateral funiculus fibres input into the lamina VII of Rexed, pst/umt to V propriospinal tracts or upper modulating tracts input into the lamina V of Rexed, v. root ventral root, ?V to lst fibres from the lamina V of Rexed on the right side coursing through the grey matter to form the lateral spinothalamic tract of the left sid, V to df output tracts from lamina V of Rexed into the dorsal funiculus, vf ventral funiculus, vf/vmf to VII input fibres from the ventral funiculus or ventromedial funiculus into the lamina VII of Rexed, vgc ventral grey commissure, vmf ventral median fissureb. A block of 17.6 mm-wide, 9.5 mm-tall, 1.5 mm-thick formalin-fixed cerebellar folium delipidated with SDS for 3.5 months, immunostained for neurofilament, and cleared with OPTIClear. The inset shows the gross appearance of the folium. Overview of immunostaining (right) and detailed view demonstrated by colour depth-coded image (below; Z-depth = 546.00 µm) from the white boxed area on the right.

Fig. 4

3D mapping of the brainstem catecholaminergic system. a. A maximum projection image of a 21.6 mm-wide, 12.2 mm-tall, 1.5 mm-thick slice of midbrain immunostained for tyrosine hydroxylase (TH) with an imaging depth of 1748.36 µm. Zoomed images showing detailed morphology of TH-immunopositive neurons. Highlighted areas are presented with analyses in subsequent subpanels. b–d Reticular formation pictured in a, b maximum projection image with 10 neurons traced and segmented, which are presented in c, based on which Sholl analysis has been performed and the results for neuron [1] shown in d. The results for other neurons is presented in Supplementary Figure 14. The cartesian coordinates (in µm) were labelled for the longest path length with the soma marked as the origin. The number of intersections was colour-coded. Nav is the average number of intersections of the entire arborisation, and is computed by dividing the area under the fitted polynomial function by the maximal radial extent of the neuronal fibres. The critical radius is radial distance from soma where the maximum number of intersections occurs. e The medial border of the sample in the anatomical location of caudal linear nucleus, where the antibody penetration was adequate for image throughout the thickness of the sample. This is exemplified by a fibre (highlighted by arrowheads) travelling caudocranially along the Z-direction. Z-depth colour-coded 3D rendering, scale bar on the left upper corner is 500 µm. f A Z-depth colour-coded 3D rendering of the substantia nigra reticulata demonstrating curliform TH-positive fibres as they are being separated by the descending tracts from the cortex. Some of these has been traced demonstrating fibre lengths and neuronal span up to 3457 µm and 2682 µm, respectively

Fig. 5

3D visualisation and quantification of basal forebrain magnocellular neurons. a Three-dimensional cresyl violet staining for the visualisation of the nucleus basalis of Meynert (nbM) on a 10.8 mm-wide, 5.5 mm-tall, 2.5 mm-thick basal forebrain block (imaging Z-depth = 1120.00 µm). Tissue block was delipidated with SDS for 3 days and stained with cresyl violet for 3 days (gross appearance shown in inset). Image on bottom left showed magnified view of selected areas from top image. Z-positions of the quantified and annotated neurons were colour-coded according to the reference bars, where the numerical values indicate the distances from the imaging objective for each end of the spectrum (bottom right). b–d Quantification of magnocellular neurons in region of interests highlighted in b (white boxes), where the coordinates of each neuron were marked and annotated (c). d 4D scatter plot of all 3528 neurons, the distance of each neuron to its nearest neighbour was calculated and represented with colour coding. More detailed statistics were provided in Supplementary Figure 15. ac anterior commissure, fx fornix, inf infundibulum, nbM nucleus basalis of Meynert

Fig. 6

Flowchart summarising the next-generation histology protocol developed in this paper. The protocol outlined here enables both fresh and archival formalin-fixed paraffin-embedded (FFPE) post-mortem tissues to be processed for tissue clearing. Formalin-fixed tissues can then be stained using immunohistochemical or non-immunohistochemical techniques such as lipophilic dye tracing with DiI or 3D chemical staining with cresyl violet. An optional SDS delipidation step is recommended before 3D immunostaining or chemical staining. Subsequently, tissues are rendered optically transparent with refractive index homogenization using OPTIClear before imaging or proceeding to other studies such as electron microscopy. H&E, haematoxylin & eosin; RI, refractive index; SDS, sodium dodecyl sulphate; TEM, transmission electron microscopy