Hybrid open-top light-sheet microscope
Light-sheet-based imaging is achieved using three optical arms. The first obliquely illuminates the specimen with a light sheet at 45 deg relative to the vertical axis. The second is oriented vertically, enabling high-resolution collection of the light-sheet-generated fluorescence in a NODO configuration. Finally, the third arm is oriented at 90 deg relative to the illumination light sheet, enabling high-speed mesoscopic imaging in an ODO configuration. All three objectives are positioned below the specimen and are sealed into a custom-designed monolithic imaging chamber (machined by Hilltop Technologies). In the case of each angled air objective, a custom-fabricated solid-immersion meniscus lens (SIMlens) (fabricated by BMV Optical Technologies) is sealed into the chamber (Supplementary Fig. 6) [26]. Stage-scanning is achieved by using an XY stage (MS2000, Applied Scientific Instrumentation) attached to two Z-axis risers (LS50, Applied Scientific Instrumentation). 405, 488, 561, and 638 nm excitation laser light is provided to the illumination path by a multi-line laser package (Cobolt Skyra, HÜBNER Photonics). The NODO collection path is equipped with a motorized filter wheel (FW102C, Thorlabs) with four filters that can be used with 405-nm (FF02-447/60 − 25, Semrock), 488-nm (FF03-525/50 − 25, Semrock), 561-nm (FF01-618/50 − 25, Semrock), or 638-nm (BLP01-647R-25, Semrock) excitation. The ODO collection path is equipped with a single multi-band-pass filter (FF01-432/515/595/730 − 25, Semrock) that can be used with any of the four excitation wavelengths. The entire system is compact and fits on a portable 2’ x 3’ optical cart (POC001, Thorlabs).
Illumination optical path
The illumination optical arm is shown in Supplementary Fig. 9. The illumination optics are designed to allow for the light-sheet properties (i.e., width, thickness, and depth of focus) to be adjusted. In addition, the optical path is designed to be compatible with the refractive index, n, of all current clearing protocols (i.e., it has multi-immersion capabilities) and to minimize chromatic aberrations and defocusing (i.e., variations in the illumination focal length as a function of wavelength).
Illumination light is fiber coupled into the system with a Gaussian numerical aperture (NA) of ~ 0.12 and collimated using an objective (RMS20X, Olympus). The beam diameter is then adjusted using a 4X variable beam expander (BE052-A, Thorlabs). This serves to adjust the overall NA of the light sheet. The variably expanded beam is then passed through an electronically tunable lens (ETL) that enables axial adjustment/alignment of the light sheet (EL-16-40-TC-VIS-5D-C, Optotune). The axially adjusted beam is relayed 1:1 using a pair of lenses (AC254-75A, Thorlabs) so that it can be scanned using a pair of large-beam-diameter galvanometric scanning mirrors (GVS012, Thorlabs). One mirror is scanned to create a digitally scanned light sheet [48]. The other scanning mirror is used to align the light sheet with the focal plane of the NODO and ODO imaging paths. A pair of achromatic doublet lenses (AC508-75A and AC508-200A, Thorlabs) are then used to relay the scanned beam to the back focal plane of a 2X illumination objective with NA = 0.10 (TL2X-SAP, Thorlabs). Finally, the illumination light travels through the SIMlens (Supplementary Fig. 6) [26]. The SIMlens provides multi-immersion performance and prevents aberrations (spherical, off-axis, and chromatic) of the light sheet by minimizing refraction of the illumination rays as they transition from air into the immersion medium. In addition, since ray angles are preserved as they transition between air and the immersion medium, the SIMlens increases the NA of the illumination light sheet by a factor of n. When combined, this optical design yields a light sheet with tunable NA (0.025–0.10 x n) and tunable width (0–11 mm / n) limited by the field of view (FOV) of the illumination objective.
A digitally scanned light sheet was chosen over a cylindrical-lens approach (static light sheet), as this facilitates achieving a high level of tunability for multi-scale imaging. Moreover, large scanning mirrors were selected to fill the back focal plane of the final illumination objective and to avoid having to significantly magnify the beam after the scanning mirrors, which would reduce the lateral scanning range of the mirrors (i.e., constrain the maximum light-sheet width). In the current design, rotating the scanning mirror results in lateral scanning at a ratio of ~ 0.60 mm per deg / n. The maximum desired light sheet width (i.e., lateral scanning range) is 11 mm, corresponding to a scanning angle of ~ 18 deg, which is within the maximum scan range of the galvo scanner (20 deg).
NODO optical path
The physical layout, ZEMAX model, and objective options for this NA-maximized NODO imaging configuration are shown in Supplementary Fig. 10. The NODO optical path of our system uses a multi-immersion objective (#54-12-8, Special Optics) with a long 1-cm working distance. This objective is compatible with all clearing protocols (n = 1.33–1.56) and provides a NA of 0.483 (in air) that scales with the index of the immersion medium (e.g., NA ~ 0.75 at n = 1.56). The lens is oriented in the normal (vertical) direction with respect to the specimen holder/interface and is therefore non-orthogonal to the light sheet. To image this non-orthogonal light sheet, we use a remote-focus imaging strategy analogous to what is used for single-objective light-sheet systems, with the multi-immersion objective serving as the primary objective (O1) [28, 29].
To minimize aberrations in the remote focus relay, the overall magnification from the specimen to the remote focus (air) should be equal to the refractive index of the specimen, n. Given this requirement, the relay lenses and first remote objective (O2) must be carefully selected. For O2, a 20X objective is optimal, as a 10X or 40X objective would clip either the NA or FOV of our O1. Of the several companies that produce microscope objectives, Zeiss and Leica were avoided because chromatic aberrations are partially corrected in the tube lenses produced by these companies, which would complicate selection of the two relay tube lenses. Therefore, only objectives from Olympus and Nikon were considered, where 20X objectives from Olympus have a focal length of 9 mm, and 20X objectives from Nikon have a focal length of 10 mm. Factoring in the effective focal length of O1 (12.19 mm / n), the required relay lens magnification is ~ 1.219X for Nikon and ~ 1.354X for Olympus. Note that the magnification of the multi-immersion objective in our system scales inversely as a function of n. This allows us to use a fixed set of relay lenses and always satisfy the remote focusing magnification requirement. However, this would be problematic with alternative multi-immersion objectives, where the effective focal length does not vary with n, and therefore the magnification of the relay lenses would need to be adapted for each immersion medium.
Well-corrected tube lenses are available with a limited selection of focal lengths (100, 165, 180, and 200 mm). Although custom tube lens assemblies are possible [33], we found that off-the-shelf 200-mm (Nikon Tube Lens #58–520, Edmund Optics) and 165-mm (TTL165-A, Thorlabs) tube lenses provide a magnification of ~ 1.212X, which matches the requirement for Nikon. Therefore, we decided to select a 20X Nikon objective for O2. To avoid the need for a cover glass, we narrowed our selection to Nikon objectives designed for use without a cover glass. This yielded one option, the LU (now TU) Plan Fluor EPI 20X (NA = 0.45). We chose this objective for the O2 in the current system.
To remotely correct for spherical aberrations introduced by the index mismatch of the specimen holder, it would also be possible to use a Nikon objective with a correction collar. There are two options, the CFI S Plan Fluor ELWD 20XC (NA = 0.45) with a correction collar for a cover glass of t = 0–2 mm, and CFI S Plan Fluor LWD 20XC (NA = 0.70) with a correction collar for a cover glass of t = 0–1.8 mm. Both objectives could be explored in a future design (although spherical aberrations were not found to be an issue in the current design). In the case of the CFI S Plan Fluor LWD 20XC, the full 0.483 NA of our O1 objective could be transmitted to the remote focus, unlike the chosen LU Plan Fluor EPI 20X objective or CFI S Plan Fluor ELWD 20XC objective, which both clip things down to 0.45 NA.
The goal of O3 is to maximize light collection when tilted at the angle required to orthogonally image a remote version of the oblique light sheet within the specimen. As mentioned previously, in a single-objective light-sheet design, the light sheet angle would be limited by our chosen O1 to a maximum of 28.9 deg. This would require O3 to be tilted by at least 61.1 deg. At this extreme tilt angle, one way to prevent light loss is through a custom solid- or liquid-immersion objective [32, 33]. However, a benefit of our NODO design is that the crossing angle can be increased to 45 deg, which reduces the tilt of O3 to 45 deg. At this tilt angle, an objective with NA = 0.95 is able to capture light up to NA = 0.45 and provide NA-maximized imaging. Therefore, we opted to use a NA = 0.95 air objective for our O3. This use of a tertiary air objective makes alignment more straightforward and stable than in the case of a solid- or liquid-immersion objective [32, 33]. This is especially the case for liquid-immersion objectives, where there may be evaporation or leakage of a liquid medium over time.
In terms of selecting an optimal O3 for NA-maximized imaging, we only considered objectives from Olympus and Nikon for the same reasons as mentioned previously. Both companies offer two types of objectives with NA = 0.95. These include 40X life-science objectives with a correction collar for cover-glass thicknesses ranging from t = 0.11–0.23 mm, and 50X metrology objectives for imaging with no cover glass. Although it would be possible to permanently align or adhere a cover glass to the 40X objectives, we decided to select the CF IC EPI Plan Apo 50X objective from Nikon for our O3 for simplicity and ease-of-use. When combined with a 100-mm tube lens (TTL100-A, Thorlabs), our NODO imaging path provides a total magnification of 25 x n. This yields a sampling rate of ~ 2.71 (slightly better than Nyquist) when using a sCMOS camera with pixels spaced by 6.5 µm (pco.edge 4.1, PCO Tech). The corresponding FOV is ~ 0.53 mm / n (FOV = 0.40 − 0.34 mm when n = 1.33–1.56), which is not clipped by the 0.40-mm FOV of the 50X objective. In this configuration, the back aperture of the illumination objective is filled, yielding an illumination NA of ~ 0.10 x n. This corresponds to a confocal parameter of ~ 40–50 µm and a pixel height of ~ 256 pixels for each raw camera image. This also corresponds to an axial re-focusing range that is well within the range of operation for an idealized objective, as specified by Botcherby et al. [28]. The pixel width of each raw camera frame is the full 2048 pixels of the sCMOS camera, corresponding to a FOV of ~ 0.34–0.40 mm.
While the above O3 selection provides NA-maximized imaging, our O1 also provides a 1-mm FOV to enable FOV-maximized imaging (with slightly worse resolution). To achieve this, O3 can be changed to a 20X objective with a matched 1-mm FOV. To minimize light loss, 20X objectives with the highest NA were considered. For both Olympus and Nikon, the upper limit is NA = 0.75 (although the recent UPLXAPO20X objective from Olympus offers NA = 0.80). In a single-objective light-sheet configuration, the light sheet angle in the specimen would be limited by the NA of O1, requiring a larger remote tilt angle, which for this particular O3 would result in an effective NA ~ 0.12 x n. In contrast, with our NODO configuration this results in an effective NA ~ 0.26 x n. This reduced effective NA (in contrast to the NA-maximized case) enables close-to-Nyquist sampling of a 1-mm FOV with the same sCMOS chip as described earlier (with 2048 x 2048 pixels). To match parfocal lengths and thus enable a quick interchange of objectives, we selected the Nikon CFI Plan Apo Lambda 20X for our FOV-maximized O3 objective. When paired with the same 100-mm tube lens (TTL100-A, Thorlabs), this provides a total magnification of 10 x n, which yields a near-Nyquist sampling rate of 2.02. The effective FOV is ~ 1.33 mm / n (FOV = 1.00–0.85 mm when n = 1.33–1.56), which is not clipped by the 1.00 mm FOV of the 20X objective (O3). The physical layout, ZEMAX model, and objective options for this FOV-maximized NODO imaging configuration are shown in Supplementary Fig. 11. See Supplementary Note 2 for more discussion of NA- versus FOV-maximized imaging. As mentioned previously, an alternative design could use the CFI S Plan Fluor LWD 20XC and a customized O3 to make use of the full 0.483 NA of our O1 (see Supplementary Fig. 12 and Supplementary Note 5).
ODO optical path
The physical layouts and ZEMAX models of the ODO imaging configurations are shown in Supplementary Fig. 13–14. While the NODO imaging path can provide sub-micrometer resolution, the FOV of the multi-immersion primary objective (O1) is restricted to 1 mm, which is insufficient for fast mesoscopic imaging. This tradeoff between NA and FOV is standard across all currently available microscope objectives (e.g., no current clearing-compatible objectives can simultaneously offer sub-micrometer resolution over a mesoscopic FOV) (Supplementary Fig. 8) [49]. Therefore, in our hybrid system, we achieve low-resolution imaging with a second independent ODO imaging path.
The ODO collection path uses the same objective as the illumination optical path (TL2X-SAP, Thorlabs). The objective is similarly used in conjunction with a SIMlens (fabricated by BMV Optical), which provides multi-immersion performance and prevents axial chromatic aberrations in the ODO imaging path [26]. In addition, the SIMlens increases the NA of the collection path by a factor of n. This yields an effective NA = 0.10 x n.
The current set point of the ODO collection path lies between the NA- and FOV-maximized imaging extremes and uses a tube lens with a 200-mm focal length (TTL200-A, Thorlabs). This provides a magnification of 2 x n and an effective FOV of 6.6 mm / n, which undersamples the collection NA of the system when using a sCMOS camera with a pixel spacing of 6.5 µm (pco.edge 4.1, PCO Tech, 2048 x 2048 pixels). In this configuration, the back aperture of the illumination objective is underfilled to yield an illumination NA of ~ 0.04 x n. This results in a confocal parameter of ~ 250–300 µm, which corresponds to a digital image height of ~ 128 pixels for each raw camera image. The width (in pixels) of each raw camera image is the full 2048 pixels of the sCMOS camera, corresponding to a distance of ~ 4.25–5 mm in the sample. This combination of illumination and collection NA provides near-isotropic resolution (Supplementary Fig. 16Error! Reference source not found.).
The magnification of the ODO imaging path is easily adjusted by changing the tube lens. NA-maximized imaging is achieved using a tube lens with a 400-mm focal length (AC508-400-A, Thorlabs). This provides a magnification of 4 x n and an effective FOV of 3.3 mm / n, which corresponds to a near-Nyquist sampling rate of ~ 2.1 when using a sCMOS camera with a pixel spacing of 6.5 µm. FOV-maximized imaging is achieved using a tube lens with a 100-mm focal length (TTL100-A, Thorlabs). This provides a magnification of 1 x n and an effective FOV of 13.33 mm / n. See Supplementary Note 2 for more discussion of NA- versus FOV-maximized imaging modes.
Image acquisition and post processing
Image strips are collected with a combination of stage-scanning and lateral/vertical tiling. The stage-scanning firmware is used to send a TTL SYNC trigger signal from the XY stage to the sCMOS camera for reproducible initiation of each imaging strip. After triggering, the camera is set to free-running mode and acquires the desired number of frames for a given image-strip length (as the sample is scanned by the stage at a constant velocity), with a spacing between adjacent frames that is identical to the sampling pitch of the raw camera images. For each raw frame, the camera uses the standard rolling shutter, where the shutter rolls from the center of the camera chip in both directions, as opposed to the light-sheet readout mode, where the shutter rolls from the top to the bottom of the camera chip. Using the standard rolling shutter, the rolling directions are oriented along the light sheet propagation direction. This orientation allows the imaging speed to be increased when the pixel height of each raw frame is cropped to match the confocal parameter of the illumination light sheet.
At the start of each exposure, the camera sends a trigger to an analog output DAQ card (PCIe-6738, National Instruments). The DAQ card then sends output voltages to the lasers, galvos, and ETL (for alignment only, not axial sweeping). To reduce motion blur, the lasers and galvos are triggered with a delay to only illuminate once the shutter has rolled across the full pixel height of the camera frame, resulting in a strobing effect. For the 128- or 256-pixel height of the ODO and NODO paths respectively, the rolling time, troll, is 0.625 and 1.25 ms. The exposure time, texp, is set to 3 x troll, resulting in a total exposure time, ttot, of 4 x troll, or 2.5 and 5 ms for the ODO and NODO paths. This corresponds to a data rate that is 1/4 the maximum data rate of the sCMOS camera. The lateral scanning mirror is actuated with a sawtooth waveform and completes a single period within the total exposure time, ttot, of the raw camera frame, corresponding to a frequency of ~ 400 and ~ 200 Hz for the ODO and NODO paths. The ETL and second mirror are set to a pre-calibrated DC voltage for the entire image strip to yield an in-focus light sheet that is axially aligned to the center of the camera chip.
Raw camera frames are streamed from the camera to RAM and subsequently saved directly as a single HDF5 file and XML file with the associated metadata, which is suitable for immediate processing with BigStitcher via ImageJ [50, 51]. This involves on-the-fly saving of down-sampled copies of the image strip (2x, 4x, 8x, and 16x) in the hierarchical HDF5 file format. Additionally, the GPU-based B3D compression filter can be optionally added to the HDF5 writing process to yield 5-10X compression with negligible loss of usable information content [52]. However, both processing steps slow down the net data rate from RAM to disk. On-the-fly saving of down-sampled copies of the image strips slows the data rate by ~ 2x, and the B3D algorithm slows down the process by another ~ 2x. This net speed reduction of ~ 4x motivated our selection of the camera frame rate mentioned previously. For our experiments, this reduction in post-processing times and data-storage requirements are worth the reduction in imaging speed. However, it is important to note that the imaging speeds could be increased by 2x if the aforementioned processing steps are omitted from the acquisition procedure, in which case the imaging speed would become limited by the duty cycle of the illumination strobing, where texp = troll. The imaging speeds quoted in Fig. 1g assume the 4x-reduced data rate.
A ~ 15% overlap is used for both vertical and lateral tiling. Different wavelength channels are acquired sequentially. For each image strip “tile” that is acquired by laterally scanning the specimen in the y-direction, all channels are acquired by cycling through various laser/filter combinations and re-scanning that image strip for each laser/filter setting, before moving to the next tile position. When tiling vertically, the laser power is increased with depth per a user-defined exponential relationship, P = P0 × exp(z/µ), to account for the attenuation of the illumination light sheet as it penetrates deeper into the specimen (typically µ = 5–20 mm− 1). Finally, if desired, all of the imaging tiles can be aligned and fused into one contiguous 3D image as an HDF5 or TIFF file output using BigStitcher [50]. The entire image acquisition is controlled using a custom-written Python program that is available from the authors upon request.
Computer hardware
During acquisition, the images are collected by a dedicated custom workstation (Puget Systems) equipped with a high-specification motherboard (Asus WS C422 SAGE/10G), processor (Intel Xeon W-2145 3.7GHz 8 Core 11MB 140W), and 256 GB of RAM. The motherboard houses several PCIe cards, including 2 CameraLink frame grabbers (mEIV AD4/VD4, Silicon Software) for streaming images from the camera, a DAQ card (PCIe-6738, National Instruments) for generating analog output voltages, a 10G SFP + network card (StarTech), and a GPU (TitanXP, NVIDIA). Datasets are streamed to a local 8 TB U.2 drive (Micron) that is capable of outpacing the data rates of the microscope system. Data is then transferred to a mapped network drive located on an in-lab server (X11-DPG-QT, SuperMicro) running 64-bit Windows Server, equipped with 768 GB RAM and TitanXP (NVIDIA) and Quadro P6000 (NVIDIA) GPUs. The mapped network drive is a direct-attached RAID6 storage array with 15 × 8.0 TB HDDs. The RAID array is hardware based and controlled by an external 8-port controller (LSI MegaRaid 9380–8e 1 GB cache). Both the server and acquisition workstation are set with jumbo frames (Ethernet frame), and parallel send/receive processes matched to the number of computing cores on the workstation (8 physical cores) and server (16 physical cores), which reliably enables ~ 1.0 GB sec− 1 network-transfer speeds.
Preparation of ECi-cleared mouse brain
Labeling and clearing was carried out as previously described [43]. For sparse labeling, a Slc17a7-Cre mouse (8 weeks old, female) received a systemic injection, via the retro-orbital sinus, of a mixture of Cre-dependent Tet transactivator (PHP-eB-Syn-Flex-TRE-2x-tTA) and a reporter virus (PHP-eB-CAG-TRE-3xGFP) [43, 53]. High-titer (> 1012 GC/mL) viruses were obtained from the Janelia Research Campus Molecular Biology Core and diluted in sterile water when necessary.
Transfected mice were anesthetized with an overdose of isoflurane and then transcardially perfused with a solution of PBS containing 20 µg/mL heparin (Sigma-Aldrich #H3393) followed by a 4% paraformaldehyde solution in PBS. Brains were extracted and post-fixed in 4% paraformaldehyde at 4℃ overnight (12–14 hrs) and washed in PBS to remove all traces of excess fixative (PBS changes were performed at 1 hr, 6 hr, 12 hr, and 1 day).
For amplification by immuno-labeling, brains were delipidated with a modified Adipo-Clear protocol [54]. Brains were washed with a methanol gradient series (20%, 40%, 60%, 80%, Fisher #A412SK) in B1n buffer (H2O/0.1% Triton X-100/0.3 M glycine, pH 7; 4 mL / brain; one hr / step). Brains were then immersed in 100% methanol for 1 hr, 100% dichloromethane (Sigma #270997) for 1.5 hrs, and three times in 100% methanol for 1 hr. Samples were then treated with a reverse methanol gradient series (80%, 60%, 40%, 20%) in B1n buffer for 30 min each. All procedures were performed on ice. Samples were washed in B1n buffer for one hr and left overnight at room temperature; and then again washed in PTxwH buffer (PBS/0.1% Triton X-100/0.05% Tween 20/2 µg/mL heparin) with fresh solution after one and two hrs and then left overnight.
After delipidation, selected samples were incubated in primary antibody dilutions in PTxwH for 14 days on a shaker (1:1000, anti-GFP, Abcam, #ab290). Samples were sequentially washed in 25 mL PTxwH for 1, 2, 4, 8, and three times for 24 hrs. Samples were incubated in secondary antibody dilutions in PTxwH for 14 days (1:600, AlexaFluor® 488 conjugated donkey-anti-rabbit IgG) and washed in PTxwH similar to descriptions above. Finally, the tissue was dehydrated in ethanol grades (25, 50, 75, 100%) for 8 hr per grade. The 100% grade was repeated to ensure removal of all water from the tissue. Finally, the tissue was cleared in ethyl-cinnamate (Sigma-Aldrich #112372) for 8 hr before imaging. All experimental protocols were conducted according to the National Institutes of Health guidelines for animal research and were approved by the Institutional Animal Care and Use Committee at Howard Hughes Medical Institute, Janelia Research Campus.
Preparation of MDA-231/OS-RC-2 and αSMA-labeled mouse brains
Human breast cancer cells, MDA-MB-231-5a-D (MDA-231) are highly metastatic clones from MDA-MB-231 [55]. Human renal cell carcinoma, OS-RC-2 were kindly provided by Prof. Tatsuro Irimura (Juntendo University, Japan). MDA-231 cells were cultured in Dulbecco’s modified eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin and 100 µg/ml streptomycin as previously described [55]. OS-RC-2 cells were maintained in RPMI1640 containing 10% FBS and penicillin/streptomycin [56].
To establish cancer cells stably expressing firefly luciferase and mCherry under the EF-1 promoter, a lentiviral expression system was used (kindly provided by Dr Hiroyuki Miyoshi, deceased, formerly Keio University, Tokyo, Japan) as described previously (Miyoshi et al., Proc. Natl. Acad. Sci., 1997). Briefly, 293FT cells were transfected with a vector construct encoding the expression protein, VSV-G, a Rev-expressing construct (pCMV-VSV-G-RSV-Rev), and a packaging construct (pCAG-HIVgp). The culture supernatants containing viral particles were collected and used as lentiviral vectors.
BALB/c-nu/nu mice (4 weeks old, female) were purchased from Japan SLC (Shizuoka, Japan). All experiments were approved and carried out according to the Animal Care and the Use Committee of the Graduate School of Medicine, The University of Tokyo. For developing experimental brain metastasis models by intracardiac (i.c.) inoculation, BALB/c-nu/nu mice were injected with MDA-231 or OS-RC-2 cells (5 × 105 cells /mouse) by puncture into the left ventricle of the heart.
Clearing, 3D staining, and imaging of whole mouse brain samples was performed with CUBIC clearing and CUBIC-HistoVision (CUBIC-HV) staining protocols [7, 10]. An updated CUBIC-HV staining protocol was used (HV1.1, commercialized by CUBICStars Co. and Tokyo Chemical Industry (TCI), TCI #C3709, #C3708). In brief, the PFA-fixed whole mouse brains were treated with CUBIC-L for 4 days at 37°C, washed with PBS, stained with SYTOX-G (1/2500) in CUBIC-HV nuclear-staining buffer (included in TCI #C3709) for 5 days at 37°C, and then washed / immersed in 50 % and 100% CUBIC-R + for 1 day and 3 days, respectively, at room temperature. We used CUBIC-R+(M) [45 wt% of antipyrine (TCI #D1876), 30 wt% of N-methylnicotinamide (TCI #M0374), and 0.5% (v/w) N-butyldiethanolamine (TCI #B0725), adjusted to pH ~ 10] for brains containing cancer cells, and CUBIC-R+(N) [45 wt% of antipyrine, 30 wt% of nicotinamide (TCI #N0078), and 0.5% (v/w) N-butyldiethanolamine, adjusted to pH ~ 10] for the αSMA immunostained brain, respectively.
For the whole mouse brain immunostained with anti-α-SMA (Sigma, #A5228) antibodies, the brain was subjected to CUBIC-HV immunostaining. The brain was first treated with 3 mg/mL hyaluronidase in CAPSO buffer (pH 10) for 24 h at 37°C. After washing with hyaluronidase wash buffer [50 mM Carbonate buffer, 0.1% (v/v) Triton X-100, 5% (v/v) Methanol, and 0.05% NaN3] and HEPES-TSC buffer [10 mM HEPES buffer, pH 7.5, 10% (v/v) Triton X-100, 200 mM NaCl, 0.5% (w/v) casein, and 0.05% NaN3], the brain was immersed in 500 µL of HEPES-TSC buffer containing a primary antibody (6 µg for anti-α-SMA), a secondary Fab fragment (FabuLight, Jackson immunolab, Alexa Fluor® 594 Goat Anti-Mouse IgG1 #115-587-185, Alexa Fluor® 594 Goat Anti-Rabbit IgG #111-587-008, Alexa Fluor® 594 Goat Anti-Mouse IgG2a #115-587-186, 1:0.75 of weight ratio), and 3D immunostaining additive (1x) (included in TCI #C3708). Then, the sample was incubated with gentle shaking for 10 days at 32°C. After staining, the sample was additionally incubated in the same buffer for 1 day at 4°C to stabilize the Fab binding. Then, the sample was washed and post-fixed according to the protocol of CUBIC-HV immunostaining kit (TCI #C3708) before being index-matched with CUBIC-R+. The cleared sample was embedded in CUBIC-R-agarose for imaging and storage [57]. This animal experimental procedures and housing conditions of the animals were approved by the Animal Care and Use Committees of the Graduate School of Medicine of the University of Tokyo.
Statistics and reproducibility
For statistical analysis, we reported the mean, standard deviation, and number of observations. For repeatability, only one sample was used for each imaging experiment, unless otherwise noted.