Tissue clearing and 3D imaging – putting immune cells into context

During immune homeostasis and inflammatory challenges, the communication between cells is crucial for a concerted immune response. The precise positioning of individual cell types in specific anatomical locations (the immune niche) influences this communication (Grant et al., 2020; Qi et al., 2014). Spatial information of cells, as well as their distribution within their respective tissue niches, is thus necessary to understand and manipulate cell–cell and cell–niche communication. Nevertheless, our current understanding of these processes still mainly relies on single-cell analyses, including flow cytometry (Dittrich and Göhde, 1968; Fulwyler, 1965) and more recent and advanced methodological developments such as mass cytometry (Bandura et al., 2009) and single-cell RNA-seq (Tang et al., 2009). These methods offer a diverse collection of readouts regarding cell identity and state but require the mechanical or enzymatic extraction of single cells from their specific microenvironment. In this process, information about cellular interactions, cell behaviour and localization is lost. Furthermore, this process is prone to introducing a selection bias, as only cells withstanding the isolation procedure can be observed. Imaging methods can circumvent such caveats by permitting a direct and undisturbed insight into tissues. Conventional histology (Mayer, 1819) and immunohistochemistry (IHC) with immunofluorescence (IF) microscopy (Coons et al., 1941; Coons and Kaplan, 1950; Heimstädt, 1911; Köhler, 1904) have driven countless discoveries over the past century and remain routine methods in biomedical laboratories. However, these conventional imaging techniques usually analyse only a small part of an organ, often in 2D, are limited by the selected sections and suffer from involuntary field-of-view selection bias. As such, most of the contextual information is lost and the likelihood of detecting rare events and cell populations is diminished.

Consequently, the accurate study of cellular organization in tissues requires a method that allows the visualization of larger tissue volumes and reveals the 3D associations of cells and their surroundings. Imaging of thin serial tissue sections (2D) followed by 3D reconstruction has provided volumetric images of the lung (Cabeza-Cabrerizo et al., 2019; Gomariz et al., 2018), but due to the sample preparation method is a very time-consuming approach. In addition, sectioning itself, as well as optical reconstruction, might introduce artefacts. Another approach is optical tissue clearing, which relies on manipulating the optical properties of the samples to achieve a ‘transparent’ tissue sample or organ, and hence an enhanced imaging depth and resolution of the intact, complete tissue (Figs 1 and 2; Box 1). A multitude of optical clearing methods followed by 3D imaging have been developed. In this Review we will provide the background to the optical challenges and their solutions and discuss published protocols for tissue clearing, methods for 3D imaging of immune cells and their limitations, and image analysis. We will touch on possible applications and future developments with the aim of raising awareness of the power and advantages of tissue clearing and 3D imaging for the analysis of immune cells within their respective niches.

The origins of tissue clearing date back to the turn of the twentieth century and can be found in publications by the German anatomist Oskar Schultze (Schultze and Koelliker, 1897) and the physician Werner Spalteholz (Spalteholz, 1911). They applied the idea of refractive index (RI) matching (see Box 1) by extended immersion of tissues in various organic solutions that were available at the time to produce large-scale, transparent anatomical specimens for educational purposes. After being mostly forgotten, tissue clearing was ‘rediscovered’ in light of new imaging methods, such as light-sheet microscopy, that are capable of imaging large samples (Dodt et al., 2007). Over the past decades, this renaissance of sorts has steadily continued, largely driven by neurobiological research.

Tissue clearing approaches can be roughly classified into three conceptual categories: hydrogel-based, hydrophobic and hydrophilic methods. All these methods aim to achieve almost completely transparent organs by removing lipids (delipidation) and pigments (decolourization) and by matching the RI of the samples with that of the surrounding medium (Fig. 1).

Hydrogel-based tissue-clearing techniques aim to completely remove lipids from tissues and, at the same time, secure the physical location of biomolecules and minimize structural damage. This is achieved by crosslinking biomolecules in acrylamide-based hydrogels and subsequent lipid extraction through either electrophoresis or passive diffusion. Among the three groups of tissue-clearing techniques, hydrogel-based methods are the most time-consuming and require specific equipment. They are, however, often used to induce tissue swelling or expansion, which enhances tissue structures and allows for multiplexed labelling and RNA imaging due to high preservation of cellular structures, as well as RNA and DNA retention (Sylwestrak et al., 2016). Cleared lipid-extracted rigid immunostaining/in situ hybridization-compatible tissue hydrogel (CLARITY) (Chung et al., 2013), passive CLARITY technique (PACT) and perfusion-assisted agent-release in situ (PARS) (Yang et al., 2014) compose a family of hydrogel-based methods within which further subtypes have been developed for specific applications or tissues, such as Bone CLARITY (Greenbaum et al., 2017), as reviewed in detail elsewhere (Gradinaru et al., 2018).

Hydrophobic methods of tissue clearing commonly rely on the use of an organic solvent for RI matching, similar to the original approaches. Most hydrophobic tissue-clearing approaches consist of an initial dehydration step followed by immersion in organic solvents, which aid delipidation and RI matching. Hydrophobic methods are fast and robust and are well suited for the preservation of the tissue over extended periods of time.

The initial simple methods that revisited optical clearing mainly focussed on replacing the water content of the tissue. This was achieved through methanol-based tissue dehydration followed by incubation in a clearing medium containing one part benzyl alcohol and two parts benzyl benzoate (BABB) (Dent et al., 1989; Dodt et al., 2007). In subsequent protocols, BABB was exchanged with tetrahydrofuran dehydration and dibenzyl ether for clearing (THF–DBE; Becker et al., 2012). Further collaborative efforts have resulted in the creation of 3D imaging of solvent-cleared organs (3DISCO), which in recent years has enjoyed wider adoption and inspired a long and growing list of variants and evolutions (Belle et al., 2014; Cai et al., 2019; Jing et al., 2018; Kirst et al., 2020; Pan et al., 2016, 2019; Renier et al., 2014; Zhao et al., 2020). While the family of DISCO-related tissue-clearing methods provides results in a short amount of time, the organic solvents used are harmful, volatile and even toxic, and hence independent efforts have introduced the natural aromatic compound ethyl cinnamate (ECi) (Klingberg et al., 2017) as a potential optical clearing solution. Further optimizations of dehydration and pigment decolourization have led to the development of a second-generation ECi-based method (2ECi) (Masselink et al., 2019) and bleaching-augmented solvent-based non-toxic clearing (BALANCE) (Merz et al., 2019), respectively. Our recent addition to ECi-based clearing methods, named efficient tissue clearing and multi-organ volumetric imaging (EMOVI; Hofmann et al., 2021), introduces the digestion of the extracellular matrix (ECM) by hyaluronidase in order to minimize light scattering and autofluorescence caused by differences in the macromolecular content (lipids, ECM) of the tissue (Kim et al., 2018; Susaki et al., 2020). Other recent methods involve the adoption of methyl salicylate as a clearing agent chemically similar to ECi (Messal et al., 2021) or ECi-containing composite-clearing solutions (Biswas et al., 2021 preprint). Such cleared tissues can be re-hydrated and re-used for histology or cryo-immunofluorescence (Biswas et al., 2021 preprint; Hofmann et al., 2021; Messal et al., 2021).

Parallel to these techniques, a group of hydrophilic clearing methods, which make use of aqueous solutions to homogenize the RI, are used. Hydrophilic methods use water-soluble agents that are less destructive to the tissue. Compared to hydrophobic clearing approaches, they often require prolonged incubation times and might lead to swelling of the tissue. The subgroups of hydrophilic clearing methods employ specific agents for decolourization, delipidation or RI matching, such as urea or urea with sorbitol in Scale and ScaleS, respectively (Hama et al., 2015, 2011), fructose in See Deep Brain (SeeDB; Ke et al., 2013), and urea and fructose in FRUIT (Hou et al., 2015). The most widely applicable agents among those used in hydrophilic methods, due to their flexibility, are the amino alcohol-based chemical cocktails developed by Ueda and colleagues, which define the family of methods that includes clear, unobstructed brain imaging cocktails and computational analysis (CUBIC; Susaki et al., 2014) and its numerous optimized and tissue-adapted variants (Hasegawa et al., 2019; Matsumoto et al., 2019; Murakami et al., 2018; Susaki et al., 2020, 2015; Tainaka et al., 2014, 2018). Other hydrophilic clearing approaches use several compounds to achieve optimal clearing results, such as Histodenz-based clearing-enhanced 3D (Ce3D) (Li et al., 2017, 2019) or the FlyClear (Pende et al., 2018) and DEpigmEntation-Plus-Clearing (DEEP-Clear) methods (Pende et al., 2020).

As diverse as the clearing methods themselves are their potential applications. The initial focus of many studies since the beginning of the clearing ‘renaissance’ was the study of the central nervous system architecture (Dodt et al., 2007). Attempts to understand the connectome of neurons and the vasculature of the brain drove the development of most of the early technical advances, which remain a centrepiece of many CUBIC and DISCO approaches (reviewed in Molbay et al., 2021; Ueda et al., 2020a,b). Other studies have employed clearing to elucidate the large-scale structural composition of the brain (Renier et al., 2016), the vertebral column lymphatic network in mice (Jacob et al., 2019) or even early human development (Belle et al., 2017; Renier et al., 2014), while further studies have used clearing to characterize the distribution and changes in vasculature throughout bone, heart or adipose tissue (Gilleron et al., 2020; Gruneboom et al., 2019; Merz et al., 2019).

In recent years, tissue clearing has also gained growing attention from the field of nephrology, where it is used to dissect the complex kidney micro-architecture (Puelles et al., 2019, 2016; Su et al., 2019; Unnersjö-Jess et al., 2021, 2016) and macro–architecture (Hasegawa et al., 2019; Klingberg et al., 2017; Renier et al., 2014; Zhao et al., 2020). Investigations in pathophysiological settings is of course not limited to nephrology, and a number of studies have made use of clearing techniques to describe tumour biology (Cuccarese et al., 2017; Kubota et al., 2017; Lee et al., 2017; Messal et al., 2021; Pan et al., 2019; Rios et al., 2019; Stoltzfus et al., 2020) and the effect and distribution of infectious pathogens (Amich et al., 2020; Fretaud et al., 2021; Song et al., 2021), amongst others.

In an attempt to group the more than 70 published clearing protocols, we defined 13 parameters characterizing protocol specifics, the resolution achieved and the method of fluorescent labelling used (Fig. 2). We next employed clustering algorithms [uniform manifold approximation and projection (UMAP) combined with Phenograph in FlowJo] to group published protocols, obtaining five groups. We found that the group 1 protocols were uniquely characterized by the standalone parameter ‘cellular resolution’, which we define as a method that provides quantifiable cellular resolution with intact cellular morphology (Fig. 2). Here, we therefore focus on protocols that can be used to clear tissues with sufficient resolution to detect and quantify cellular structures.

Interestingly, in this group of cellular-resolution clearing methods, all three approaches (i.e. hydrogel-based, hydrophilic and hydrophobic) are present (Fig. 2B,C). Hydrogel-based methods [the family of CLARITY-based methods, MAP (Ku et al., 2016) and SHIELD (Park et al., 2019)] as well as some hydrophilic [SeeDB (Ke et al., 2013), Clear^T2 (Kuwajima et al., 2013) and Simplified CLARITY (Lai et al., 2016)] or hydrophobic [uDISCO (Pan et al., 2016), iDISCO (Renier et al., 2014)] clearing approaches are well suited to preserve the neural architecture in the brain and are compatible with endogenous fluorescent reporters or immunolabelling of nerves and cell nuclei (for a more detailed description see Ueda et al., 2020a,b). Similarly, other hydrophilic or hydrophobic clearing agents are compatible with endogenous fluorescent reporters and immunolabelling and have been used to image microglia (FACT; Xu et al., 2017), detailed kidney structures (Unnersjö-Jess et al., 2021) or mammary tissue (FUnGI; Rios et al., 2019).

Compared with the analysis of structural components in the brain, the in-depth analysis of immune cells and their respective niches remains understudied. This might be because immune cells are rather small, compared to vessels or nerves, and labelling immune cells in their respective niches has proven to be challenging. Furthermore, imaging with a resolution that allows single cells to be distinguished and quantified requires higher magnification, is more time-consuming in large volumes and necessitates computational processing power. Recent approaches to study immune cells in volumetric tissues have employed both hydrophilic [Ce3D (Li et al., 2017), FUnGI (Rios et al., 2019)] and hydrophobic [EMOVI (Hofmann et al., 2021), FLASH (Messal et al., 2021)] clearing approaches (Fig. 2C).

Below, we review recent labelling and imaging approaches and discuss the challenges of data analysis when imaging immune cells in context.

In order to visualize immune cells in their microenvironments, the cells of interest need to be tagged by a fluorochrome. The excitation and emission spectra of fluorochromes to be used is determined by the imaging method of choice. To obtain a robust and homogenous labelling of immune cells compatible with tissue clearing, many studies employ mice expressing transgenic reporters for a specific subset of immune cells (Fig. 3A) (Ballesteros et al., 2020; Erturk et al., 2012; Klingberg et al., 2017; Merz et al., 2019; Puelles et al., 2019; Rios et al., 2019). Although well-established reporter systems provide ideal and specific labelling, they require either the possession, acquisition or even the lengthy de novo generation of the desired mouse lines. Furthermore, most conventional transgenic models only allow labelling of a single marker or cell population. Successful multicolour 3D imaging has also been demonstrated using mice expressing a multitude of transgenic reporters, such as the ‘confetti’ mice (Livet et al., 2007). Of note, native fluorescence of reporter proteins may be lost during certain clearing procedures. A summary of compatibility of clearing methods with endogenous fluorescent reporters can be found elsewhere (Richardson and Lichtman, 2015; Seo et al., 2016).

Another possibility for the visualization of immune cells is labelling them with antibodies conjugated to a fluorochrome (also referred to as immunolabelling; Fig. 3A). Similar to their use in flow cytometry, fluorescently labelled antibodies allow for multiplexing in volumetric imaging and hence offer the greatest flexibility in cell detection. For immunolabelling, fluorochromes that show certain biophysical properties, such as photostability, small size and (chemical) polarity are favourable. Most of the published protocols therefore rely on Alexa Fluor dyes for this purpose (Hofmann et al., 2021; Li et al., 2019; Renier et al., 2014). However, the increased availability of small and easily penetrating nanobodies holds much promise and has recently been used to label epitopes in an entire animal (Cai et al., 2019).

Immunolabelling requires the delivery of antibodies to achieve consistent staining throughout the entire tissue (Fig. 3B). A common technique used in many protocols is the intravenous (i.v.) injection of antibodies into the blood circulation through the tail vain shortly before or after euthanizing mice (Klingberg et al., 2017; Merz et al., 2019). This not only yields robust staining but is also the fastest staining method available. Nevertheless, such injection of antibodies limits staining to surface antigens of one set of cells in the entire animal, thereby restricting the use of different tissues for multiple lines of investigation. In addition, this procedure necessitates certain legal requirements as well as technical expertise. Injection of antibodies is also not suited for work with human tissue samples or other biopsy material.

A more flexible and accessible way to label immune cells using fluorescently labelled antibodies is immersion staining post tissue fixation. This can circumvent the issues of i.v. application and allows for the use of small amounts of highly multiplexed antibody cocktails tailored to the specific need of the tissue sample. Here, antibody penetration is determined by the tissue composition, with ECM components, fat tissue and muscles influencing penetration. Tissue density as determined by the number of cell layers, cellular distribution and size might also influence antibody penetration (Fig. 3C). Consequently, immune cells in tissues such as lymph nodes (Duckworth et al., 2021; Li et al., 2019; Stoltzfus et al., 2020), lung (Amich et al., 2020; Ballesteros et al., 2020), or white adipose tissue (Hofmann et al., 2021) are more readily stained compared to those in kidney, liver, heart or gut tissue. Permeabilization of the tissue is thus critical for increasing the penetration depth of the antibodies into the tissue. Protocols used for this purpose include a variety of detergents and solvents, such as saponin, Triton X-100, DMSO, urea or CHAPS (Hama et al., 2015; Li et al., 2019; Messal et al., 2021; Renier et al., 2014; Rios et al., 2019; Zhao et al., 2020). To improve antibody diffusion throughout such dense tissues, and in addition to the permeabilization step, we have introduced hyaluronidase treatment of the tissue to remove hyaluronic acid – a component of the ECM (Hofmann et al., 2021). Hyaluronic acid is a polysaccharide that provides structure to tissue architecture and increases viscosity and hydration in interstitial collagenous matrices. Treatment of tissue with hyaluronidase hence removes the ‘filling’ but keeps structural components, such as collagen, fibronectin and laminin, intact. Other methods have described the digestion of ECM components using acetic acid and guanidine-HCl (Zhao et al., 2020) or collagenase (Biswas et al., 2021 preprint; Susaki et al., 2020).

Further approaches to enhance immunolabelling rely on active transport of antibody solutions by electrophoresis (Lee et al., 2016), pressurized perfusion systems for antibody delivery (Yang et al., 2014) or increased antibody supply (Chung et al., 2013), but these have not been tested for the labelling of immune cells. Whereas active transport of antibody solutions reduces staining time, it still requires a more sophisticated setup than immersion staining.

Using immersion staining post tissue fixation provides the greatest flexibility for analysing immune cells in their microenvironment. Care needs to be taken when choosing detergents for permeabilization and ECM digestion because not all of these solutions preserve antigens for antibody labelling. Similarly, not all clearing agents preserve fluorescence equally well, with some even completely quenching the fluorescence of antibodies used for immunolabelling (Li et al., 2017). Recent studies have taken these limitations into account, demonstrating that antibody multiplexing, long-term preservation of fluorochromes, and even further processing of the tissue for sectioning and traditional IF is feasible (Biswas et al., 2021 preprint; Hofmann et al., 2021; Messal et al., 2021).

With the continuing developments in large-scale organ clearing, it is crucial to also consider the best mode of acquiring images of these large samples. Many different types of microscopes exist, from the conventional widefield fluorescence (Heimstädt, 1911; Köhler, 1904) and two-photon microscopes (Abella, 1962; Denk et al., 1990; Göppert-Mayer, 1931) to modern optical sectioning microscopes, such as confocal laser scanning microscopes (Brakenhoff et al., 1979; Minsky, 1957, 1988) or light-sheet fluorescence microscopes (LSFM) (Dodt et al., 2007; Huisken et al., 2004; Siedentopf and Zsigmondy, 1902), which have inspired the tissue-clearing renaissance. Each of these methods presents a compromise between image resolution and acquisition speed, and have their own advantages and disadvantages. Many reviews provide help in finding the best suited method for each application (Feuchtinger et al., 2016; Husna et al., 2019).

Detecting immune cells during volumetric imaging further requires a certain resolution due to the small size of the cells. In addition to choosing the correct imaging modality for this purpose, the fluorescence signal of the cells should have a good signal-to-noise ratio. Approaches to quench tissue autofluorescence and thereby improve the signal-to-noise ratio in cleared tissue range from haem decolourization [Scale (Hama et al., 2011), CUBIC (Susaki et al., 2014) and SHANEL (Zhao et al., 2020)]) to decalcification [PEGASOS (Jing et al., 2018), PACT (Yang et al., 2014) and CUBIC (Tainaka et al., 2018)] or depigmentation (tissue bleaching) using peroxide (Hofmann et al., 2021; Merz et al., 2019; Renier et al., 2014). In our hands, the concentration of peroxide needs to be adapted to the fluorochromes and the staining methods used, but tissue bleaching is helpful to gain a better intensity of the emission signal, resulting in enhanced imaging depth (Hofmann et al., 2021). In the following sections, we will give a short overview of imaging modalities used for the imaging of immune cells in 3D (Fig. 4).

LSFM has been widely used for structural analysis of large organs, such as the kidney, heart and brain, and even of whole animals (Kirst et al., 2020; Klingberg et al., 2017; Merz et al., 2019; Pan et al., 2019; Susaki et al., 2020; Ueda et al., 2020a). Due to its high resolution in z, LSFM is the method of choice to image and quantify the distribution of vessels, nerves or endothelial cells (Biswas et al., 2021 preprint; Messal et al., 2021). The detection of single cells is more challenging using LSFM but has been achieved in the kidney and heart, mainly using transgenic mice expressing fluorescent markers for certain immune cell types (Klingberg et al., 2017; Merz et al., 2019). Furthermore, subsets of monocytes have been shown in Aspergillus-infected lungs using immunolabelling followed by LSFM (Amich et al., 2020). Recent work has demonstrated the use of LSFM to study the distribution of T cells in lymph nodes during infection using a combination of endogenous reporters and immunolabelling (Duckworth et al., 2021). The preferred fluorochromes for LSFM emit in the red-to-infrared spectrum, as light in this range travels best through tissue. This, however, limits multiplexing options and hinders the analysis of the samples on different microscopes, as light emitted in this spectrum is more difficult to detect using most confocal microscopes.

In order to image immune cells in large volumes, confocal microscopy has been the method of choice to date. In addition to providing the maximum cellular resolution, and thus enabling the most in-depth analysis, confocal microscopes are also the most widely accessible microscope in immunological laboratories. Another advantage of most confocal microscopes is their compatibility with dyes commonly used in flow cytometric analysis, thereby contributing to the flexibility of this imaging approach. Disadvantages are the small field of view and the limited working distance of objectives usually installed on confocal microscopes, which can be circumvented by using dedicated objectives for cleared tissues, such as two-photon or motorized objectives (Cai et al., 2019; Li et al., 2017, 2019; Merz et al., 2019; Pan et al., 2019; Puelles et al., 2019; Saritas et al., 2018; Yang et al., 2014). Imaging immune cells in large-volume samples using confocal microscopy is also a time-consuming process (several hours to days) and hence is mainly feasible for smaller organs, such as lymph nodes or tissue pieces (Hofmann et al., 2021; Li et al., 2017, 2019; Stoltzfus et al., 2020). Using a variety of clearing methods, recent publications highlight the power of confocal microscopy to image and quantify immune cells in their native environment, such as innate and adaptive immune cells in lymph nodes (Rios et al., 2019; Stoltzfus et al., 2020), neutrophils in the lung (Ballesteros et al., 2020), infiltrating lymphocytes in mouse mammary tumours (Lee et al., 2017) or macrophages in mammary gland tissue (Dawson et al., 2020).

Thus, confocal microscopy remains the imaging method of choice if a high resolution is needed for subsequent cellular quantification, but it is also the most time-consuming imaging technique. A faster alternative for detecting immune cells in at least some tissues is high-end widefield microscopy.

Typically, widefield microscopes are used in IHC and are thus widely available in clinical settings. However, these microscopes are incompatible with fluorescence imaging of large-volume samples as they detect of out-of-focus light, which results in ‘blurred’ or ‘hazy’ images. We recently described widefield microscopy with real-time computational clearing as a valuable alternative to confocal microscopy for rapid image acquisition (Hofmann et al., 2021). We adapted novel software algorithms designed to remove the blurring of the widefield images and demonstrated that imaging of large volumes, such as whole lymph nodes, half a kidney or large pieces of fat tissue, as well as the detection and quantification of rare cell populations or loosely spread cells, is possible (Hofmann et al., 2021). Limitations of widefield microscopy include a restricted resolution in z due to the nature of the method (mode of illumination and image acquisition). In addition, widefield microscopy is susceptible to tissue autofluorescence, and we thus have found acquisition of single cells in cleared lung, liver and heart challenging. Another consideration is the size of the data generated. Widefield microscopy can generate large datasets in a short time. In addition, images obtained by widefield microscopy require deconvolution to regain spatial resolution (Schumacher and Bertrand, 2019). Thus, when imaging with a widefield microscope, data storage and processing power of the computer used for analysis need to be considered.

In summary, several imaging technologies can be used to detect single cells in cleared, large-volume samples, with LSFM, confocal microscopy and advanced widefield techniques providing good resolution (Fig. 4). A comparison of imaging speed, resolution and data size generated by these microscopes, as well as advantages and limitations to detect single cells has been provided previously (Hofmann et al., 2021). We believe that for best cellular resolution and in-depth analysis, confocal microscopy is still the method of choice.

The switch from simpler 2D imaging to full volumetric imaging of large tissue sections, whole organs or even entire specimens does not only push the limits of current hardware but also sets new challenges for processing and analysis software. The increase in dimensionality inflates the generated data size of each acquisition substantially, and even the simple display of large 3D microscopy files requires good computational resources. Hence, 3D imaging not only demands a solution for the transfer and storage of big data, but also specialized software tools. An ever-growing variety of commercial tools, such as IMARIS (imaris.oxinst.com) and Amira (www.thermofisher.com), or freely available applications, for example Fiji (Rueden et al., 2017; Schindelin et al., 2012) and Vaa3D (Peng et al., 2010), offer a selection of analytical capabilities as described elsewhere (Piccinini et al., 2020) (Fig. 5A). When choosing a specific software option, users should take into account the intended biological question as well as the amount of available documentation and required technical expertise.

Multiple specialized analytical pipelines have been published in connection with clearing protocols, and these are focussed on the fully automated detection of structures or cells in large-volume datasets. In particular, the field of neurology has seen a variety of extensive whole-brain mapping strategies to create complete and detailed catalogues of vasculature (Di Giovanna et al., 2018; Kirst et al., 2020; Todorov et al., 2020), as well as to profile cell distribution and organization (Friedmann et al., 2020; Matsumoto et al., 2019; Murakami et al., 2018; Susaki et al., 2020; Zhao et al., 2020). There also have been efforts to detect and quantify heart vasculature (Merz et al., 2019) and micro-metastases (Pan et al., 2019). Of note is the rising use of neural network-based selection automatization, which can outperform conventional algorithms and might offer even greater analytical options in the future (Wainberg et al., 2018). Although these analysis tools can perform highly sophisticated analysis with respect to their designed purpose, they cannot be easily applied to other settings where different sets of data are obtained from different tissue types. Additionally, they utilize highly advanced algorithms that not only require substantial expertise in setup and operation, but also significant processing power.

Other tools for image analysis offer a more intuitive, streamlined approach to quantify cells in their tissue niches (Berg et al., 2019; McQuin et al., 2018; Stoltzfus et al., 2020). Amongst these is a technique called histo-cytometry, which was first introduced for the analysis of immune cells in 2D images (Gerner et al., 2012), and recently has been implemented for the analysis of immune cells in Ce3D-cleared tissues (Li et al., 2017, 2019; Stoltzfus et al., 2020). Histo-cytometry maps the position and phenotypic identity of cells stained with multiplexed antibodies; hence, this method operates in a manner analogous to flow cytometric analysis but additionally provides information about the spatial organization of immune cells in tissues. Using confocal microscopy followed by histo-cytometry, the distribution of several immune cell subsets in murine lymph nodes has been extensively studied (Li et al., 2017, 2019; Stoltzfus et al., 2020) (Fig. 5B). The most challenging step of this analysis pipeline, as is true for all image-analysis approaches, is the segmentation of immune cells sitting in close proximity in their tissue niches. Future developments will hopefully provide better algorithms for this purpose.

When observing single cells separated from each other, the challenging segmentation can be omitted and a simplified analytical pipeline can be used to perform complex investigations of tissue niches (Fig. 5B). We have used our EMOVI staining and clearing protocol followed by confocal microscopy to quantify changes in sparse cell populations in the lung during systemic inflammation (Hofmann et al., 2021). Furthermore, a combination of surface creation in IMARIS and population gating in FlowJo using histo-cytometry allowed us to accurately define and visualize antigen-presenting cells surrounding or infiltrating glomeruli of nephritic kidneys (Hofmann et al., 2021). Combining 3D imaging and histo-cytometry can thus be a powerful tool to investigate the changes in distribution of immune cells in their respective niches during homeostasis or pathological processes.

It is becoming more and more clear that immune cells not only adapt to but also shape their microenvironment. In addition, the exact localization of cells in the tissue as well as local crosstalk with other immune cells or endothelial cells contributes to immune cell function. This information on cell localization and cell–cell interaction is lost during the harsh treatment of tissues to obtain single-cell suspensions. Furthermore, tissue digestion not only destroys epitopes on the cell surface of cells but often also leads to loss of fragile cell populations. This might be especially true for inflamed or pathologically altered tissues in autoimmune diseases or other pathophysiologies. Tissue clearing and large-volume imaging provides a way to investigate cell–cell interactions and cell–niche crosstalk in their native environment.

Recent advances in this method enable the staining of subsets of immune cells in whole organs or large tissue pieces, as well as subsequent volumetric imaging and detailed quantification of immune cells during homeostatic or pathophysiological processes. Some challenges such as obtaining homogenous labelling and imaging large tissue pieces remain, but promising new methods with the potential to ease large-scale volumetric imaging are in development. These include, amongst others, widefield microscopy with computational clearing (Schumacher and Bertrand, 2019) and multi-photon LSFM (Maioli et al., 2020), as well as optimizations of the established methods, especially with regard to better objectives and longer wavelength excitation and emission capabilities (Cheng et al., 2019; Schweikhard et al., 2020). With all these possibilities, advancing hardware and data storage capacities to handle the data will be crucial.

Future developments might also see the combination of tissue clearing and volumetric imaging of large tissue pieces with mass cytometry or single-cell RNA sequencing with cells isolated from those same tissues. As imaging mass cytometry and transcriptome analysis of thin layers of stained tissue are feasible but still limited to 2D, it will be interesting to see the development of these techniques to obtain spatial information of proteins and RNA in 3D.

We thank all the members of the Inflammation and Immunity laboratory, MRI, TUM, for scientific discussions and critical reading of the manuscript.