Scalable cytoarchitectonic characterization of large intact human neocortex samples

We describe MASH (Multiscale Architectonic Staining of Human cortex): a simple, fast and low-cost cytoarchitectonic labeling and optical clearing approach for human cortex samples, which can be applied to large formalin fixed adult brain samples. A suite of small-molecule fluorescent nuclear and cytoplasmic dyes in combination with new refractive index matching solutions allows deep volume imaging. This enables highly scalable human neocortical cytoarchitecture characterization with a large 3D scope.

Optical volume imaging of human cerebral cortex is challenging at the cellular scale owing to the large size of the human brain and the 3-dimensional geometry of the cortex. The 2-4mm thick cortical sheet is highly curved and packed with billions of neurons, organized in layers each hundreds of micrometers thick. Traditionally, studies on human cortical cytoarchitecture have been performed on sections with a thickness of less than 100µm. However, thin sections have no clear geometric relation to the curved cortical sheet, mostly slicing it non-orthogonal to the layer organization. Due to shape distortions and tearing inherent to the sectioning process, serial sections are extremely difficult to align post-hoc into a volume dataset.
Although a recent surge of optical clearing techniques has transformed microscopic 3D imaging of small transgenic or antibody stained rodent brains [1][2][3][4][5][6] , translation of these techniques to the much larger adult human brain has remained a challenge.
More specifically, volume imaging and cytoarchitectonic characterization of large, adult formalin fixed brain samples, scalable in terms of time and cost to thousands of cubic millimeters in order to cover a significant portion of a human cortical area, has so far remained out of reach.
Here, we report MASH (Multiscale Architectonic Staining of Human cortex): a novel scalable nuclear and cytoplasmic labeling and optical clearing approach.
MASH is suitable for 4-5mm thick archival (i.e. formalin fixed and long-term stored) adult human cortex samples and enables high-throughput deep 3D optical imaging.
MASH consists of two innovations: 1) a set of low-cost small-molecule fluorescent dyes and cleared tissue cytoarchitecture labelling protocols (MASH dye protocols) and 2) a set of adjustable refractive index matching solutions (MASH RIMS) which enable the use of the MASH dye protocols in cleared tissue. For the MASH dye protocols, we identified four small organic compounds: acridine orange (AO), methylene blue (MB), methyl green (MG) and neutral red (NR), previously used as cytoplasmic and nuclear labels in traditional light microscopy studies e.g. 7,8 . We developed adapted protocols for their use as fluorescent labels in large cleared human brain specimen: MASH-AO (green spectrum cell-body label), MASH-NR (red spectrum cell-body label), MASH-MB (far-red spectrum cell-body label) and MASH-MG (far-red spectrum cell-nucleus label). In order to apply the MASH dye protocols in a wide range of human cortex samples, the clearing process must be: 1) potent enough to clear highly myelinated adult human brain tissue up to 4-5 mm thickness within reasonably short time, 2) compatible with MASH dye protocols and, ideally, 3) applicable to archival samples available from brain banks and other academic and clinical tissue storing facilities. The DISCO family of clearing protocols 3, 5, 9 have short clearing times and they have been applied to freshly frozen 10  We demonstrated that MASH can clear and label thick (4-5mm) archival adult cortex samples and that cytoarchitecture can be imaged at a variety of wavelengths, depths and magnifications (Fig. 1). Volume imaging can be performed in the green spectrum (MASH-AO, Fig. 1a,b,d; Supplementary Video 1), red spectrum (MASH-NR, Fig. 1c,g,i-k) and far-red spectrum (MASH-MB, MASH-MG, Fig. 1e,f; Supplementary Video 5). This imaging can be performed at sub-micron resolutions using Two-photon microscopy (TPM) to delineate single neuron cell-body morphology and cortical layer borders (Fig. 1a-d). Likewise, imaging can be performed over larger fields-of-view up to the entire cortical sheet with light sheet fluorescence microscopy (LSFM; Fig. 1e,f). Clearing and staining can be performed on 5mm thick samples in a matter of 10 days and result in imaging of cytoarchitecture with low background and high signal over the full imaging depth (Fig. 1g-k). MASH cell body labelling is well co-localized with standard Nissl stain cresyl violet (CV) in thin sections showing its labeling specificity and suitability for cytoarchitecture characterization ( Fig. 1l-n). Labeling specificity was further validated by verifying colocalization with the DAPI nuclear stain and bright-field cytoplasmic stains ( Supplementary Fig. 1,2).
We demonstrated the efficacy of MASH in high-throughput 3D LSFM characterization of human cortical cytoarchitecture by applying it to large (~40x30x5mm) archival samples surrounding the calcarine sulcus (Fig. 2). The samples were large enough to be easily localized in a magnetic resonance imaging (MRI) reconstruction of the entire host occipital lobe and to contain parts of both primary (V1) and secondary (V2) visual cortical areas (Fig. 2a). Clearing was highly effective, rendering the entire 5mm thick samples transparent with a slight amber-like tint (Fig. 2b,g). Labeling could be achieved over the entire depth of the samples with the far-red MASH-MB cell-body stain for the anterior sample ( Fig. 2b-f) and a twocolor labeling with the red MASH-NR cell-body stain and the far-red MASH-MG nuclear stain for the posterior sample ( Fig. 2g-i). Multi-spectral LSFM imaging of layered cytoarchitecture could be performed at low magnification over 10-12mm long stretches of V1 and V2 cortical sheet (Fig. 2h). Higher magnification LSFM produced mesoscale imaging volumes with both the resolution to resolve single neurons and the field-of-view to contain all layers of the cortical sheet (Fig 2c,i; Supplementary Video 4). Planes from these volumes allowed for classification of cortical layering and sub-layering, corresponding to a histological reference atlas 11 . They displayed distinctive cytoarchitectonic features such as the cell-poor layers IVb and V in V1 ( Fig. 2i) and the large pyramidal cells in layer IIIb of V2 ( Fig. 2c). High magnification light sheet imaging ( Fig. 2d; Supplementary Video 2,3) showed soma morphology features of individual neurons in the context of a deep 3D imaging stack ( Fig. 2e) with sufficient resolution for a surface reconstruction of a pyramidal neuron cell body ( Fig. 2f).
Applied together, the MASH dye protocols and MASH RIMS allow clearing and labeling of thick adult human brain samples and deep volume imaging of cytoarchitecture. The entire MASH protocol for clearing and labeling of 5 mm thick samples takes approximately 10 days. MASH dye solution costs are low, less than 1$ per sample. This makes it easily scalable to the investigation of the human brain and feasible to apply in standard lab environments. Moreover, MASH is capable of clearing and labeling adult human archival brain samples, even after prolonged storage in formalin (current samples had been fixed for 14 to 30 months), making it applicable to tissue stored in brain banks rather than being limited to fresh or freshly frozen tissue. The low corrosive MASH RIMS can be adapted to various RIs which potentially allows their application in a wider range of optical clearing protocols.
Furthermore, samples can be stored long-term in MASH RIMS before imaging because they maintain transparency and fluorescence and do not solidify at low temperatures (2°C to 7°C). In the context of smaller and thinner samples, where antibody labels are effective, or in transgenic animal experiments, the MASH labels can serve as counterstains providing highly useful contextual information in complementary spectral bands. Furthermore, they could be combined with deeply penetrating small-molecule pathology labels e.g. 10 , to investigate human disease pathology in its full cytoarchitectonic context. We demonstrated LSFM volume microscopy imaging over a variety of scales with a commercially available light sheet system, both at high resolution and at large field-of-view. MASH can be combined with advances of LSFM technology, such as the dual inverted selective plane illumination microscopy (diSPIM) system geometry 12 , and large-volume image stitching 13 . This opens new possibilities of cellular level volume imaging of entire human cortical subsystems. Combination with other imaging modalities, such as MRI, can additionally provide correlative multi-modal multi-scale data on cytoarchitecture in the human brain.   Table 1). All tissue was manually blocked with anatomical trimming blades, then cut into 2 to 5 mm thick slices in coronal orientation and immediately processed.
MASH protocol for clearing and labelling of human brain samples. For clearing of formalin fixed adult brain tissue an adaptation of the iDISCO+ method 14  These solutions render archival human brain samples highly transparent with a slight remaining amber color (Fig. 2 b,g; Supplementary Fig 5) typical for many solventbased clearing approaches. The transparency achieved with the recently described ECi 16 is similar to the WGO/CA and TDE/CA RIMS (Supplementary Fig. 3).
However ECi has a melting point of 6-8 ⁰C and samples cannot be stored in the fridge once immersed in the liquid. Overall the Overall the TDE/CA RIMS was preferred, because it is even less corrosive for plastic equipment than WGO/CA and had better properties for long-term cold storage (low melting point) than ECi.  Figure 1 a -i). For experiment 2 on MASH-MG, the same procedures were followed, but sections were co-labelled with DAPI as described above for DAPI counterstaining (Supplementary Figure 1 j -l). Quantitative T 2 * estimation was performed by fitting a mono exponential decay model and a 3D surface reconstruction was created using Brain Voyager QX v2.8.
Light-sheet and two-photon microscopic imaging. Before two-photon microscopic imaging (TPM), stained and cleared human brain samples were transferred to a glass petri dish, and a cover slip with water drop was placed on top of each sample. For TPM imaging experiments, a two-photon laser scanning microscope (Leica TCS SP5 MP, Leica Mikrosysteme Vertrieb GmbH, Wetzlar, Germany), equipped with a HCX APO L 20x/1.00W water immersion objective was used. Working distance of the objective was 2 mm and the excitation source was a 140 fs-pulsed Ti:sapphire laser (Chameleon Ultra II, Coherent Inc., Santa Clara, CA, USA), mode-locked at 800 nm.
To avoid photobleaching and tissue damage, laser power was kept at 11% resulting in approx. 25-50 mW, at the sample surface. Images and image stacks were acquired with Leica Application Suite Advanced Fluorescence (Leica Microsystems).
TPM Image acquisition settings are detailed in Supplementary Table 3. Volume rendering was performed using Bitplane IMARIS. Cell body and tissue background signal analysis variation over imaging depth (Figure 1h) was performed in MATLAB using the Open Microscopy Environment (OME) MATLAB toolbox. At each depth of 500µm, 1500µm, 2500µm, 3500µm and 4500µm from the surface of the tissue, three consecutive data planes were used from the imaging stack for analysis (15 planes total). The middle 50% (in the light propagation direction) of each image plane was used to discard data away from the light sheet waist. At each of the five depths, a total of fifteen regions-of-interest (five per plane, three planes) of 3x3 pixels were selected for brightness assessment in both cell bodies and tissue background. The signal for each region-of-interest was taken as the average over the 3x3 pixel area. For both cell bodies and tissue background, boxplots were created depicting mean ('+' sign), median (box center line), 25th and 75th percentiles (box edges) and 9th and 91th percentiles (whiskers) of the n=15 region-of-interest signals at each depth.
Data availability. The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.