3D molecular phenotyping of cleared human brain tissues with light-sheet fluorescence microscopy

Abstract

The combination of optical tissue transparency with immunofluorescence allows the molecular characterization of biological tissues in 3D. However, adult human organs are particularly challenging to become transparent because of the autofluorescence contributions of aged tissues. To meet this challenge, we optimized SHORT (SWITCH-H₂O₂-antigen Retrieval-TDE), a procedure based on standard histological treatments in combination with a refined clearing procedure to clear and label portions of the human brain. 3D histological characterization with multiple molecules is performed on cleared samples with a combination of multi-colors and multi-rounds labeling. By performing fast 3D imaging of the samples with a custom-made inverted light-sheet fluorescence microscope (LSFM), we reveal fine details of intact human brain slabs at subcellular resolution. Overall, we proposed a scalable and versatile technology that in combination with LSFM allows mapping the cellular and molecular architecture of the human brain, paving the way to reconstruct the entire organ.

Fig. 1. Effect of autofluorescence elimination reagents—AR and NeuN antibody penetration in processed human slices.

a, b Comparison of seven different autofluorescence elimination reagents and AR in 500 μm-SWITCH processed slices using LSFM with the excitation light at 405, 488, 561, and 638 nm. For H₂O₂ and CuSO₄, we examined three and two different concentrations, respectively. Statistical analysis (n = 20) was performed between two transformation protocol methods (SWITCH vs SHIELD) (a) and classic SWITCH-processed tissue vs autofluorescence elimination treatments on SWITCH-treated slices (a, b). A Mann–Whitney test was performed (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Error bars show mean ± SD. Abbreviation: AR antigen retrieval, SB Sudan black. c, d Effect of temperature (4 °C, 37 °C) and tissue processing (SWITCH, SHIELD, and SHORT), e, f autofluorescence treatments (SB, NaBH₄, and CuSO₄) on NeuN immunofluorescence labeling. The plot profiles in c–f show the mean intensity ± SD of three different regions.

Fig. 2. Outline and timing of SHORT.

a The entire SHORT protocol, from the whole human brain sample to imaging, requires three weeks. This consists mainly of incubation time, reflected in the ‘Time’ column for each step (b). During days 1-3, SHORT processing slices were performed. On days 4-8, the removal of the lipid takes place. On day 9, peroxide oxygen treatment (1 h) and AR (10 min) were performed. On days 9–15, the primary antibodies were incubated. The secondary antibody incubation was performed during days 15-19. Finally, the samples were acquired using LSFM and, then, reconstructed using Zstitcher. Additionally, the processed samples can undergo multiround staining for proteome analysis.

Fig. 3. Multiplexed staining and antigen preservation of SHORT-processed tissue.

a Labelling efficiency of two neuron markers, PV and CR, acquired by LSFM. Mean and SD were calculated by taking the signal of 8 different neurons and background in three different regions: 0–240 μm corresponding to the maximal light exposure (0 μm), 240–470 μm corresponding to the middle of the slices (350 μm), and 490–700 μm which was opposite to the excitation side (700 μm). Abbreviations: S, signal; B, background. High-resolution images, scale bar = 100 μm. b Different examples of optimized costaining in SHORT processed slices. Maximum intensity projection (MIP) of 20 slices. Downscaled images, scale bar = 100 μm. c Three distinct rounds of immunostaining for probing neurons (NeuN), inhibitory markers (Vip, SST, PV, CR) vasculature (VIM) and glial cells (GFAP). MIP of 70 slices. Downscaled images neuron, scale bar = 100 μm. d SNR was quantified for SST immunolabeling in round 1, round 2, and round 3 (n = 30 neurons). MIP of 70 slices. Downscaled images, scale bar = 100 μm.

Fig. 4. SHORT combined with LSFM enables single cell-resolved imaging of large portions of adult human slices.

a Image of a typical 500 μm-thick adult hippocampus slice before and after SHORT. b Maximum intensity projection (number slices: 140) image shows a mesoscopic reconstruction of the processed hippocampus slice labeled for NeuN (excitation light 488 nm) and (c) GAD67 (excitation light 561 nm), and (d) merged channels. Scale bar = 1 mm. e Higher magnification on the dentate gyrus and (f) example of high-resolution single cell resolved imaging with our custom-made LSFM. Scale bar = 100 μm. (g–i) MIP of 20 slices of Broca’s area labeled for CR (g), SST (h), and NeuN (i). j Merged channels. Scale bar  = 1 mm. k–q MIP images (slice number 20) of the frontal cortex labeled using the nuclear marker DAPI (k), CR (l), SST (m), and NeuN (n). o Merged channels. Scale bar = 4 mm. p Magnified images of the white matter (blue rectangle), characterized by a high distribution of SST-immunoreactive neurons, and (q) grey matter (orange rectangle). Scale bar = 100 μm.

Fig. 5. SHORT allows multiple rounds of staining, imaging, and stripping of large portions of human brain slices.

a Seven different antibodies were used across 3 sequential rounds of immunostaining, with SST employed as fiducial markers in each round. Round 1: PV—SST—VIP; round 2: CR- SST—NeuN; round 3: VIM—SST—GFAP. Scale bar = 1 mm. b The merged image shows a particular of the white and grey matters labeled with VIP, SST, PV, CR, NeuN, GFP, VIM. MIP of 70 slices. Scale bar = 100 μm.