Click3D: Click reaction across deep tissues for whole-organ 3D fluorescence imaging

Abstract

Click chemistry offers various applications through efficient bioorthogonal reactions. In bioimaging, pretargeting strategies have often been used, using click reactions between molecular probes with a click handle and reporter molecules that make them observable. Recent efforts have integrated tissue-clearing techniques with fluorescent labeling through click chemistry, allowing high-resolution three-dimensional fluorescence imaging. Nevertheless, these techniques have faced a challenge in limited staining depth, confining their use to imaging tissue sections or partial organs. In this study, we introduce Click3D, a method for thoroughly staining whole organs using click chemistry. We identified click reaction conditions that improve staining depth with our custom-developed assay. The Click3D protocol exhibits a greater staining depth compared to conventional methods. Using Click3D, we have successfully achieved whole-kidney imaging of nascent RNA and whole-tumor imaging of hypoxia. We have also accomplished whole-brain imaging of hypoxia by using the clickable hypoxia probe, which has a small size and, therefore, has high permeability to cross the blood-brain barrier.

Click3D, an optimized method for fluorescent staining with click chemistry, enables whole-organ 3D imaging of clickable probes.

INTRODUCTION

Click chemistry is a highly valuable chemical technology that enables the bioorthogonal functionalization of molecules under biological environments. In preclinical bioimaging conducted in vitro, ex vivo, and in vivo, pretargeting strategies have often been used. These strategies use click reactions between target molecules with a click handle and reporter molecules that make them observable (1). Small click handles, such as alkyne and azide, allow for visualizing the biodistribution of target molecules while largely preserving their intrinsic properties, which is a notable improvement over the direct conjugation of larger reporter molecules such as fluorophores (2).

In in vivo bioimaging, including magnetic resonance imaging (MRI) and positron emission tomography (PET), click chemistry facilitates three-dimensional (3D) imaging of the distribution of target molecules including biomacromolecules, ligands, and drugs (1, 3). This is achieved by click reactions between reporter molecules, such as MRI contrast agents or PET tracers, and target molecules with click handles that are preadministered in vivo. Such click chemistry–coupled PET or MRI allows molecular analysis at the whole-tissue and whole-body levels. However, these methods are limited in resolution, ranging from hundreds of micrometer to millimeter scales, thus hindering the cellular-level observation (4).

On the other hand, fluorescence imaging, using visible light as the measurement modality, provides higher resolution at the scale of tens to hundreds of nanometers, enabling detailed visualization of the in vivo distribution of molecules (5). However, this approach faces a different challenge: the low tissue penetration of light, which restricts imaging depth. Although attempts have been made to overcome this limitation through the development of long-wavelength fluorescent dyes (6) and two-photon fluorescence microscopy (7), high-resolution analysis is generally limited to tissue surfaces and sections, making it difficult to analyze entire organs in detail with cellular-level resolution.

Tissue clearing, which increases light penetration by making biological tissues transparent, represents a state-of-the-art technology for whole-organ fluorescent imaging at cellular resolution (8). This technology has successfully visualized the localization of endogenous biomolecules such as proteins and nucleic acids through techniques such as immunostaining, fluorescence in situ hybridization, and genetically encoded fluorescent proteins (9). Integrating tissue-clearing technology with click chemistry holds promise for enabling high-resolution 3D imaging of the in vivo biodistribution of a broader range of molecules. Recently, 3D fluorescence imaging of covalent drugs and DNA synthesis markers with alkyne click handles was achieved by in situ fluorescence staining using Cu-catalyzed azide-alkyne cycloaddition (CuAAC) and tissue clearing (10, 11). These breakthroughs provide valuable insights yet are limited by a maximum observation depth. This limitation restricts their use to tissue slices (<500 μm thick) or partial organs, thereby hindering the imaging of large tissues at the scale of entire mouse organs. The limited observation depth may be largely attributed to the low tissue permeability of click reagents, which was not extensively discussed in conventional studies. To achieve whole-organ 3D imaging with click chemistry, it is crucial to understand the factors influencing the permeability of click reagents. Furthermore, it is necessary to develop click chemistry specifically tailored for staining 3D tissues.

Here, we report Click3D, a method that realizes efficient fluorescent staining of whole tissues and organs using click chemistry. We first developed an original assay for the semiquantitative evaluation of click staining depth in 3D tissues, leading to the identification of key conditions that substantially enhance the staining depth. Using these insights, we established the Click3D method for fluorescent staining of clickable molecules in whole organs or tissues, as shown in Fig. 1A. Notably, the optimized Click3D method enabled 3D click staining of a whole organ, including a brain (Fig. 1B). This method was successfully applied to whole-kidney mapping of nascent RNAs. In addition, using an alkyne-tagged hypoxia probe, we achieved high-resolution 3D mapping of hypoxia in whole tumors and brains, which could not be achieved by conventional click staining methods.

Fig. 1. Overview of this study.

(A) Schematic illustration of the whole-organ imaging strategy using a chemical probe in combination with Click3D, a whole-organ staining method based on CuAAC. (B) Previous works and this work for click chemistry–based 3D staining with tissue clearing. Created using BioRender.com.

RESULTS

Construction of a semiquantitative assay for click reaction conditions in deep staining

To date, two methods have been reported for achieving 3D high-resolution fluorescence imaging of target molecules in adult mouse tissues, leveraging both tissue clearing and click chemistry. These are the clearing-assisted tissue click chemistry (CATCH) method, developed by Pang et al. (10), and another method by Lazutkin et al., hereinafter referred to as Method A (11). Both methods used CuAAC with azide fluorescent dyes on alkyne-labeled biological samples. The CATCH method was applied to visualize the distribution of covalent drugs, while Method A was used to image neurogenesis with a DNA synthesis marker, 5-ethynyl-2′-deoxyuridine. However, these methods were demonstrated only up to a maximum thickness of approximately 500 μm to 1 mm. Our initial objective was to assess the effectiveness of these two existing methods for 3D fluorescence imaging of larger tissues.

We devised an experimental assay using paraformaldehyde (PFA)–fixed and delipidated cylindrical liver tissue blocks to assess the staining depth of the click reaction (Fig. 2A). To semiquantitatively assess the click staining depth, we used liver tissue blocks with homogeneously distributed alkynes. This was achieved by reacting iodoacetamide (IAM) with endogenous thiols present in tissues. Specifically, IAM-hexynyl, an alkyne-conjugated IAM, was used to modify thiol groups in the tissue blocks (Fig. 2, B and C). However, the permeation of IAM-hexynyl into the tissue blocks was slower than its reaction, resulting in strong labeling only at the tissue surfaces. To address this, we simultaneously used IAM-hexyl, a competitive reactant with thiols, with IAM-hexynyl. Under conditions of 1 mM IAM-hexyl and 2 μM IAM-hexynyl (0.2% of total IAM concentration), cylindrical liver tissue blocks, approximately 2 mm in diameter, were homogeneously labeled with alkynes (fig. S1).

Fig. 2. Development of Click3D method.

(A) Schematic illustration of the click staining depth assay. (B) Synthesized compounds used for the assay. (C) Homogeneous alkyne labeling using IAM-hexynyl and IAM-hexyl. (D to G) Dependency of click staining depth on the concentration of each click reagent. Constant conditions other than the reagents tested: (Cu preincubation) 0.3 mM CuSO₄, 0.6 mM BTTP, 150 mM NaCl, 10% (v/v) DMSO in 10 mM Hepes (pH 7.3), 37°C, 2 days; (Reaction) 0.3 mM CuSO₄, 0.6 mM BTTP, 5 μM Cy5-azide, 2.5 mM NaAsc, 150 mM NaCl, 10% (v/v) DMSO in 10 mM Hepes (pH 7.3), room temperature (r.t.), 1 hour. Scale bar, 1 mm. (H) Dependency on the species of Cu ligand. Conditions: (Cu preincubation) 2 mM CuSO₄, 4 mM ligand, 150 mM NaCl, 10% (v/v) DMSO in 10 mM Hepes (pH 7.3), 37°C, 6 hours; (Reaction) 2 mM CuSO₄, 4 mM ligand, 15 μM Cy5-azide, 100 mM NaAsc, 150 mM NaCl, 10% (v/v) DMSO in 10 mM Hepes (pH 7.3), r.t., 30 min. Scale bar, 1 mm. (I) Dependency on NaCl concentration. Conditions: (Cu preincubation) 2 mM CuSO₄, 4 mM THPTA, 150 to 900 mM NaCl, 10% (v/v) DMSO in 10 mM Hepes (pH 7.3), 37°C, 6 hours; (Reaction) 2 mM CuSO₄, 4 mM THPTA, 15 μM Cy5-azide, 100 mM NaAsc, 150 to 900 mM NaCl, 10% (v/v) DMSO in 10 mM Hepes (pH 7.3), r.t., 30 min. Scale bar, 1 mm. (J) A Click3D protocol. O/N, overnight. (K) Benchmark of Click3D (only 2 hours for reaction in this experiment) against conventional methods, CATCH and Method A. For fair comparison of staining efficiencies, in all the methods, Cy5-azide and Click3D-W were used for dye and washing solution, respectively. Scale bar, 1 mm. The intensity profiles in (D) to (I) and (K) indicate means ± SD. (C) and (J) created using BioRender.com.

Subsequently, the alkyne-labeled cylindrical liver tissue blocks were subjected to CuAAC with an azide-fluorescent dye under various conditions. The tissue was then made transparent with BABB (benzyl alcohol/benzyl benzoate = 1:2), an organic solvent–based refractive index matching solution (RIMS) (12), and imaged using light sheet fluorescence microscopy (LSFM). A z-slice deep enough from the top surface of the block (approximately 0.6-mm depth) was selected, and the staining depth was semiquantitatively evaluated by plotting fluorescence intensity from the tissue surface to the center (Fig. 2A, right). Higher fluorescence intensity at the tissue edge indicates higher reactivity, while higher fluorescence intensity at the tissue center represents better permeability of the click reagents.

Using this assay, we initially evaluated the CATCH method and Method A (fig. S2, A to C). The results indicated that click reactions had limited staining depths of approximately 200 to 300 μm. This suggested that while these existing methods are effective for relatively thin sections, they are not suitable for large organ-level samples.

Construction of the Click3D protocol

In our study, we used the CATCH method as a reference to identify factors impeding effective 3D staining via CuAAC in large tissues. In the CATCH protocol, AF647-picolyl azide (5 μM), CuSO₄ (0.3 mM), and 3-(4-((bis((1-(tert-butyl)-1H-1,2,3-triazol-4-yl)methyl)amino)methyl)-1H-1,2,3-triazol-1-yl)propan-1-ol (BTTP) (0.6 mM) were preincubated with PFA-fixed and alkyne-labeled tissues in 10% dimethyl sulfoxide (DMSO)/phosphate-buffered saline (PBS) for 2 days. This was followed by click reaction in the presence of sodium ascorbate (NaAsc) (2.5 mM), and subsequent fluorescence imaging after the tissues were made transparent. When applying the CATCH method for fluorescent staining of alkyne-unlabeled cylindrical liver tissues in our original assay, we observed that the samples colored blue in bright-field images (fig. S2D). We hypothesized that this was due to the interaction of cationic Cu(II) ions with the PFA-fixed and delipidated tissue, which is reported as anionic (13), causing the ions to remain even after staining and washing. These residual Cu(II) ions can absorb excitation light (14), posing a particular problem in LSFM imaging where the excitation light sheet is irradiated from the side of tissues, leading to nonuniform imaging. To address this issue, we washed the tissue samples with a 0.1 M aqueous solution of EDTA, a copper chelator. After washing, we noted substantial decolorization of the tissue as the Cu(II) ions were removed and a remarkable recovery of fluorescence (fig. S2E). This finding underscores the importance of removing residual copper ions with EDTA wash for effective LSFM fluorescence imaging in 3D tissues.

In our pursuit to enhance click staining depth, we addressed the issue of low permeability of cationic copper ions, which were predicted to interact strongly with tissue as described above. Similar to the CATCH method, we opted for a 2-day preincubation with the copper catalyst, but without the dye, to minimize background signals. Considering the potential of the Cu-chelating azide dye (AF647-picolyl azide) used in the original CATCH method to decrease permeability when complexed with cationic copper ions (15), we instead used a normal azide dye (Cy5-azide). In addition, to avoid the issues of copper precipitation in phosphate buffers, we used Hepes buffer instead of PBS (16).

Under these modified conditions, we experimented with varying concentrations of dye (Cy5-azide), copper catalyst ([CuSO₄]:[BTTP] = 1:2), and reducing agent (NaAsc). The reaction time was limited to 1 hour to assess the states where the staining was not saturated. The results indicated that changes in the copper catalyst concentration did not notably affect the staining depth under these conditions (Fig. 2D). A high dye concentration resulted in an increase in fluorescence within roughly 0.2 mm of the tissue surface, but there was no substantial increase in fluorescence intensity beyond this depth, indicating no improvement in the staining depth (Fig. 2E). Contrastingly, the high concentration of NaAsc substantially improved the staining depth (Fig. 2F). Along with an increase in copper catalyst concentration from 0.3 to 2 mM, the high concentration of NaAsc resulted in even greater staining depth (Fig. 2G). These results indicate that the copper catalyst and NaAsc have limited tissue permeability, which is the cause of the difficulty in applying click staining to large samples. Therefore, increasing the concentrations of these chemicals shows promise for application in whole organs and tissues.

We next examined various copper ligand species to enhance click staining depth (Fig. 2H). Besides BTTP (17), used in the original CATCH method, we tested the water-soluble copper ligands 2-(4-((bis((1-(tert-butyl)-1H-1,2,3-triazol-4-yl)methyl)amino)methyl)-1H-1,2,3-triazol-1-yl)acetic acid (BTTAA) (18) and tris((3-hydroxypropyl-1H-1,2,3-triazol-4-yl)methyl)amine (THPTA) (19) (fig. S3), with staining conditions set as Cy5-azide (15 μM), copper catalyst (2 mM), and NaAsc (100 mM) (preincubation for 6 hours and reaction for 30 min). Given that small molecules with a negative charge have been reported to have high tissue permeability (13), BTTAA, which has a carboxylate group, seemed promising. However, contrary to expectations, BTTAA showed high permeability but substantially lower reactivity, resulting in less staining depth than the other ligands. Although the hydrophilic carboxylate of BTTAA might enhance the permeability of the copper catalyst, the unique environment of 3D tissue and high concentration conditions might have negatively affected the reactivity of BTTAA. Comparing BTTP and THPTA, the bulky tert-butyl group of BTTP stabilizes the Cu(I) ion and may contribute to higher reactivity (20, 21). However, its increased hydrophobicity may have led to more interaction with tissues and reduced permeability. THPTA, meanwhile, showed the best balance between reactivity and permeability. THPTA was deemed more suitable for click reactions in 3D tissues than BTTP and BTTAA.

We also investigated whether increasing salt concentration could improve the permeability of click reagents (Fig. 2I). It has been reported that increasing salt concentration greatly enhances the tissue permeability of antibodies (13). In our case, higher salt concentrations slightly decreased the reactivity, likely because of chloride ions slowing the reaction by coordinating with copper (22). However, it enhanced the permeability possibly by reducing Donnan’s potential and electrostatic interactions, which impede the permeation of molecules. These results suggest that increasing salt concentration has a certain effect, albeit not very substantial, on the tissue permeability of click reagents.

After additional minor condition optimization, we finalized the Click3D protocol (Fig. 2J). This protocol consists of Click3D-C for Cu preincubation, Click3D-R for reaction, and Click3D-W for washing. The tissue is initially incubated in Click3D-C with a high concentration of copper catalyst for 2 days to allow sufficient permeation. Subsequently, it is incubated with Click3D-R, containing fluorescent dye and a high concentration of NaAsc, for the minimum staining time required. Last, Click3D-W with EDTA is used to thoroughly remove the copper catalyst that could interfere with LSFM imaging. Using this optimized Click3D method, we confirmed that almost all alkynes were subjected to the click reactions (fig. S4) and uniformly fluorescently labeled in a whole tissue block, and the staining depth was markedly improved in contrast to conventional methods (Fig. 2K). Click3D enabled deep staining and imaging of 3D tissues through efficient permeation processes with optimized solution conditions and a washing process that sufficiently removes Cu(II) ions (fig. S5).

Whole-tissue imaging of nascent RNA

We then used the developed Click3D method to image nascent RNA in mice, using an RNA synthesis marker 5-ethynyluridine (5-EU), an alkyne-labeled uridine analog (Fig. 3A). Because 5-EU is incorporated into nascent RNA in vivo, its subsequent click staining enables the visualization of RNA transcription. However, imaging of a whole mouse organ using this technique had not been demonstrated.

Fig. 3. Whole-kidney imaging of nascent RNAs using Click3D.

(A) Chemical structure of 5-EU. (B) Schematic illustration of the assay. Delipidation: MeOH. RI matching: BABB. (C) Whole-kidney 3D mapping of nascent RNAs. Upper left: 3D-rendered image. Bottom left: A representative 2D section. Right: 200-μm-thick section of 3D-imaged kidney. OSOM, outer stripe of outer medulla; ISOM, inner stripe of outer medulla. Scale bar, 1 mm (zoomed: 500 μm). (D) LSFM imaging result of vehicle control sample, stained by Click3D. Scale bar, 1 mm. (E) LSFM imaging results of 5-EU administered sample, stained by conventional methods, CATCH and Method A. The contrasts of them are higher than that of Click3D results in (C). Scale bar, 1 mm. Voxel size of the LSFM image is 4.13 μm by 4.13 μm by 4.1 μm.

Following the intraperitoneal administration of 5-EU, mouse kidneys were harvested after 5 hours. The organs were then stained with Cy5-PEG₄-azide, a hydrophilic dye we previously developed (23), using the Click3D method and cleared using BABB for 3D imaging (Fig. 3B). The LSFM imaging results successfully exhibited 3D visualization of the incorporated 5-EU (Fig. 3C, upper left). Cross-sectional analysis revealed that the signal was particularly strong in the cortex, especially on the inner side (Fig. 3C, bottom left). However, not all tubules were detected, suggesting potential variability in RNA transcriptional activity among different or even the same types of tubules (Fig. 3C, right). In a vehicle control group without 5-EU, the signal was negligible under the same imaging conditions (Fig. 3D). When the 5-EU–administered samples were stained using conventional methods such as CATCH and Method A, limited staining was observed, demonstrating the superior staining depth of Click3D (Fig. 3E). Therefore, Click3D has been confirmed to enable effective 3D staining in whole organs. As shown in Fig. 3C, by combining in vivo labeling of nascent RNA with 5-EU, RNA transcriptional activity in kidney tubular cells was visualized comprehensively. Kidney tubules, which form intricate microstructures along with other components such as blood vessels and collecting ducts, have been the focus of several studies aiming to elucidate their morphology and function using tissue clearing–based imaging (24). The 3D mapping of RNA transcriptional activity, correlating with cell division rate and metabolic activity, will contribute to further analysis of the complexity of the function and morphology.

Whole-tumor imaging of hypoxia

In our next experiment, we focused on applying a functional chemical probe to target hypoxia, a critical microenvironment implicated in various diseases including cancers and central nervous system disorders. We used Pimo-yne, a clickable probe for detecting hypoxia that we previously developed for 2D section imaging (Fig. 4A) (23). Pimo-yne is an alkyne-tagged derivative of pimonidazole (Pimo), a gold standard for staining hypoxic regions (25), and exhibits similar staining properties to Pimo (23). Under hypoxic conditions, the nitroimidazole moiety of pimonidazole is reduced to an electrophilic species, which then covalently binds to nucleophilic residues in neighboring proteins, enabling Pimo-yne to label hypoxic regions (23, 26).

Fig. 4. Whole-tumor hypoxia imaging of HeLa/5HRE-d2EGFP tumor using Click3D.

(A) Chemical structure of Pimo-yne. (B) Schematic illustration of the assay. Blood vessels were stained by intravenous injection of AF647-conjugated anti-CD31 antibody 10 min before sacrifice. Delipidation: 10% 1,2-hexanediol (HxD), 5% Triton X-100, 1 mM EDTA, 10 mM N-butyldiethanolamine (NBDEA). RI matching: CUBIC-R+(M). Created using BioRender.com. (C) Whole-tumor 3D imaging of mild hypoxia (HIF-1, EGFP), severe hypoxia (Pimo-yne, Cy3), and blood vessels (CD31, AF647). Scale bar, 1 mm. (D) A merged image of all three channels displayed in (C). The white box shows the ROI. Scale bar, 1 mm. (E) A representative 2D section in the ROI shown in (D). Scale bar, 200 μm. (F) Surface rendering. For HIF-1 and Pimo-yne, the ROI is shown in (D). For CD31, the surface analysis was performed in larger regions than the ROI for HIF-1 and Pimo-yne, to properly calculate the distances between hypoxia and blood vessels. White arrows show Pimo-yne–positive cores on the tumor surface. Scale bar, 300 μm. (G) Violin plot of the distance between voxels in HIF-1– or Pimo-yne–positive regions [inside the surface-rendered regions in (D)] and the nearest voxel in CD31-positive regions (blood vessel). Voxel size of the LSFM image is 5.16 μm by 5.16 μm by 5.2 μm.

Our first attempt involved whole-tumor hypoxia imaging using Pimo-yne. In cancer, oxygen consumption/supply imbalance, due to increased consumption and incomplete vascular network formation, leads to hypoxia in regions distant from blood vessels (27). Hypoxia-inducible factor–1 (HIF-1), activated under hypoxic conditions, is crucial in cancer, as it induces genes involved in angiogenesis, metastasis, and invasion (28). HIF-1–positive regions, reported to be distributed in mild hypoxia near blood vessels, differ from Pimo-positive regions, reported to be distributed in severe hypoxia far from blood vessels (29). Thus, detailed 3D mapping of HIF-1, Pimo, and blood vessels can deepen our understanding of tumor hypoxia. For this purpose, we used tumor-bearing mice with HeLa/5HRE-d2EGFP cells, which express d2EGFP under the control of HIF-1–responsive 5HRE promoter, staining HIF-1–positive regions in green fluorescence (30). After intravenous administration of Pimo-yne, tumors were harvested (Fig. 4B). For vascular staining, an AF647-labeled anti-CD31 antibody was administered intravenously shortly before perfusion. The tissues were then subjected to Click3D staining with Cy3-PEG₄-azide, a hydrophilic dye (23), and cleared using an aqueous RIMS, CUBIC-R+(M) (31). The aqueous RIMS was used to observe fluorescent proteins. The results successfully demonstrated simultaneous staining of HIF-1 (green), Pimo-yne (red), and blood vessels (cyan) (Fig. 4, C and D, and movie S1). It should be noted that signal intensity was attenuated in deeper tumor areas due to necrotic regions and low transparency of the tumor with an aqueous-based clearing method. The intensity of HIF-1 reporter–enhanced green fluorescent protein (EGFP) and CD31 antibody AF647 remained constant, with or without Click3D treatment (fig. S6), suggesting that Click3D is compatible with 3D tissue clearing–based imaging using fluorescent proteins and antibodies.

A detailed analysis revealed closely located but distinct positive areas for HIF-1 and Pimo-yne, indicating different oxygen concentration ranges detected by HIF-1 and pimonidazole (Fig. 4, D and E). We created a region of interest (ROI) near the tissue surface (approximately 2-mm depth) with high resolution, and surface-rendered HIF-1–, Pimo-yne–, and CD31-positive regions (Fig. 4F and movie S1). The analysis showed that hypoxic cells were densely located in the areas without blood vessels, with HIF-1–positive mild hypoxic regions on the exterior and Pimo-yne–positive severe hypoxic regions on the interior. On the tumor surface in contact with the epidermis, the Pimo-yne–positive core was exposed to the surface of hypoxic regions (as indicated with white arrows in Fig. 4F, left). In addition, the shortest distance from blood vessels to each voxel within the HIF-1– and Pimo-yne–positive regions was quantified (Fig. 4G). The results revealed that HIF-1–positive areas were closer to vessels (median, 103 μm) than Pimo-yne–positive areas (median, 136 μm), indicating that pimonidazole labels more severely hypoxic regions compared to HIF-1. Distance analysis, previously limited to 2D sections, was now successfully conducted unbiasedly in 3D. The simultaneous 3D mapping of chemical probes, fluorescent proteins, and immunostaining could be a potent tool for revealing heterogeneous microenvironments, such as those found in tumors, in a single comprehensive analysis.

Whole-brain hypoxia imaging in disease mouse models

We then addressed the challenge of whole-brain hypoxia imaging. To date, there have been no reports of 3D imaging of hypoxia in a whole brain at single-cell resolution. Acute hypoxia in the brain, caused by conditions such as stroke and traumatic brain injury, seriously affects the central nervous system (32, 33). Therefore, it is crucial to clarify the hypoxia-related pathology at the cellular level. Pimonidazole is known for its sufficient brain uptake and ability to label hypoxic brain parenchyma regions (34). Pimo-yne, a minimally tagged derivative of pimonidazole, was expected to cross the brain through blood-brain barrier (BBB), detect brain hypoxia, and enable whole-brain 3D imaging of hypoxia through Click3D. To test this, we used a mouse model of hypoxemia. Mice treated with Pimo-yne were exposed to 8% O₂ for 30 min to induce hypoxemia and then returned to room air. After 1.5 hours, brains were harvested following blood removal and PFA fixation by perfusion (Fig. 5A). For vascular staining, an AF647-labeled anti-CD31 antibody was administered intravenously shortly before perfusion. Following Click3D staining and BABB clearing, LSFM imaging was performed. The results successfully detected fluorescent signals induced by hypoxemia across whole brains (fig. S7A). Costaining of brains with Pimo-yne and pimonidazole in 2D sections supported that Pimo-yne stained almost the same hypoxic regions as pimonidazole (fig. S7B). We used the rolling-ball algorithm for background subtraction, commonly used in tissue clearing–based imaging, allowing us to overlay hypoxic regions on vascular vessels stained by CD31 antibody, revealing the spatial relationship between hypoxic cell clusters and cerebral blood vessels (Fig. 5, B and C, and movie S2).

Fig. 5. Whole-brain hypoxia imaging of hypoxemia model mouse using Click3D.

(A) Schematic illustration of the assay. Delipidation: 10% HxD. RI matching: BABB.Created using BioRender.com. (B) Whole-brain 3D imaging of hypoxia (Pimo-yne, Cy3) and blood vessels (CD31, AF647) in hypoxemia model with Pimo-yne. The images are after rolling-ball algorithm background subtraction. The images without background subtraction are shown in fig. S7A. Blood vessels were stained by intravenous injection of AF647-conjugated anti-CD31 antibody 10 min before sacrifice. Scale bar, 1 mm. (C) Upper left: A merged image of two channels displayed in (B). Upper middle: Horizontal view (300 μm of z projection). Upper right: Zoomed view of ROI (i). Middle right: Zoomed view of ROI (ii). Bottom left: Sagittal view (300 μm of x projection). Bottom middle: Coronal view (300 μm of y projection). Scale bar, 1 mm (zoomed: 200 μm). (D) Chemical structures of Pimo-Me-Si-Rhodol and Pimo-BODIPY. (E) Whole-brain 3D imaging in hypoxemia model with Pimo-Me-Si-Rhodol. Scale bar, 1 mm. (F) Chemical structure of Pimo-yne. Voxel size of the LSFM image is 6.45 μm by 6.45 μm by 6.5 μm.

For comparison, we conducted a similar experiment using pimonidazole-fluorophore conjugates, Pimo-Me-Si-Rhodol and Pimo-BODIPY, which we had previously developed for tissue clearing–based imaging of hypoxia (Fig. 5D) (35). However, these probes yielded few substantial signals of cerebral hypoxia (Fig. 5E and fig. S7C versus fig. S7A), suggesting that these compounds lack BBB permeability and are thus unsuitable for 3D brain hypoxia imaging.

The BBB strictly restricts the penetration of compounds into the brain parenchyma, generally allowing only small and lipophilic compounds to permeate (36). The threshold for BBB permeability is around 400 Da for molecular weight and 8 for the number of hydrogen bonds (hydrogen bond donors and acceptors) (37). Conventional functional fluorescent probes, which are designed as conjugates of a sensing moiety and a fluorophore, often exceed this threshold because of the large size of fluorophores. In addition, achieving a balance between lipophilicity for BBB permeability and practical water solubility poses a challenge. In contrast, minimally modifying the sensing moiety with a clickable tag, as in the case of Pimo-yne, can preserve BBB permeability if the sensing moiety is inherently BBB-permeable. Pimo-Me-Si-Rhodol and Pimo-BODIPY, which did not translocate into brain, have molecular weights of 681 and 557 Da, respectively, whereas Pimo-yne, which successfully translocated into brain, has a molecular weight of only 308 Da (Fig. 5, D and F). This characteristic of Pimo-yne contributed to achieving high-resolution whole-brain 3D imaging of hypoxia. Our findings suggest that reducing the molecular weight of chemical probes based on the Click3D platform is a highly effective strategy for brain imaging.

DISCUSSION

In this study, we successfully developed the Click3D method, an innovative approach for efficient fluorescent staining of whole organs using the CuAAC reaction. The Click3D protocol was established by optimizing the various conditions with our custom-developed click staining depth assay. This Click3D surpasses the limitations of conventional 3D click staining, which is confined to tissue sections or partial organs, enabling whole-organ 3D staining of clickable probes.

We demonstrated the effectiveness of Click3D using the RNA synthesis marker 5-EU and the hypoxia probe Pimo-yne, achieving unprecedented whole-organ/whole-tissue 3D imaging of nascent RNA and hypoxia. In addition, Click3D has proven to be compatible with fluorescent proteins and vascular staining using CD31 antibodies. The simultaneous use of fluorescent proteins, which is a standard method in tissue-clearing imaging, further emphasizes the versatility of Click3D. Moreover, the integration of Click3D with immunostaining and in situ hybridization for RNA is expected to make notable contributions to biomedical research.

Particularly notable is the application of Click3D in combination with Pimo-yne, facilitating high-resolution 3D whole-brain hypoxia imaging in a hypoxemia model. This contrasts sharply with the limitations of existing tissue-clearing compatible hypoxia probes, such as Pimo-Me-Si-Rhodol or Pimo-BODIPY, which failed to detect brain hypoxia. The minimal modification of Pimo-yne ensured its BBB permeability and effective translocation into the brain. The complexity of the brain, with its myriad of distinct regions, underscores the importance of comprehensive 3D imaging. Click3D could provide essential design guidelines for developing BBB-permeable probes suitable for tissue-clearing imaging. It should be noted, however, that the staining regions of Pimo and Pimo-yne in the brain overlap considerably but are slightly different (fig. S7B). This may be due to differences in biodistribution, including BBB permeability, and fluorescent staining methods. Future development of clickable chemical probes that more accurately detect hypoxic regions would provide further progress in this field.

The Click3D method, developed in this study, is a technique that integrates the versatile scope of click chemistry into whole-organ tissue-clearing-based imaging. The efficient labeling of targets in deep sites of tissue, achieved by the optimized Click3D protocol, is a substantial advantage over the conventional antibody-based tissue staining method. Extending beyond the traditional scope of anatomical and biomolecular localization analysis in tissue clearing–based imaging, Click3D expands the range of observable targets, including drug distributions and microenvironments detected by chemical probes. This technology will substantially advance biomedical research, enhancing our understanding of biological phenomena and disease pathology.

MATERIALS AND METHODS

Click staining depth assay

BALB/c mice were anesthetized by isoflurane and transcardially perfused with heparinized PBS and 4% (w/v) PFA in 0.1 M phosphate buffer (PB). The livers were extracted, fixed in 4% (w/v) PFA in 0.1 M PB at 4°C overnight, and washed with PBS. The liver tissues were delipidated with nonbuffered 10% (v/v) 1,2-hexanediol (HxD) in water overnight twice at 45°C with shaking at 100 rpm. After PBS wash, the liver tissues were cut into cylindrical tissue blocks (ϕ ca. 2 mm) with a stainless steel circle punch. The tissue blocks were immersed in a mixture of 1 mM IAM-hexyl and 2 μM IAM-hexynyl in 10 mM PB (pH 7) containing 1 M NaCl and 1% (v/v) DMSO overnight at room temperature (r.t.) with gentle shaking. After PBS wash, the blocks were immersed in the Cu preincubation solution and the reaction solution (for detailed conditions, see the legend of Fig. 2 and the table in fig. S2B) sequentially with shaking. Unless otherwise noted, the blocks were washed with Click3D-W for 1 hour twice and overnight. After PBS wash, they were dehydrated with 60% (v/v) MeOH/H₂O for 1 hour, 80% (v/v) MeOH/H₂O for 1 hour, 100% MeOH overnight, and 100% MeOH for 1 hour, at r.t. with gentle shaking. Then, the blocks were cleared with BABB for 2 hours twice at r.t. with shaking. The whole-tissue images were acquired with a custom-built LSFM. A z-slice deep enough from the top surface of the block was manually picked, ensuring there were no large holes or cracks. Then, three lines along the diameter of the circular tissue slice were manually drawn and the intensity profiles were calculated. Each of the three profiles at half their length was held to make a total of six profiles, and the mean intensity and SD were graphed by custom-written Python code.

Mouse models

For nascent RNA labeling experiments, 5-EU (80 mg/kg) in sterile PBS (20 mg/ml) was intraperitoneally injected into BALB/c mice. After 5 hours, the mice were anesthetized by isoflurane and transcardially perfused with ice-cold sterile heparinized PBS and 4% (w/v) PFA in 0.1 M PB.

For the tumor-bearing mouse model, HeLa/5HRE-d2EGFP cells were subcutaneously inoculated to both hindlegs of BALB/c-nu/nu mice. After 8 days, Pimo-yne (30 mg/kg) in the sterile vehicle solution [5% (v/v) DMSO, 45% (v/v) polyethylene glycol, molecular weight 400 (PEG-400), 50% (v/v) PBS, 6 mg/ml] was intravenously injected into the mice. After 2 hours, the mice were anesthetized by isoflurane and transcardially perfused with ice-cold sterile heparinized PBS and 4% (w/v) PFA in 0.1 M PB. For blood vessel staining, 10 μg of AF647-conjugated anti-CD31 antibody in sterile PBS (0.1 μg/ml) was intravenously injected 10 min before the sacrifice.

For the hypoxemia mouse model, Pimo-yne (60 mg/kg) in the sterile vehicle solution [5% (v/v) DMSO, 45% (v/v) PEG-400, 50% (v/v) PBS, 12 mg/ml] was intravenously injected into BALB/c mice under normal air conditions. After 5 to 10 min, the mice were exposed to 8 to 9% O₂ atmosphere for 30 min and then returned to room air. Two hours after the probe administration, the mice were anesthetized by isoflurane and transcardially perfused with ice-cold sterile heparinized PBS and 4% (w/v) PFA in 0.1 M PB. For blood vessel staining, AF647-conjugated anti-CD31 antibody was intravenously injected as described above.

Click3D

For Cu preincubation, the delipidated tissues were immersed in Click3D-C at 37°C for 2 days (for brain samples, the solution was refreshed after 1 day) with shaking at 100 rpm. For reaction, the tissues were immersed in Click3D-R at r.t. for 2 hours twice (three times for brain samples) with gentle shaking. The tissues were washed with Click3D-W at r.t. for 2 hours twice and overnight with gentle shaking. Then, they were washed with PBS at r.t. for 1 hour three times with gentle shaking.

Supplementary Materials

This PDF file includes:

Supplementary Methods

Figs. S1 to S7

Legends for movies S1 and S2

References

Other Supplementary Material for this manuscript includes the following:

Movies S1 and S2