Unlocking the potential of large-scale 3D imaging with tissue clearing techniques

Abstract

The three-dimensional (3D) anatomical structure of living organisms is intrinsically linked to their functions, yet modern life sciences have not fully explored this aspect. Recently, the combination of efficient tissue clearing techniques and light-sheet fluorescence microscopy for rapid 3D imaging has improved access to 3D spatial information in biological systems. This technology has found applications in various fields, including neuroscience, cancer research and clinical histopathology, leading to significant insights. It allows imaging of entire organs or even whole bodies of animals and humans at multiple scales. Moreover, it enables a form of spatial omics by capturing and analyzing cellome information, which represents the complete spatial organization of cells. While current 3D imaging of cleared tissues has limitations in obtaining sufficient molecular information, emerging technologies such as multi-round tissue staining and super-multicolor imaging are expected to address these constraints. 3D imaging using tissue clearing and light-sheet microscopy thus offers a valuable research tool in the current and future life sciences for acquiring and analyzing large-scale biological spatial information.

Tissue clearing and light-sheet fluorescence microscopy allow for rapid 3D imaging of biological systems, providing insights into various biomedical research. This technology holds promise for enabling 3D spatial omics analysis. Emerging techniques may overcome current limitations, making it a valuable tool for analyzing large-scale biological spatial data.

Introduction

Recent advent of modern tissue clearing and three-dimensional (3D) fluorescence microscopy techniques has offered huge opportunities in analyzing large-scale multicellular systems with unprecedented clarity [1,2]. They facilitate the study of complex tissue architectures and their function in its native 3D context, which is critical for understanding organ development, homeostasis and disease pathogenesis. Neuroscience is the field that has first adopted these technologies to map neuronal circuits and activities that aids in understanding brain function and disorders [3,4]. Visualizing embryogenesis and organogenesis in cleared embryos and organoids also provides insights into tissue morphogenesis and cellular dynamics [5–8]. Cancer researchers also utilize these technologies to detect and analyze micrometastasis, tumor microenvironments and drug distributions [9–14]. 3D pathology has recently started approaching clinics in an effort to expand traditional two-dimensional (2D)-based examinations in 3D, with the goal of improving the accuracy and sensitivity of disease tissue assessment [15–18].

Furthermore, integrating 3D imaging data with various omics technologies, such as single-cell sequencing techniques, can provide a comprehensive view of biological systems by bridging the gap between molecular content information and spatial context information with full coverage. Recent techniques for spatial omics (e.g. Visium, Xenium, GeoMx, CosMx, spatially resolved transcript amplicon readout mapping and Multiplexed Error Robust Fluorescence In Situ Hybridization) have allowed to overlay molecular profiles onto spatial tissue maps, providing correlation of gene expression patterns, protein localization and cellular interactions within the native tissue environment [19]. However, such current spatial omics techniques provide limited coverage in the 2D section. Tissue clearing and 3D imaging technologies also have a potential to expand the axes of omics approach with its competence of collecting and analyzing biological spatial information in 3D [20,21]. The 3D spatial omics will further improve our understanding of cellular heterogeneity and signaling among them in a native biological space, revealing details about complex biological processes and disease mechanisms.

Beyond the analysis of spatial structures and molecular expression in a restricted selection of biological samples, the upcoming biomedical research should broaden its scope to include comprehensive examinations of entire organs and whole organisms. This review discusses the potential of tissue clearing and 3D imaging technologies that can contribute to such future direction. We particularly emphasize large-scale 3D imaging of whole organs/bodies and the emerging field of 3D spatial omics, highlighting their roles in shaping the future of biomedical research.

Whole-organ/body 3D imaging with tissue clearing

Tissue and organ clearing techniques are essential for preparing biological samples for large-scale 3D imaging with optical microscopes. The origins of these techniques can be traced back to the trials in the early 1900s by Lundvall and Spalteholz [22,23]. The early tissue clearing reagent by combining methyl salicylate and benzyl benzoate enabled 3D anatomical observations of human embryos and adult organs. After a long period without significant advancements, Dent reported the use of BABB (a mixture of benzyl alcohol and benzyl benzoate) for clearing frog embryos in 1989 [24], opening the modern tissue clearing applications. In the 1990s, Tuchin and colleagues studied tissue clearing methods from an optical physics perspective [25–27], while Chiang and colleagues developed and commercialized clearing protocols for insect tissues [28].

Light-sheet fluorescence microscopy (LSFM) is a technology with both historical roots and modern applications as well [29]. Interestingly, the development of this unique device has progressed almost in parallel with tissue clearing techniques. In 1902, Siedentopf and Zsigmondy proposed the concept of light-sheet microscopy to observe light scattering in solutions [30]. A modern LSFM with laser light as an illumination source was first reported by Spelman and colleagues, called orthogonal-plane fluorescence optical sectioning. They demonstrated obtaining a 3D image of cleared guinea-pig cochlea [31]. Stelzer and colleagues further introduced Selective Plane Illumination Microscopy (SPIM) and Digital Scanning Light-sheet Microscopy [32,33]. These LSFMs were primarily developed to observe intrinsically transparent biological samples, such as animal embryos, in 3D and 4D.

The integration of these two key technologies for organ-wide 3D imaging began with an epoch-making work by Dodt et al. in 2007 [34]. They reported 3D observations of entire organs with cellular resolution using the BABB-based tissue clearing and a macrozoom LSFM system. Ignited by this work, numerous clearing protocols were developed in the following decade [35–48], primarily classified into three approaches: organic solvent-based reagents, hydrophilic compound-based reagents and tissue hydrogel chemistry. Organic solvent-based methods allow rapid and effective clearing. Hydrophilic compound-based methods excel in safety, retention of fluorescent proteins and adaptability for a wide range of research purposes. Tissue hydrogel chemistry methods fix tissues rigidly, enabling multiplex staining and multimodal observation of proteins and RNA.

Keeping pace with these advancements, several LSFM systems for cleared sample imaging have been developed. Those have particularly focused on solving trade-offs in optics, such as balancing field of view (FOV), high numerical aperture (NA) and long working distance (WD). Generating light sheets with submicron thickness and centimeter-scale width has been another challenge. Dual-view inverted SPIM (diSPIM) and open-top light sheet microscopy place the optical path for image detection above or below the stage, allowing samples to be prepared on conventional glass coverslips and enabling rapid and high-resolution 3D imaging of cm³-scale cleared tissue [49–55]. Axially swept light-sheet microscopy and tiling light-sheet microscopy cover a large FOV with the thinnest part of the light sheet (the beam waist), adapting to large volumetric cleared sample imaging with isotropic micron to submicron resolution [56–58]. Another LSFM with the moving observation and efficient real-time autofocus (MOVIE) system was developed to rapidly obtain entire mouse organ images with subcellular resolution by continuous movements of the sample stage and automatic selection of region-of-interest [59]. To avoid the degradation of image data quality during a long imaging time, several systems incorporated (semi-) automated adjustments of imaging parameters [56,59–61]. Finally, mesoscale SPIM and desktop-equipped SPIM for cleared specimens were developed as open-source LSFM systems for observing cleared tissues, providing relatively easy access to these technologies and opportunities for end users [11,62,63].

Processing LSFM image data typically involves three main steps: preprocessing (including intensity correction and image stitching), feature detection (such as segmentation and cell coordinate detection) and quantification. It is occasionally possible to quantitatively compare organ-wide 3D data from multiple samples by registering and aligning them to a standardized reference [55,64,65]. However, these computational procedures often face challenges due to the large data size, ranging from gigabytes to terabytes. To manage LSFM data effectively, researchers need to consider several factors, such as substantial computational requirements (including large memory and storage capacities, high-speed I/O devices and powerful CPU and GPU resources), specialized software (such as image stitching and AI-driven analysis software for large-volume image data [14,66–69]), and a carefully planned experimental design that optimizes sample size and the required resolution. These considerations are crucial for handling the extensive datasets generated by LSFM and extracting meaningful biological insights.

Recent examples of large-scale 3D imaging and analysis with tissue clearing and LSFM

Beyond routine applications by general end users, several advanced researchers employ these technologies to collect and analyze large-scale 3D data encompassing entire animal bodies, thick primate and human brain sections, or entire human organs. These state-of-the-art examples necessitate the development of a specialized tissue clearing protocol, histological labeling, microscopy systems and efficient data processing workflows.

The whole-body clearing and imaging of animals is one of the ultimate goals of clearing techniques. By 2014, several attempts at whole mouse body clearing had been reported, marking a significant milestone in the early phase of tissue clearing development [70,71]. Recent improvements in tissue clearing, whole-sample labeling and LSFM imaging methods have enabled complete whole-body observation at cellular resolution [47,72–74]. A significant example is wildDISCO [immunolabeling of wild-type mice and DISCO (3D imaging of solvent-cleared organs) clearing] [74], which achieved efficient whole-animal antibody staining by applying β-cyclodextrin to enhance cholesterol extraction and improve tissue penetration of regular IgG antibodies. Using LSFM optimized for large sample imaging, the researchers successfully traced neural projections and mapped lymphatic systems in entire 2.0-cm-thick mouse bodies (Fig. 1a). In addition, they successfully explored and quantified rare pathological lesions, such as tumor-associated tertiary lymphoid structures, on a whole-body scale. The LSFM used in this example was paired with a macro zoom microscope that featured an objective lens with a ×4 magnification and a NA of 0.28. The entire mouse body was captured as 4 million TIFF images, resulting in a data size of approximately 3.5 terabytes when stored as 16-bit images. A Fiji plugin was used to stitch the imaging data. A major current challenge is imaging time, with continuous scanning of an adult mouse’s entire body taking days to weeks.

Fig. 1.

Large-scale tissue clearing and 3D imaging examples. (a) An example of whole mouse body clearing, immunostaining and 3D imaging with wildDISCO [74]. The immunostained signals of pan-neuronal marker PGP 9.5+ with depth color coding shows neuronal projections at different z levels in the 2.0-cm-thick whole mouse body. The figure was adopted under the Creative Commons CC BY license of the original work. (b)–(d) An example of constructing a cellular resolution census of the human brain [55]. The custom-built dual-view LSFM system utilized in this study is illustrated in the panel (b). Mutual light-sheet illumination and image detection could be achieved by two optic paths, enabling isotropic resolution imaging of the planer-shape cleared sample. The panels (c) and (d) depict a representative slice of a 500-μm-thick human brain slice that has been cleared and stained with anti-calretinin (CR) and anti-NeuN antibodies, as well as a nuclear stain, propidium iodide (PI). The cellular resolution of the whole section data is represented by the high-resolution images in (d). The figure was adopted under the publisher’s permission (license number 5837590429309). (e) Multiscale imaging of the expanded whole mouse brain by ExA-SPIM [84]. Sparsely labeled neurons expressing tdTomato were imaged and traced from centimeter (whole brain) to nanometer (individual dendritic spines and axons) scales. (f) The ExA-SPIM microscope system developed in the study. Diffraction-limited and aberration-free imaging over a wide FOV and WD was achieved by incorporating a detection lens and a large-format cMOS camera from the metrology industry. The copyright for the panel (e) and (f) is held by Glaser et al. (2023). The figure was adopted with some modifications under the author’s permission and the Creative Commons Attribution License.

The analysis of entire animal bodies has expanded to include 3D imaging and analysis of human organs [48,75]. Brain tissue requires particularly precise data collection and analysis. Recently, several research groups have been working on clearing and multiscale data collection of human brains [48,54,55,76–79]. Typically, the brain is divided into planar blocks (slabs) of several hundred micrometers to a few millimeters thickness for clearing, autofluorescence elimination and staining. Microscopes using oblique light sheet illumination, such as diSPIM or OTLC, are well-suited for imaging these thick planar tissues. For example, a custom-built dual-view inverted LSFM was used to create a cell census of the Broca’s area in the human cerebral cortex at micrometer resolution [55] (Fig. 1b–d). This device featured two optical axes, each with immersion objective lenses (×12 magnification, NA 0.53, WD 8.5–11 mm) and scientific cMOS cameras. The system achieved 3.3 μm isotropic resolution after postprocessing through the mutual illumination and detection. Data were acquired at 800 megabytes/second at 47 flame per second, quickly reaching terabytes in size. For efficient storage and sharing, the large-scale data were compressed using JPEG2000 lossy compression at a 1:20 ratio. The custom stitching software ZetaStitcher was used for subsequent processing (https://github.com/lens-biophotonics/ZetaStitcher). In all, 15 of 500 μm thick slices (total volume 1.5 cm × 1.3 cm × 0.75 cm) were finally analyzed. The detected neuronal subtypes were registered to an MRI-based atlas coordinate system to generate a regional cell census. The neuron populations in various cortical layers were approximately estimated to be between 20 000 and 85 000 based on this dataset. Collecting more comprehensive human brain data presents ongoing challenges, including slab preparation, clearing, labeling and inter-slab stitching after image acquisition [79].

Morphology analysis of a whole neural cell with long-range neurites necessitates submicron resolution whole-brain imaging without tissue sectioning. While still challenging for human brains, this has been achieved at least in mouse brains. Expansion Microscopy is a unique technique that enhances effective resolution by embedding tissue in a hydrophilic gel and expanding it, enabling visualization of structures below the optical diffraction limit [80–83]. This process also improves tissue transparency. However, expanding large tissue samples like whole organs makes microscope limitations more prominent, particularly trade-offs among FOV, WD and spatial resolution. The expansion-assisted selective plane illumination microscope (ExA-SPIM) was developed to address these limitations and obtain nanoscale resolution data across the entire brain [84] (Fig. 1e and f). This system incorporated optics and a detector developed for the electronics metrology industry. It featured a lens with ×5 magnification, a relatively high NA (0.305) and long WD (35 mm). The detector captures a 10.6 mm × 8.0 mm FOV using a 14 192 × 10 640 pixel sensor (3.76 µm/pixel), offering 38 times more pixels than typical life science cMOS cameras. A tissue expansion protocol was also optimized for centimeter-scale samples (protocol.io dx.doi.org/10.17504/protocols.io.n92ldpwjxl5b/v1). Finally, ExA-SPIM provided a 100-fold larger FOV and 10-fold longer WD compared to standard biological microscopes. The system successfully imaged an entire 3-fold expanded mouse brain in only 15 tiles, which contrasts with conventional LSFM systems requiring over 400 tiles for equivalent resolution. The system achieves an effective resolution of 300 nm laterally and 800 nm axially for cleared and 3-fold expanded brains, generating approximately 100 terabytes of data per brain. This necessitated the development of customized operation software, dedicated hardware and data management pipelines, including high-speed networks, efficient I/O processes, data compression (OME-Zarr format) [85] and machine learning-based analysis. Nevertheless, the researchers successfully imaged and reconstructed single neurons and their long-distance axonal projections in whole mouse brains, macaque motor cortex and human neocortex and white matter.

The capability to analyze large-scale and complex biological 3D information, as shown by these examples, highlights the crucial role of tissue clearing and LSFM technologies in life sciences. On the other hand, significant expertise is required for developing and optimizing sample preparation protocols, imaging devices and large-scale data collection and analysis pipelines. These challenges currently hinder general end users from adopting the advanced tissue clearing-based 3D imaging techniques. Similar to single-cell RNA-Seq and spatial transcriptomics, commercialization and support by regional core facilities may potentially reduce the burdens.

Tissue clearing and 3D imaging as essential for cell-omics and 3D spatial omics

The suffixes -ome and -omics have been frequently used for expressing the comprehensive set of biological information and its large-scale analysis in life sciences [86]. Since the advent of systems biology in the post-genome sequence era [87,88], molecular omics has been central to collecting and analyzing biological information, especially at the molecular level. However, recent advances in spatial omics have emphasized the equivalent significance of tissue structure. Cellomics, a field distinct from molecular omics, enables the collection and analysis of spatial information in biological tissues and organs, encompassing the entirety of all cells in a multicellular system (cellome) [21]. Tissue clearing and LSFM imaging can achieve cellomics through 3D microscopic observation and analysis of all labeled cells in 3D space (Fig. 2a).

Fig. 2.

Expanding tissue clearing and 3D imaging technologies for cellomics and 3D spatial omics. (a) Diagram showing axes of biological information and corresponding omics technologies. Sequencing-based omics (e.g. single-cell RNA-Seq) collect comprehensive molecular information, typically lacking spatial context. Recent spatial omics (e.g. spatial transcriptomics) incorporate some spatial information with limited coverage. Cellomics targets 3D spatial information of whole organs and bodies (cellome), constrained in molecular information due to limited color channels in histology and microscopy. Future 3D omics may be achieved by expanding spatial coverage of current spatial omics and increasing molecular information in imaging-based cellomics. (b) Common reference-based cellomics. For organs with a fixed reference (e.g. anatomical atlas), such as the brain, analysis involves registering whole-organ images to the reference, then quantifying and comparing labeled cells/structures by anatomical regions. A part of the figure materials (brain atlas) was created from the CUBIC-Atlas data [100]. (c) 3D point cloud-based, atlas-free cellomics. For multicellular systems without a common reference, such as organoids, labeled cells are detected and reconstituted as coordinate sets (point cloud). Quantitative analysis follows by collecting structural features of the 3D point cloud, similar to other omics analyses where biological feature matrices (e.g. gene expression matrix) are prepared for bioinformatic analysis. A part of the figure materials (three organoids) was created with BioRender.com. (d) Potential strategies for increasing molecular information in microscopy-based 3D spatial omics.

So far, cellomics approaches have been primarily demonstrated in whole-brain imaging and analysis of neural circuits and activities in mouse brains (Fig. 2b). Neural circuitry can be labeled using adeno-associated virus or rabies virus vectors for anterograde and retrograde analysis, respectively [89]. The labeled neurites or somas are then segmented and quantified according to their distribution in each anatomical atlas region. For comparative analysis, virally labeled multiple brain datasets can be registered and aligned to a common anatomical reference [64,90–93]. Similar workflows have been widely applied to neural activity labeling using whole-brain c-Fos imaging and subsequent comparative analysis. Examples include studies on brain-wide changes in c-Fos expression following administration of MK-801 (an N-methyl-D-aspartate receptor antagonist) [64,72,94], alcohol [95] and weight-lowering drugs [96]. The similar workflows have also been applied for fear conditioning-associated whole-brain activity analyses [97,98]. The Allen Brain Atlas [99] provides a volume-based anatomical reference for the whole mouse brain. Additionally, clear, unobstructed brain/body imaging cocktails and computational analysis (CUBIC)-Cloud was developed as a common coordinate system encompassing all cell positions in the whole mouse brain (over 1 million cells) [100]. This latter reference has demonstrated potential for discovering novel cell populations correlated with specific experimental conditions, independent of predefined anatomical regions [100].

The brain is a unique organ with largely equivalent structures across individuals, enabling the use of a common reference for sample comparison. While not extensively tested yet, an atlas-free approach can also be employed by extracting multidimensional features from the spatial distribution patterns of cells within the sample itself (Fig. 2c). This approach is particularly relevant when analyzing soft tissues (e.g. the intestine), structurally variant tissues (e.g. cancer tissue) [18] or multicellular samples such as spheroids or organoids [5], for which a fixed common reference cannot be defined. Generally, omics data represent features of biological information, such as gene expression levels. In cellomics, 3D cell position information (coordinates) forms a 3D point cloud of the entire tissue or organ sample, serving as another data modality. 3D point cloud technology is increasingly used in engineering fields, such as automobile image analysis, with ongoing research in feature extraction and machine learning-based classification [101]. Similarly, in biological applications, preparation of a multicellular 3D point cloud allows for the collection and analysis of multiple features of cellome-wide spatial information. This enables classification within heterogeneous samples without relying on predefined anatomical references.

To fully realize the potential of tissue clearing and 3D imaging technologies as a 3D spatial omics platform, both spatial context and molecular content information need to be comprehensively collected. Current cellomics approaches are limited to obtaining cell position information (cell coordinates) with only a few molecular profiles from a limited number of imaging color channels. Future developments are expected to introduce new methods to increase the number of detectable molecules (Fig. 2d). Recent studies have reported promising approaches for 3D multi-round immunostaining and fluorescence in-situ hybridization (FISH) for protein and mRNA detection, respectively. Immunostained signals can be bleached/delabeled and re-stained using techniques such as photobleaching [102,103], robust tissue fixation and heating [46,78,83,102] or stripping compounds [46,104]. Such multi-round immunolabeling has even allowed for the detection of dozens of molecules, resolving a general issue regarding cross-talk between the primary antibodies produced by the same host species. A notable example demonstrated 28 neuronal marker labels in a 2-mm-thick hypothalamus slice of the mouse brain over seven staining and imaging rounds [104]. Hybridization Chain Reaction (HCR)-based FISH has also shown promise for mRNA detection in 3D volumes [105–107]. A recent study successfully imaged a whole mouse brain hemisphere after multi-round HCR labeling for 10 different genes [107]. Furthermore, single-molecule RNA FISH-based spatial profiling of gene expression has begun to extend into 3D space, although current coverage is still limited to volumes on the order of µm³ [108].

Super-multicolor imaging, which involves obtaining over 5 to 10 color channels in a single imaging procedure, is another approach to enable multichannel molecular expression profiling. This approach relies on adapting multispectral imaging and unmixing techniques. Several examples demonstrate the potential of this method in 3D spatial omics. For instance, confocal-based multispectral imaging combined with linear unmixing has enabled eight-color volume imaging of whole organoids [109]. In another application, multispectral LSFM, which integrates a camera with a diffractive unit, has achieved five-color imaging and linear unmixing of whole zebrafish larvae expressing fluorescent proteins [110]. Additionally, LSFM with multiple filters has been used for cleared mouse embryo imaging, employing principal component analysis-based unmixing [111]. The phasor-based hyperspectral imaging combined with LSFM was also tested for 3D live imaging, which uses a pair of filters with sine/cosine transmission spectra for decoding [112]. Despite these advancements, achieving dozens to hundreds of color channels in 3D images still requires further development in several areas, including the improvement of multispectral imaging and unmixing methods for organ-wide 3D data, as well as the incorporation of near-infrared or longer wavelengths for excitation.

Realizing tissue clearing-based 3D spatial omics also faces challenges in standardizing protocols and analysis pipelines to manage large volumes of imaging and molecular data reproducibly [20,21]. The success of current molecular omics methods, such as RNA-seq and single-cell RNA-seq, has been supported by efforts to standardize workflows, improving accessibility, reproducibility and comparability. In 3D spatial omics, similar standardization is essential at all stages. The number of existing protocols for tissue clearing and 3D staining apparently need to be consolidated into a few optimized, automated protocols specific to 3D spatial omics. Standardized LSFM imaging and data digitization processes are also crucial for ensuring reproducibility across experiments and enabling the use of common bioinformatics tools, as in the sequencing-based omics. Developing a unified software platform for integrating spatial and molecular data is also critical. This integration involves merging large-scale multichannel images onto the 3D point cloud of cell coordinates, an analytical framework that has not yet been fully discussed in the research community. Machine learning-based approaches show potential for extracting molecular and spatial features from such large-scale image data [68,69], while generating and curating training data will be labor-intensive. Another significant challenge lies in ensuring that features obtained through machine learning are biologically interpretable. Future progress in 3D spatial omics is, at least in part, likely to build on current experiences with 2D spatial omics.

In summary, current molecular omics technologies, such as single-cell transcriptomics, provide comprehensive molecular information about cells and tissues, while they sacrifice spatial context. Recent advancements in spatial omics have begun to address this limitation by incorporating spatial information, albeit within restricted 2D areas. The development of 3D spatial omics is therefore crucial for capturing both the molecular content and spatial context of biological information in its intrinsic 3D form. While challenges persist, the potential of tissue clearing-based 3D spatial omics has been increasingly recognized in the field.

Conclusion

The innovations in tissue clearing and 3D imaging techniques have significantly advanced the study of large-scale multicellular systems. This approach has proven valuable for comprehensive organ- and organism-wide analyses and the extension of various omics methods, as discussed. However, challenges such as managing data volume and complexity, computational resource demands and image analysis difficulties highlight the need for new solutions. Additionally, democratizing the whole workflow remains essential for broader adoption, as previously seen with RNA sequencing technologies. Future research should address these issues while refining existing methods and exploring new applications. As these technologies progress, they offer the potential to reveal new biological insights and significantly advance scientific understanding across multiple fields of biomedical research.

Funding

This study was supported by the Japan Agency for Medical Research and Development (AMED)-Project for Promotion of Cancer Research and Therapeutic Evolution (P-PROMOTE) (JP22ama221517); AMED-Research on Development of New Drugs (JP21ak0101181); AMED-Brain/MINDS (JP21wm0425003); Japan Science and Technology Agency (JST)-CREST (JPMJCR23B7); Japan Society for the Promotion of Science (JSPS)-KAKENHI grant-in-aid for scientific research (B) (JP22H02824); JSPS-KAKENHI Grant-in-Aid for Transformative Research Areas – Platforms for Advanced Technologies and Research Resources ‘Advanced Bioimaging Support’ (JP22H04926); JSPS-KAKENHI for International Leading Research (23K20044); Operating Costs Subsidies for Private Universities, Grants-in-Aid from Nakatani foundation for advancement of measuring technologies in biomedical engineering, The Uehara Memorial Foundation, UTEC-UTokyo, the Takeda Science Foundation and collaboration research funding with Kantum Ushikata Co., Ltd. A part of figure materials in Fig. 2c was created with BioRender.com (agreement number: XB274L06DH).

Conflicts of interest

E.A.S. is an inventor on patents and patent applications owned by RIKEN covering the CUBIC tissue clearing reagents, is employed by CUBICStars Inc. that offers services based on CUBIC technology and has received collaboration funding from Kantum Ushikata Co., Ltd.