C57BL/6 male and pregnant female (TP12.5) mice were purchased from KOATECH, Inc. Thy1-eYFP, Tau-GFP (Jackson Laboratory) transgenic mice were used in this study. Pregnant Sprague-Dawley rats (TP18.5) were purchased from Orient-Bio, Inc. All animal care and euthanasia protocols were in accordance with guidelines from Korea University and have been approved by members of the Korea University Institutional Animal Care and Use Committee (KUIACUC-20150520-1).

Male mice (8-weeks-old) were anesthetized with ketamine (30 mg/kg) and xylazine (10 mg/kg). Then, mice were transcardially perfused with 100 ml 1× Phosphate Buffered Saline (PBS) followed by 70 ml 4% paraformaldehyde (PFA) in 1× PBS. Organs from mice were post-fixed in 4% PFA overnight at 4℃. Before electrophoretic tissue clearing, fixed samples were washed with 0.1× PBS several times. Samples were transferred into the cassette for the electrophoretic tissue clearing (ETC) chamber (X-CLARITY, Logos Biosystems, Republic of Korea) and the cassette was placed into the chamber containing 2% SDS and 200 mM boric acid in H₂O (pH 8.5). ETC was processed with following conditions: 1.2 A, 37℃, time depending on the sample thickness (3 h for 1-mm brain slice, 12 h for 2~3mm brain slices, 28 h for whole mouse brain, 12 h for whole mouse embryo, 18 h for whole mouse kidney, 37 h for whole mouse liver and 18 h for mouse liver slices, otherwise mentioned). After tissue clearing, all samples were washed in 0.1× PBS overnight at room temperature while shaking to remove SDS.

Comparison with other clearing methods was performed by following the protocols described in the original published papers [9,26]. For ACT, 8-week-old C57BL/6 mice (KOATECH, Inc, Republic of Korea) were perfusion-fixed with PFA, and the fixed brain was further incubated in hydrogel solution overnight at 4 ℃. Hydrogel-infused brain was polymerized for 2~3 h after degassing. Polymerized brain was cut into 1-mm-thick slices and electrophoresed for fast removal of lipids, using an ETC apparatus with following conditions: 1.5 A, 37 ℃ and 1 h. For FASTClear, fixed mouse slices (1-mm) were incubated in 4% SDS buffer for 12 h at 50 ℃ oven to achieve de-lipidation of the tissue.

Tissue transparency measurement was executed using an adapted protocol previously described [9,29]. Briefly, the brain slices were imaged using a conventional camera before and after clearing. We measured the size and transparency of the cleared 1-mm-thick brain slices, which were outlined and calculated using ImageJ (NIH, USA). The gray value of the cleared sample image was used to measure tissue transparency, and transparency was normalized to the background part of the image. For τ measurements, transparency was acquired by fitting with the one-phase association function.

Oil-Red-O (ORO, O0625, Sigma-Aldrich, USA) was used to measure the amount of lipid droplets in a tissue, as we previously reported [11]. Samples were immersed in ORO (2 mM in 100% isopropanol, 10 µl/mg) and incubated at room temperature for 2 h. ORO-stained tissues were washed in 1% SDS overnight and lipids were extracted by soaking in 100% isopropanol for 4 h, then the extracted fluorescence was measured with a Micro plate reader (SpectraMax Plus384, Molecular Devices, Sunnyvale, CA, USA) at 500 nm.

DiI (468495, Sigma-Aldrich, USA) was used to measure the amount of phospholipids in a tissue. The tissue was immersed in 100% dimethyl sulfoxide (DMSO) for 24 h to allow homogenous loading of DiI deep into the tissue, after which it was again immersed in 0.2 mM DiI in DMSO overnight. Then, the tissue was washed with 1% SDS overnight, followed by the extraction of lipids by soaking in 100% isopropanol for 4 h. The extracted fluorescence was measured using a Micro plate reader (SpectraMax Plus384, Molecular Devices, Sunnyvale, CA, USA) at 550 nm.

Time constant (τ) measurement was achieved using an adapted protocol described previously [11]. Briefly, the glass bottom of a 35-mm-diameter black confocal dish (102350, SPL, Republic of Korea) was covered with a thin polydimethylsiloxane (PDMS) sheet, onto which a tissue stored in PBS was pinned using a short thin tungsten wire to prevent it from moving when the solution was exchanged. The dish was mounted onto the stage of a microscope (EVOS FL, Thermo Fisher Scientific) and filled with PBS, and a reference image was acquired. The RI-matching solution was added to the dish right after the PBS was removed, and time-lapse images were acquired for 1 h with 10-sec interval. The changes in intensity with time for each image was fitted to a onephase association function (fx=y0+p−y0·1−exp−xy) to calculate the maximum values of transparency (p; plateau) and time constant (τ) of diffusion [30].

The percentage of protein loss was obtained by measuring the amount of total protein in the clearing solutions collected from all groups. Protein content was assessed with a Micro plate reader (SpectraMax Plus384, Molecular Devices, Sunnyvale, CA, USA) blanked with the respective solutions and normalized to the weight of the slices before clearing.

Thy1-YFP mice were used for fluorescence intensity measurements. Fluorescent images were captured with a TCS SP8 confocal laser-scanning microscope with a 25× lens, 488-nm excitation wavelength and the same imaging parameters for all groups. Obtained images were converted to 8-bit gray scale (fluorescence intensity from 0~255) using ImageJ, and the mean intensity in the region was calculated from thresholded pixels excluding background fluorescence.

For fluorescence quantification of immunolabeled 1 mm-thick brain slices, the z-stack fluorescence images of tissues were acquired with a TCS SP8 confocal laser-scanning microscope with a 25× lens, 488 and 568-nm excitation wavelength and the same imaging parameters for the three groups. Images were acquired at the surface of brain slices (n=7) and converted to 8-bit gray scale (fluorescence intensity from 0~255), then the total number of pixels and the number of pixels found in each gray value were calculated. Finally, we calculated the highest gray value as the ‘intensity 100’ and showed a histogram of the distribution of gray values in the whole image.

To calculate the signal-to-noise ratio (SNR) for each image, intensity profiles of single images were obtained with ImageJ. SNR was calculated using mean/standard deviation in a uniform region with this formula ‘SNR=20^*log ((intensity signal)/(intensity noise))’.

Tissues were washed with 0.1× PBS several times at room temperature with gentle shaking, and incubated with the primary antibodies in 6% (w/v) bovine serum albumin (BSA), 0.2% (v/v) Triton X-100, 0.01% (w/v) sodium azide in 0.1× PBS for 1~2 days in a 37℃ shaker. Antibodies against Calbindin (1:500, AB1778, Millipore), Collagen type IV (1:500, AB769, Millipore), DCX (1:500, SC-8066, Santa Cruz), GFAP (1:500, z0334, DAKO), GFP (1:1000, ab13970, Abcam), Iba1 (1:500, 019-19741, WACO), Laminin (1:500, L9393, Sigma) and TH (1:1000, AB152, Millipore) were used. Samples were washed several times with 0.1× PBS and incubated with secondary antibodies (1:500) for 1~2 days in a 37℃ shaker that matched the host of each primary antibody for fluorescence imaging. Samples were incubated in CUBIC-mouse solution [9] for at least 1 h. Images were acquired using a TCS SP8 confocal laser-scanning microscope (Leica, Germany). Image reconstructions were performed using LAS X software (Leica) or Imaris imaging software (Bitplane). After imaging, samples were stored in PBS at 4℃.

Data were analyzed with Graphpad Prism software, and results were presented as mean±SEM. Comparisons were performed using unpaired two-tailed Student's t-test and one-way ANOVA followed by Bonferroni post-test when appropriate. A p-value of < 0.05 was considered statistically significant.

The brief workflows for ACT, FASTClear, and FxClear are summarized in Fig. 1A. Because ACT includes a step for acrylamide monomer infusion and polymerization, the total duration for the 1-mm brain slice processing was 3 working days when the fixation step is included. FASTClear also requires 3 days since it needs longer time for passive de-lipidation. On the other hand, FxClear can be done within 2 days owing to active de-lipidation with the electrophoretic tissue clearing (ETC) device (Fig. 1A). We measured the effect of different durations of ETC on the transparency of the 1-mm brain slice with the FxClear protocol. A relatively short ETC processing time (3 h) was sufficient for achieving adequate (>90%) transparency before RI adjustment with CUBIC-Mount (CM), which was equivalent to the 1 h ATC and 12 h FASTClear samples (Fig. 1B).

The expansion of tissues during the ACT process a typical occurrence [9]. Comparing to the ACT which caused a 2-fold increase in the tissue size after 1~3 h of ETC, FxClear only caused 40% and 80% expansion of tissue size from 3 h and 6 h of ETC, respectively (Fig. 1C). This degree of tissue expansion was also seen in 12 h FASTClear, suggesting that ETC expedites the de-lipidation process and tissue expansion 2~4 times faster than the passive protocol. In all protocols, processed tissues returned to their original size after the PBS washes and subsequent RI adjustment with the CUBIC-mount solution (data not shown).

De-lipidation is required for both tissue clarity enhancement and antibody penetration for deep imaging [9,11]. Thus, to directly assess the efficiency of de-lipidation in FxClear protocol, we measured the rate of lipid removal using two different methods which favorably measure oil droplets (Oil red-O, Fig. 2A) and lipid membranes (DiI, Fig. 2B). Interestingly, de-lipidation was not sufficient and only 20% and 70% of lipid removal was achieved by 3 h and 6 h ETC. On the other hand, the ACT procedure can efficiently remove lipids with 1 h ETC. These data suggest that the acrylamide-protein complex may alter tissue chemistry favoring the SDS-based removal of lipid species. Because the contribution of RI adjustment in the tissue-clearing is relatively high in brain tissues [11], 1~3 h of ETC with FxClear technique was sufficient to achieve 95% transparency even when lipids were not completely removed.

The diffusions of the RI solutions (tau value) are affected by multiple factors including tissue porosity, lipid contents, and the micro-architecture of the tissues. Previously, we proposed that tau measurement is an efficient method to rapidly assess the quality of the cleared tissues [11], thus we compared the tau values in several specimens processed with FxClear or ACT protocols. Interestingly, the tau value of 3 h ETC samples was similar to that of 1 h ACT samples, which is consistent with the minimal ETC time of the FxClear protocol required to achieve tissue transparency (Fig. 2C).

During the ACT process, proteins are fixed with PFA and conjugated with polyacrylamide hydrogel, making proteins difficult to be removed. On the other hand, since FxClear does not include acrylamide polymerization, we wanted to examine whether proteins were appropriately preserved during the ETC step of the FxClear protocol (Fig. 3A). Quantification of protein loss during ETC revealed two different phases of protein extraction. In an early phase (0~12 h of ETC), proteins were slowly extracted from the tissue and less than 2% of total proteins were lost. On the other hand, protein removal was significantly accelerated in a later phase (12~72 h of ETC), and the loss of proteins reached >15% by 72 h ETC. Although the underlying cause of the second phase of protein extraction is unclear, FxClear with a 12 h ETC period appears to be safe with >98% of total protein preservation. Accordingly, GFP signals remaining in the 3 h FxClear-processed Thy1-YFP mouse brain slice samples were indistinguishable with samples processed with 1 h ACT (Fig. 3B). Furthermore, similar quality of deep imaging with antibody labeling was achieved with FxClear and ACT-processed samples (Fig. 3C).

Previously, we reported that acrylamide-protein conjugation reduced the immunoreactivity of several antibodies, and about 10% of antibodies which produced reliable signals in PFA-fixed conventional sections failed to exhibit significant signals in ACT samples [9]. Because FxClear does not include acrylamide, we next tested whether FxClear can save immunoreactivities better (Fig. 4A). We randomly chose 5 antibodies (Iba1, GFAP, Tyrosinehydroxylase (TH), Calbindin, DCX), and found that all their immunoreactivities in FxClear tissues were superior to those in ACT tissues. Importantly, two antibodies (Calbindin and DCX) which failed to exhibit significant immunoreactivity with ACT samples showed clear and noticeable immunoreactivity with FxClear 3 h processed samples. Furthermore, the intensities of immunoreactivities in FxClear samples were even greater (10~50% depending on the antibodies) than in conventional PFA-only, thin-sectioned samples (Fig. 4B and 4C, left). Furthermore, signal to noise ratio (SNR) of FxClear sample was also better than other groups (Fig. 4C, right), indicating that FxClear protocol is superior to other methods in the intensity and specificity of the antibody reaction.

So far, we have tested the feasibility of the FxClear protocol with 1-mm-thick brain slices. Finally, we tested whether thicker brain slices or other peripheral organs can be cleared and imaged. Since we found that an ETC longer than 12 h may cause unwanted protein loss, we fixed the ETC time at 12 h and tested the transparency with 2- or 3-mm thick brain slices (Fig. 5A). As a result, 2-mm brain slices reached to >80% transparency in 12 h ETC, while 3-mm slices achieved <60% transparency (Fig. 5B). However, we were able to visualize blood vessels, astrocytes (GFAP-positive) and Purkinje cell (Calbindin-positive) in adult mouse brain by immunostaining with collagen Type IV, GFAP and Calbindin antibodies in 2~3 mm-thick FxClear processed tissue (Fig. 5C and 5D).

In the case when we increased the ETC time to >12 h, whole organs or embryos could be cleared (Fig. 5E), although substantial protein loss was expected. The FxClear-processed tau-GFP mouse embryo (12.5 day) showed strong tau-GFP fluorescence in blood vessels and nerve fiber layer (NFL) but little-to-no fluorescence in neural retina (Fig. 5F and Supplementary Video 1). This is very similar to the tau-GFP expression reported in this developmental stage [31]. Collectively, these data demonstrated that various thick organs can be cleared with the FxClear protocol, despite the concerns about limited tissue transparency and potential loss of proteins.