The first step in CLARITY is tissue fixation. In the original CLARITY protocol, animals were transcardially perfused with HM solution (including 4% PFA, 4% acrylamide, 0.05% bis-acrylamide and 0.25% VA-044 initiator in PBS) after removing blood. Next, the tissue is incubated in HM solution at 4°C for 2–3 days followed by hydrogel polymerization with heating (14). In a separate protocol, the incubation time was shortened to 1 day (17).

PFA, acrylamide, proteins and nucleic acids chemically bond together to form a hydrogel mesh. This binding preserves intrinsic molecules (proteins and nucleic acids) and permits the penetration of labeling dyes. However, the speed of clearing and the immunolabeling process depend on the pore size of hydrogel mesh. High concentrations of PFA and acrylamide result in a dense hydrogel mesh. This mesh creates a stable hydrogel-tissue hybridization and reduced the loss of proteins. However, a dense hydrogel mesh slows both tissue clearing and antibody perfusion (17,21,28,29). Therefore, many researchers have adjusted the mixture ratio for different tissues and clearing approaches to maintain a balance between hydrogel rigidity and porosity with minimal protein loss and a faster speed for clearing and immunolabeling (19).

In ETC, the original solution (4% PFA, 4% acrylamide, 0.05% bis-acrylamide) is the most common ratio to date. Epp et al systematically researched the influence of different ETC temperatures and hydrogel composition on tissue transparency and expansion using an intact mouse brain (21). These researchers recommended a 4% formaldehyde and 4% acrylamide mixture for examining endogenous fluorescent markers and a composition of 3% acrylamide and 3% formaldehyde for immunohistochemistry. For passive CLARITY, the process of tissue fixation is the same as ETC in the original article (17). Next, Yang et al proposed PACT, which adopted a different perfusion method (28). Compared with the passive CLARITY protocol, PACT fixes tissues with 4% PFA followed by incubation at 4°C overnight in the hydrogel monomer solution (A4P0, 4% acrylamide in PBS supplemented with VA-044 initiator). These researchers also demonstrated the structural integrity of tissues throughout PACT processing using 1-mm-thick Thy1-eYFP tissue sections. The pore of the hydrogel mesh in tissues processing with PACT is larger than that observed with HM solution because of the obstruction of lipid bilayer when incubating in A4P0 (23). Thus, tissues fixed with A4P0 would be more easily deformed compared with those in original protocol. However, this provides advantages in certain dense tissues, such as bone (PACT-deCAL) (23,25). This perfusion method for PACT is also used in an improved ETC technique termed ACT-PRESTO (18). Lee et al optimized the original PACT procedure, creating psPACT (process-separated PACT) and mPACT (modified PACT: psPACT with added α-thioglycerol) (30). In psPACT, fixed tissues were incubated in A4P0 (4% acrylamide in PBS without VA-044) at 37°C for 24 h (the original protocol incubated tissues with A4P0 within VA-044 initiator and stored tissues at 4°C). Next, tissues were covered with 0.25% VA-044 at room temperature for 6 to 24 h. The authors compared psPACT, mPACT with PACT in the rat central nervous system (spinal cord and brain), as well as in other internal organs. These researchers concluded that this tissue fixation method could accelerate tissue clearing, with mPACT being faster than psPACT. We compare these different fixation methods in Fig. 3.

We have observed that there is no obvious limitation to the perfusion method used in different tissue clearing approaches (ETC vs. passive CLARITY). The most important challenge is to shorten the time of tissue clearing by adjusting the pore diameter of hydrogel mesh using different concentrations of ingredients and perfusion methods while keeping tissue structure stable. Passive CLARITY and PACT requires more time to extract lipids from tissue-hydrogel hybrids with shaking with a higher temperature than would be used in ETC when handling relatively large tissues (ETC adopted 37°C; the original passive CLARITY protocol recommended 37°C-60°C; PACT first adopted 37°C; another study concluded 42°C-47°C is a better temperature range for PACT) (17,28,31). Thus, the requirement for tissue stability in PACT is higher than that of ETC, and the concentration of acrylamide and PFA is of vital importance when implementing passive CLARITY and PACT.

The following step after tissue fixation is hydrogel polymerization. In the original protocol, nitrogen and vacuum pump are used to extract and replace air in the tissue container (as oxygen impedes hydrogel formation) (14). Notably, oxygen must be completely removed; otherwise, hydrogel formation in the inner parts of tissue would be incomplete and result in the emergence of cavities after lipids extraction. It was also reported that adding a thick layer of mineral oil over the top of the hydrogel solution before polymerization could minimize the exposure of samples to the air (32).

Since membrane lipids are the main cause of light diffraction, lipid extraction becomes a key process in CLARITY (1,20,33). Tissue transparency primarily depends on the degree of lipid elution. In general, there are two methods for lipid extraction: passive thermal diffusion and electrophoresis. Passive thermal diffusion is a way to transport SDS micelles into tissues via heat-induced diffusion. And electrophoresis applies electric fields to accelerate the penetrating of SDS micelles. All modified CLARITY techniques elute membrane lipids based on these two methods (PACT, PARS, ETC, ACT-PRESTO- and stochastic electrotransport) (14,17,18,14). In this study, we compared strengths and weaknesses of these two methods in Table I. Next, we reviewed the modified CLARITY techniques based on the two methods.

Regarding passive thermal diffusion in-passive CLARITY, tissues are incubated in 4% SDS at pH 8.5 with gentle shaking at 37°C with an increased temperature of up to 60°C for faster clearing (17). Because of dense hydrogel mesh structure and low efficiency of SDS transportation, this method is time-consuming and may not reach a high degree of tissue transparency for larger tissues, except at higher temperatures (34). However, higher temperatures may cause potential tissue deformation and damage to natural molecules. Therefore, in order to accelerate SDS transportation and to keep tissue structure stable, Yang et al proposed new methods, termed PACT and PARS (23,28). PACT adopted 8% SDS buffer at pH 8.5 and samples are gently shaking at 37°C for tissue clearing. These researchers emphasized that only an 8% SDS concentration could achieve uniform tissue clearing throughout the entire 3-mm block, while the SDS concentration in ETC and most other modified techniques is 4% (14,17). PARS is a method for whole-body clearing and labeling (23,28). After fixing tissues with PFA and A4P0, tissues of interest are not removed from animal's body and the clearing process is achieved in vivo. PARS utilized the intact vasculature to deliver and circulate the SDS buffer. For different tissues, different parts of vasculature could be selected. For brain or spinal cord clearing, PACT regents could be delivered with a subdural cannula inserted above the region of interest, and the buffer circulation would be similar to that of cerebrospinal fluid (28). The principle of PARS is also based on passive diffusion. The difference is that PARS increased the contact area between the clearing buffer and tissues using the intrinsic vasculature; therefore, the tissue clearing speed becomes faster. Additionally, PARS minimizes tissue expansion during clearing, as the skull and other physiologic structures limit the available space. However, implementation of PARS is more complicated than PACT (a cannula needs to be set and kept stable as tissue vasculature is easily destroyed and obstructed during buffer circulation). Thus, the application of PARS is limited. After that, several modified methods based on PACT emerged. For mPACT, the author added 0.5% α-thioglycerol to 8% SDS buffer for lipid extraction with gentle shaking, and it showed faster clearing speed than PACT. The use of α-thioglycerol was also reported in other articles to avoid browning and autofluorescence accumulation as a result of the Maillard reaction (10,18). The α-thioglycerol works via the sulfhydryl, suggesting that other chemical materials with sulfhydryl, such as β-mercaptoethanol, may also work (10). It has also been reported that 0–50 mM sodium sulfite could exhibit an anti-browning effect in other tissue clearing methods, such as SWITCH (35). Lee et al also combined mPACT and PARS as a whole body perfusion method (PARS-mPACT) (30). PARS-mPACT fixed tissues via the transcardial perfusion of separated mPACT regents. Next, polymerized tissues were isolated and passively cleared according to the mPACT protocol. However, PARS-mPACT and mPACT showed no differences in tissue transparency. Another method, named CLARITY2, adopted a different way to accelerate tissue clearing (36). After tissue fixation and hydrogel polymerization, brain tissue was cut to 1–1.5-mm-thick coronal slices and underwent passive tissue clearing. This approach enabled us to accelerate and simplify the clearing, staining and imaging steps compared with original protocol. However, CLARITY2 did not avoid damaging the tissues and is similar to a traditional tissue slice. However, this technique is still an ideal time-saving method for methodological studies (32,37,38). In addition, it's reported that with pancreatic lipase breaking down lipid droplets, lipid-rich tissues were better cleared when performing passive clearing (39).

And there were no obvious tradeoffs in the quality or characteristics of cleared tissues with the SDS buffer at different pH, except for a slight enhancement in the rate of clearance at a more alkaline pH (23). Thus PACT, PARS and other passive CLARITY techniques could dissolve SDS using 1× PBS at pH 7.5 for convenience, adjusting the SDS concentration according to the needs of the study to maintain a balance between tissue integrity and transparency.

Electrophoresis is faster than passive thermal diffusion, as electric fields accelerate the transport of SDS. First, the ETC system is composed of fourelements: i) A chamber for containing samples and electrodes; ii) a power supply; iii) a circulator to control flow rate and buffer temperature, which is combined with the chamber; and iv) a filter to remove byproducts.

In the original protocol, a range of conditions (10–60 V) were applied across the tissue at 37–50°C for several days with clearing solution circulating through ETC chamber (14). However, the high temperature and voltage could cause a higher risk of tissue deformation, as well as epitope and fluorescence loss (17,22). Therefore, the voltage and circulation temperature should be precisely controlled to produce better outcomes. It was also recommended that ETC be completed at 25 V and 37°C for an adult mouse brain (17). The limitations to the application of high electric fields in the ETC system was compounded by the fact that the electrical properties in different regions of a tissue are not homogeneous, leading to regions with concentrated electric fields. Thus, high electric fields may cause deformation of these regions. This suggests that electrophoresis is ineffective for promoting the transportation of SDS into large, dense samples only if high electric fields are used. Stochastic electrotransport, created by Kim et al could promote the migration of freely moving molecules with high electromobility while suppressing the displacement of endogenous biological molecules with low electromobility within the sample. This technique may be an ideal way to solve this high-voltage ETC problem (22). Compared with the stationary samples and electrodes in original ETC chamber, the author implemented stochastic electrotransport by creating a continuously rotating sample chamber with respect to two parallel electrodes (22). Next, these researchers made a contrast among mouse brains cleared by stochastic electrotransport, static electrophoresis and thermal diffusion. The mouse brain cleared with stochastic electrotransport was remarkably transparent within 3 days, while others did not show the same results (22). Lee et al created a modified clearing method termed ACT (18). ACT alternated the platinum wires to platinum plate to generate a dense regular current in the ETC chamber and adopted a two-step fixation protocol as mentioned in PACT. Therefore, ACT could clear tissues faster than ETC (2 h of ACT could achieve complete optical transparency in 1-mm brain sections and 15 h was sufficient to clear entire adult rat brain) (18). However, this method has a high risk of tissue deformation because of the strong electric fields generated and relatively unstable tissue-hydrogel hybridization structure compared to HM solution fixed tissues.

Moreover, modifications to ETC were proposed via changing the parameters of ETC (current or temperature) to achieve better results. One study reported that different optimized current of ETC for various organs (including brain, pancreas, kidney, liver, intestine and lung) (24), demonstrating that the current and voltage of electrophoresis should correspond with the needs of the study. Epp et al tested effects of different ETC temperatures on tissue transparency and proposed a combined 37/55°C clearing protocol, in which they ran ETC at 20 V for the first 4 days at 37°C and increased the temperature to 55°C on the final day (21). Using this method, tissues showed better transparency and structural integrity than those produced by running ETC at only 37°C or 55°C.

There are additional problems when repeating the original protocol. First, the by-products formed during electrophoresis would discolor tissues and colored particles may deposit on the surface of tissues (24). To solve this problem, the author in stochastic electrotransport used a temperature controller to maintain SDS buffer temperature at ~15°C to prevent browning tissues due to heat. Apart from that step, nanoporous membranes were used to contain samples and mechanically divide the circulating solution into an inner portion and outer portion (22). The lower concentrations in outer portion were designed to slow down electro-oxidation of the surfactant molecules (22). Second, the continuously emerging bubbles during electrophoresis could interrupt electrophoresis if the ETC chamber was filled (14,17). In ACT (18), the author designed a long ETC chamber, which allowed bubbles float to the top where they were removed through the top outlet. Bastrup and Larsen (29) also designed an adjustable ETC chamber to fit variable tissue sizes.

In general, there are three methods to visualization interest molecules: i) Genetic introduction of fluorescent markers; ii) in vivo labeling of cells or regions with viruses or chemicals; and iii) chemical/antibody staining (40). For the first two methods, fluorescence may quench throughout clearing process and require amplification (2,19). Therefore, chemical/antibody staining is an alternative way to label cleared tissues. With this method, multiple rounds of staining [three rounds of staining with up to 4 channels were reported (14)] are possible without damaging the preserved structures as proteins and nucleic acids are chemically bound to the hydrogel mesh (19). The clearing solution SDS buffer can be used to wash antibodies and other molecular labels out of the hydrogel-tissue hybridization in preparation for another round of staining.

In the original protocol, tissues were immunostained by incubating them at 37°C with a high concentration of primary and secondary antibodies (dilution, 1:50-1:100) and gentle shaking. Itrequires between 2 days and 2 weeks, depending on the size of the tissue sample (14). The principle of this immunostaining method is the same with passive tissue clearing, expediting the penetration of antibodies by passive thermal diffusion. Some other methods to provide external forces for accelerating the transportation and penetration of antibodies were applied to labeling tissues with a large volume. Similar to lipid extraction, the improvement of immunostaining is also based on two aspects: passive diffusion and electrophoresis (41).

Many labeling methods are based on passive diffusion as it's simple to implement. It has been reported that PARS was suitable for immunostaining and that all immunohistochemical solutions were delivered through the PARS circulation system (including blocking reagent, antibodies or fluorescently labeled molecules and wash buffers) (23,28). This delivery is target-specific and uniformly distributed throughout organs with a low background (28). However, similar to original protocol, it cannot expedite the labeling process (23). To solve this problem, another method termed PRESTO (pressure related efficient and stable transfer of macromolecules into organs) emerged (18). This method was composed of c-PRESTO (centrifugal PRESTO) and s-PRESTO (syringe PRESTO). The c-PRESTO technique applied a centrifugal force with a standard table top centrifuge (600 rcf) to expedite antibody diffusion. For example, kidney tissues incubated for 3 h with antibodies showed labeled structures 10–30 µm deep, whereas for normal tissues, 3 h was sufficient to label structures 120 µm deep (18). The s-PRESTO created a convection flow with a syringe pump, which infused labeling reagents into the specimen. This technique showed a labeling depth that was four times greater than that obtained in original passive labeling (41).

Almost all cleared tissues are applied a fluorescence-based phenotype to label and image interest tissues. But one study used a colorimetric, non-fluorescent method based on the conversion of horseradish peroxidase to diaminobenzidine to label PACT cleared tissues (42). However, this technique was only applied to 50 and 100 µm sections of adult mouse brain tissue; therefore, it did not contain the anatomical structure information of an intact brain. Furthermore, non-fluorescence labeling cannot achieve 3D reconstruction, limiting further applications in tissueimaging.

As antibodies are charged molecules in solution with certain pH, applying external electric fields can expedite the transportation of them into tissues. A study by Li et al applied a simple constant electric field across a 500-µm-thick brain section, and it decreased the delivery time of antibodies by more than 800-fold over simple diffusion without incurring structural damage (43). In this study, the 500-µm-thick brain section was stained in only 30 min at an external voltage of 25 V. However, when applying this to larger tissues (whole brains or animal bodies), heat damage caused by electrodes may occur necessitating buffer circulation to control the temperature. To ensure the concentration of dyes, the staining solution and circulated buffer should be separated. Similar to lipid extraction with ETC, static electrophoresis resulted in substantial tissue damage in large-volume sample, as it needed a longer time for electrophoresis. This problem was solved with stochastic electrotransport (22). Similar to the device for clearing in stochastic electrotransport, antibodies were confined inside of the sample chamber with nanoporous membranes and PBS was circulated in the outer chamber to control the temperature.

However, these techniques have been poorly explored with RNA studies. Yang et al first used single-molecule fluorescence in situ hybridization (sm-FISH) to detect single RNA molecules in 100-µm-thick PACT-processed mouse brain sections (28). When applying a single-molecule hybridization chain reaction (smHCR), single mRNAs could be detected within 500-µm-thick PACT-processed brain slices (44). Additionally, EDC-CLARITY used EDC (1-Ethyl-3-3-dimethyl-aminopropyl carbodiimide) to link 5′-phosphate groups with surrounding amine-containing proteins to stably retain RNAs in clarified tissues (45). With the hairpin chain reaction (HCR) amplification system, it presented validation for selecting microRNAs, cell-type markers and immediate-early genes.

CLARITY achieved tissue transparency by eluting membrane lipids and lipophilic dyes. Therefore, lipophilic fluorescent dyes, such as DiI, which stains cellular membranes, is unable to be applied with CLARITY (46,47). A study by Jensen and Berg altered the molecular structure of the dye to adhere to both membranes and proteins such that the dye remained in the tissue after tissue clearing (47). These researchers tested three Dil-analogue dyes, CM-DiI, SP-DiI and FM 1-43FX in PACT-processed spinal cords of adult rats and mice. All three dyes remained in the tissue after lipid-clearing, but CM-DiI had the sharpest and FM 1-43FX had the strongest fluorescent signals. This modification provided a new way to label neurons (retrograde or anterograde) and mark the position of extracellular electrodes after electrophysiology.

Before imaging, cleared tissues need to be incubated in solutions that match the average RI of the tissue (~1.46) to achieve RI homogenization. Afterward, tissues would be optically transparent and photon scattering from both the excitation light and the emitted fluorescence signal would be reduced. In theory, the RI of each solution can be matched to the cleared tissue or microscope objective by adjusting concentration of the main ingredient. However, the anti-swelling of cleared tissues should also be considered an important factor affecting the solution. In this study, we contrasted several solutions used in CLARITY-based techniques.

In original protocol, the author applied FocusClear (CelExplorer Labs) and 85–87% glycerol to match RI (14). These reagents both promoted tissue transparency, but the best results were achieved with FocusClear. However, FocusClear is expensive and not suitable for long-term storage (long-term storage resuls in the formation of an irreversible white precipitate) (14,17). Yang et al used a solution termed RIMS (refractive index matching solution) to achieve tissue transparency (28). It contains 88% Histodenz in 0.02 M phosphate buffer with an RI of 1.47. RIMS is stable with long-term storage and preserves fluorescent markers over months. Their group had also derived sRIMS, containing 70% sorbitol in 0.02 M phosphate buffer and cRIMS, containing 88% histodenz in 0.005 M phosphate buffer (23). Next, one study found that 63% 2,2′-thiodiethanol (TDE) in PBS was more suitable for RI matching in CLARITY cleared tissues (48). A different solutuion, 80% Nycodenz solution (nRIMS), was applied with mPACT (30). In ACT-PRESTO, a CUBIC-mount solution (50% Sucrose, 25% urea and 25% N, N, N', N'-tetrakis (2-hydroxypropyl) ethylenediamine) was used to match the RI of the tissues (18). The comparison of these RI-matching solutions can be seen in Table II. However, in a study by Poguzhelskaya et al imaging without an RI-matching solution but in PBST was also possible (36).

Although all of these solutions were intended to match the RI of cleared tissues, for different samples, it may still fail to match the RI well after tissue clearing, as various elements differ in inner tissues. For example, a study by Liu et al reported that 47% TDE in phosphate buffer (not PBS) was more suitable for human brain tissues (37). Therefore, if the concentrations reported in articles do not make the interest tissue ideally transparent, different concentrations and solvents (PBS, phosphate buffer or normal saline) should be tested. Moreover, solutions with a high concentration could reduce tissue swelling. When imaging without an RI-matching solution but in simple purified water or PBST, the tissue will remain swollen, and fail to get the structural information. Therefore, RI-matching and anti-swelling should both be considered when choosing RI-matching solutions.

Finding an optimized, high-resolution deep-imaging approach for large-scale samples is still a major challenge. Furthermore, imaging a 3D volume is a time-consuming task. Therefore, when selecting a microscope and objective, imaging time, resolution and sample size should all be taken into account. In this study, we compared different microscopes and objectives applied to cleared samples.

There are three primary microscopes used for three-dimensional fluorescence imaging: i) The standard confocal microscope; ii) the two-photon microscope; and iii) the light sheet microscope. The standard confocal microscope achieves optical sectioning by using a pinhole in front of the photomultiplier tubes and scans samples point to point. Therefore, this microscope's processing time is notably slow, particularly in large samples, and long-time imaging can lead to photobleaching (20,49). A two photon microscope results in lower photobleaching of fluorophores and provides greater imaging depth than a confocal microscope, but point-by-point scanning is still slow. For the light sheet microscope, it uses fast sCMOS or CCD cameras to image a selectively illuminated focal plane and provides a high imaging speed (2–3 orders of magnitude faster than point-scanning methods), high signal-to-noise ratio and low levels of photobleaching (17,50). Tomer et al developed SPED (SPherical-aberration-assisted Extended Depth-of-field) light sheet microscopy, which combined an extension of the depth-of-field with the optical sectioning of a light sheet microscope, thereby eliminating the need to physically scan detection objectives for volumetric imaging (51). It enabled the scanning of thousands of volumes per-second, limited only by the camera acquisition rate.

The key microscope objective parameters are working distance (WD), numerical aperture (NA) and RI. WD is the distance between the objective lens and the focal plane. A long WD avoids the sample directly contacting with lens and ensures imaging of large samples (52). The NA of an objective relates to the collection of emitted signal, and higher NA enables higher resolution (17). The RI of the cleared tissues is ~1.46; therefore, it is better to choose objectives designed with an RI near 1.46 when using a non-optimized objective.