Advanced CUBIC protocols for whole-brain and whole

PROTOCOL
Advanced CUBIC protocols for whole-brain and
whole-body clearing and imaging
Etsuo A Susaki1–3,7, Kazuki Tainaka1–3,7, Dimitri Perrin4,7, Hiroko Yukinaga3, Akihiro Kuno1,2,5,6 &
Hiroki R Ueda1–3
1Department of Systems Pharmacology, The University of Tokyo, Tokyo, Japan. 2Japan Agency for Medical Research and Development (AMED) – Core Research for
Evolutionary Science and Technology (CREST), AMED, Tokyo, Japan. 3Laboratory for Synthetic Biology, RIKEN Quantitative Biology Center (QBiC), Osaka, Japan.
4School of Electrical Engineering and Computer Science, Science and Engineering Faculty, Queensland University of Technology (QUT), Brisbane, Queensland, Australia.
5Department of Anatomy and Embryology, Faculty of Medicine, University of Tsukuba, Ibaraki, Japan. 6PhD Program in Human Biology, School of Integrative and Global
Majors, University of Tsukuba, Ibaraki, Japan. 7These authors contributed equally to this work. Correspondence should be addressed to H.R.U. (uedah-tky@umin.ac.jp).
© 2015 Nature America, Inc. All rights reserved.
Published online 8 October 2015; doi:10.1038/nprot.2015.085
Here we describe a protocol for advanced CUBIC (Clear, Unobstructed Brain/Body Imaging Cocktails and Computational analysis).
The CUBIC protocol enables simple and efficient organ clearing, rapid imaging by light-sheet microscopy and quantitative imaging
analysis of multiple samples. The organ or body is cleared by immersion for 1–14 d, with the exact time required dependent
on the sample type and the experimental purposes. A single imaging set can be completed in 30–60 min. Image processing and
analysis can take <1 d, but it is dependent on the number of samples in the data set. The CUBIC clearing protocol can process
multiple samples simultaneously. We previously used CUBIC to image whole-brain neural activities at single-cell resolution
using Arc-dVenus transgenic (Tg) mice. CUBIC informatics calculated the Venus signal subtraction, comparing different brains at a
whole-organ scale. These protocols provide a platform for organism-level systems biology by comprehensively detecting cells in a
whole organ or body.
INTRODUCTION
Since the discovery of the cell as the basic unit of living organisms,
people have been seeking a way to observe all cells inside the body.
Comprehensive analysis of cells in organs and whole organisms
is expected to provide information about type, position, number
and activity of cells and cellular networks. Tissue clearing followed by 3D imaging is one approach that enables the analysis of
multiple cells simultaneously in organs. Thus, the development of
this and related technologies has become a recent trend1,2.
Development of tissue-clearing methods
Early tissue-clearing methods used organic chemicals (e.g., benzyl
alcohol–methyl salicylate, benzyl alcohol–benzyl benzoate
(BABB), and solvents used in 3D imaging of solvent-cleared
organs (3DISCO)) for this purpose3–6. Some of these methods
achieved high transparency within a few days by removing lipids
and homogenizing refractive indices (RIs) of the tissue, and they
were shown to be compatible with whole-mount immunohistochemical analysis7. However, concerns about the quenching
of fluorescent proteins and safety issues led to further method
development. A recent publication addressed this issue, reporting
that pH control and temperature during clearing are crucial to
stabilizing fluorescent proteins8. Alternative techniques, such as
Scale9, use a hydrophilic chemical urea, and more recently developed tissue-clearing methods use other hydrophilic reagents,
including SeeDB10, ClearT (ref. 11), or 2,2`-thiodiethanol12,13
and FRUIT14. These methods are easy and safe, and they allow
fluorescent signals to be retained; however, they have a relatively
low clearing capability. The introduction of CLARITY enabled
both fluorescence retention and high transparency by embedding
a tissue into hydrogel polymer and removing most of the lipids
by electrophoresis15. Possible drawbacks of CLARITY included
its technical difficulty and limited scalability because of the need
to use a specific device. However, these difficulties have been
addressed by the development of passive clearing protocols that
increased the scalability16,17.
In this protocol, we describe how to perform CUBIC. CUBIC
offers a high-performance and device-free tissue clearing
method based on hydrophilic reagents, which preserves
fluorescence. It enables reproducible whole-organ and wholebody clearing. We have used CUBIC for clearing and rapid 3D
imaging of whole mouse brains, a whole marmoset hemisphere,
whole mouse organs (e.g., lung and heart) and whole mouse
body. These images were used for image analyses for extracting
biological information18,19.
Methods for imaging cleared tissues
Tissues cleared using the above methods can be imaged in 3D
with optical microscopies. Because some of the above clearing
methods render tissues highly transparent, light-sheet fluorescence microscopy (LSFM) has also been used for imaging1,2.
This type of microscopy can collect z-stack images in a rapid
manner, and it has been applied to 3D and 4D imaging, such
as a time-lapse imaging of developing embryos or whole-brain
calcium dynamics20–23. One of the earliest cases of whole mouse
brain imaging was rapid whole-brain imaging of a BABB-cleared
brain using a macrozoom-compatible light-sheet unit (ultramicroscopy)4. More recently, COLM (CLARITY-optimized lightsheet microscopy) has been used for whole-brain-scale imaging
of CLARITY-processed brains16. Thus, the use of LSFM after an
efficient tissue-clearing method facilitates a high-throughput
collection of multiple 3D images.
Rapid 3D imaging with LSFM can be used after whole-organ
and whole-body clearing by CUBIC. CUBIC also provides
processing and analysis of 3D images for extracting biological
NATURE PROTOCOLS | VOL.10 NO.11 | 2015 | 1709
PROTOCOL
© 2015 Nature America, Inc. All rights reserved.
information. Therefore, CUBIC presents a platform of wholeorgan or whole-body imaging and image informatics, which
enables a wide range of users to perform experiments targeting
cellular and organ layers with multiple samples.
Development of CUBIC for efficient and reproducible
whole-organ or whole-body clearing
In developing a clearing technique for whole-brain and wholebody imaging, we considered two main criteria: one, efficiency
and transparency with the preservation of fluorescence for a rapid
whole-brain/body imaging with LSFM, and, two, reproducibility
for comparative analysis of multiple samples . Because a clearing
method with hydrophilic reagents had the potential to fulfill these
criteria, we started by modifying the Scale recipe9. For this purpose, we constructed a new chemical screening method in which
reduction of turbidity of a fixed-brain suspension was measured
before and after mixing with a candidate chemical solution 18.
This screening enables nonbiased discovery of brain-clearing
chemicals. We screened 40 Scale-related chemicals and found that
aminoalcohols, in addition to urea and Triton X-100 in Scale,
clear tissue with minimal fluorescence quenching18. In the CUBIC
clearing protocol, we prepared two reagents, ScaleCUBIC-1
(reagent-1) and ScaleCUBIC-2 (reagent-2), which also minimize light scattering inside the tissue. The first reagent works as
a potential lipid remover. Lipid is thought to be the main lightscattering material inside tissue, and its removal is correlated with
the degree of transparency. A fixed brain was treated with the
first reagent for ~1 week, washed with buffer and then immersed
into the second reagent, which has an RI close to ~1.49, which is
similar to that of the SeeDB reagent10. Moving from the buffer to
the second reagent matched the RIs between the sample and the
reagent, which further-reduced light scattering within the tissue.
Thus, a whole mouse brain became transparent within ~14 d
(ref. 18). However, some chemicals seem to have additional or
different roles during the procedure, and thus further studies are
needed to elucidate tissue-clearing mechanisms.
In addition to light scattering, light absorption is another challenge in tissue clearing. We accidentally found that aminoalcohols
can remove heme in blood and tissues19; thus, CUBIC is able
to decolorize tissue. We used perfusion of the CUBIC reagent
(CB-perfusion) to efficiently penetrate the mouse body and
to accelerate the clearing and decoloring procedure. The CBperfusion protocol enabled not only faster clearing of dissected
tissues but also whole-body clearing of infant and adult mice19.
Because of its efficiency and reproducibility, the CUBIC protocol can be applied to multiple samples in a single experiment,
and it is scalable from subcellular structures (e.g., neuronal
axons or spines) to marmoset brains to whole animal bodies 18,19.
Furthermore, whole-organ counterstaining with a nucleic dye
enables precise positioning of genetically labeled cells in the
whole-organ structure, extraction of specific anatomical structures and alignment of different samples for comparing signal
intensities18,19.
CUBIC for whole-organ or whole-body imaging and image
informatics
CUBIC-cleared samples can be used in LSFM. We use an
optimized Ultramicroscope (LaVision BioTec) for this purpose.
In rapid whole-organ imaging, a single, cleared, whole mouse
1710 | VOL.10 NO.11 | 2015 | NATURE PROTOCOLS
brain can be imaged within 30–60 min per color and orientation.
Fluorescence wavelength affects the quality of imaging results;
in general, red wavelengths can penetrate deeper in tissue, and
thus better imaging results are obtained with red fluorescence
than green fluorescence, particularly in deeper regions. To ensure
that weaker signals are detected in deeper regions, the sample is
imaged in two orientations; in the case of whole-brain imaging,
we took z-stack images of the dorsal-side-up (D-V) and ventralside-up (V-D) directions.
Image visualization software such as Imaris can be used for
depicting the reconstituted 3D image. Imaris implements numerous image analysis functions including spot counting and surface
extraction. We performed extractions of anatomical structures
in the 3D images for comparison of Langerhans islets in normal
and diabetic pancreases19. For more complicated analyses, we
implemented image informatics often used in functional MRI
(fMRI) analysis18. First, structural images via counterstaining
were registered to a reference brain to calculate transformation
parameters. Next, the transformation parameters were applied to
the corresponding signal images (transgenes, etc.) for alignment.
These aligned images could be merged with each other to calculate
signal subtraction between samples. This analysis was performed
with open-source software such as Advanced Normalization Tools
(ANTS)24 and ITK-SNAP25, but it requires advanced informatics and computer science skills. For the user’s convenience, we
provide an easier analysis pipeline with prepared scripts in
this manuscript. As an example of the comparative analysis,
we demonstrate 3D image analysis of Arc-dVenus Tg mouse brains
with or without light stimulation and calculate the signal subtraction18,26. The final subtraction data clearly depict regions and
cells in the whole brain in which neurons responded to the light
stimuli. Such direct comparative analysis by using whole-brain
fluorescent 3D images was first reported using CUBIC informatics18. CUBIC informatics enables quantitative identification of
stimulus- or timing-dependent neural activities, and it will help
delineate structural abnormalities in disease samples at the wholeorgan and whole-body scale.
Overview of the CUBIC pipeline
CUBIC provides a platform for a comprehensive analysis of cells
in a whole organ or body. Here we focus on describing the following: (i) the advanced CUBIC clearing protocols by simple
immersion and CB-perfusion (Steps 1–2); (ii) whole-brain and
whole-organ imaging with a LSFM (Steps 3–6); and (iii) CUBIC
informatics for preprocessing and comparison of different brain
samples (Steps 7–15, see also Fig. 1). Although CUBIC is also
applicable to staining with small chemicals or antibodies over days
to weeks, as described previously18,19, we focus here on imaging of
fluorescent proteins together with nuclear counterstaining.
Tissue clearing. Here we provide three clearing procedures:
Step 2A, a simple immersion protocol for dissected whole organs;
Step 2B, the CB-perfusion and immersion protocol for faster
clearing of whole organs; and Step 2C, the CB-perfusion protocol for whole-body clearing. The immersion protocol in our
first CUBIC report18 has been improved to an advanced version
(Fig. 2) in which the clearing speed and efficiency are increased.
The CB-perfusion protocol (Fig. 3) is almost identical to that
in our second CUBIC report19, but more detail is given here.
PROTOCOL
Two-photon imaging, etc.
Clearing by
simple immersion
Day 0
Day 1
Days 2–8
Fixation
Lipid removal
RI matching
Step 1–2A(ii)
Step 2A(iii–vi)
Step 2A(ii,iii)
Step 2B(v–vii)
Step 2B(viii,ix)
Clearing by
CB-perfusion
Imaging
with LSFM
Steps 3–6
5–10 min
1d
Perfused with PBS
and 4% PFA
Postfixation
in 4% PFA
1–7 d
Dissected
organs
Steps 1–2C(i)
Day 8
7d
Whole
body
Day 15~
1d
7d
Reagent-1
refresh/1 d
Preprocessing
Immersion
oil
Reagent-1
refresh/2–3 d
Signal subtraction
Steps 9 and 10
Acquision
Steps 11 and 12
D-V images
Use
dorsal
side
Create
composite
NIfTI-1 data
(viewed in ITK-SNAP)
Immersion
oil
Lipid removal/decolorization
Step 2C(ii,iii)
30 min~
Steps 7 and 8
Raw TIFF
~10 min
Reagent-2
refresh/1 d
Wash
Day 1
Fixation / CB-perfusion
Steps 1–2B(iv)
2d
~1 d
Reagent-1
refresh/2–3 d
Perfused with PBS, 4% PFA
and 1/2-diluted reagent-1
Image analysis
~Day 11
1–3 h per sample
Day 0
© 2015 Nature America, Inc. All rights reserved.
~Day 9
Acquision
V-D images
Steps 13–15
Sample A
Sample B
Reference
Structure
Subtracted
Use
ventral
side
Signal
~20 min
~3.5 h
~3 h
0.5–1.5 h
Convert to NIfTI-1/
downscaling
Create D-V + V-D
composite
Alignment
Normalization and subtraction
between samples
Figure 1 | Overview of the CUBIC pipeline. CUBIC is composed of three major stages (clearing, imaging and analysis). For efficient and reproducible clearing,
we provide three protocols: a simple immersion protocol (Step 2A), which takes ~11 d for a whole brain from an adult mouse (but varies according to the
experimental purpose and organs); a CB-perfusion protocol for the whole adult mouse (Step 2C), which takes ~14 d; and the CB-perfusion and immersion
hybrid protocol (Step 2B), in which dissected organs after CB-perfusion can be continuously cleared according to the simple immersion protocol. Rapid 3D
imaging can be performed with an LSFM. The collected data are processed and analyzed, such as by signal comparison between samples, as described in this
manuscript. Images of actual samples correspond to the samples in Figures 2 and 3. All animal experiments here were approved by the Animal Care and Use
Committee of the RIKEN Kobe Institute and The University of Tokyo, and all animals were cared for in accordance with institutional guidelines.
The clearing performance of CB-perfusion surpasses the immersion protocol, particularly in heme-rich organs (heart, muscle,
kidney or liver)19, but it tends to cause decreased signal intensity
because of a short fixation time. Incubation period can be varied,
and it is dependent on the organ and imaging methods to be used
(Figs. 1–3). Users may select either of these options and determine
the desired final transparency for their experimental purposes.
In either option, a paraformaldehyde (PFA) fixation is needed.
Thus, animals are transcardially perfused with 4% (wt/vol)
PFA, and then organs are dissected for postfixation (Step 1).
Alternatively, the fixed body can be further perfused with diluted
reagent-1. Samples are subsequently treated with reagent-1 and
reagent-2. Counterstaining is performed during and after reagent-1
treatment18,19. Samples can be stored at various points in the
PROCEDURE, which are indicated as PAUSE POINTS.
Whole-organ or whole-body imaging. We use a commercially
available LSFM instrument (Ultramicroscope, LaVision BioTec)
supplied with an optimized macrozoom microscope (MVX-ZB10,
Olympus) and a scientific complementary metal oxide semiconductor (sCMOS) camera (Andor NEO 5.5, 2,560 × 2,160 pixels).
A customized sample holder is also used for larger brain and body
samples and soft abdominal organs (Fig. 4). A suitable pair of
excitation laser and emission filters is also installed. We typically
use 100 mW of 488-nm laser-ET525/50 emission filter for green
fluorescence, and 50 mW of 588-nm laser-ET650/60 emission
NATURE PROTOCOLS | VOL.10 NO.11 | 2015 | 1711
PROTOCOL
a
b
c
d
Pancreas
Spleen
Kidney
© 2015 Nature America, Inc. All rights reserved.
Complete
dissolution
Before
dissolution
structural images via whole-organ counterstaining should be collected. To ensure
sharpness of signals throughout the 3D
image, data of the same sample from different directions (D-V and V-D, in the case
of whole-brain imaging) are also collected.
This is recommended because the ventral
horizontal slices are sharper in V-D images
and dorsal horizontal slices are sharper in
PBS washed
1/2 reagent-2
1/2 reagent-1
Reagent-1
Reagent-2
e Fixed brain
D-V images (Fig. 5)18.
(day 8 and 9)
(day 9, 6.5 h)
(day 1)
(day 1, 6 h)
(day 7)
(day 11)
If a proper LSFM instrument is not
available, widely used confocal or twophoton microscopes can also be used.
Partially cleared samples by one-step
immersion in reagent-1 for 1–3 d are even
applicable to deep region imaging with
two-photon microscopy18, because the
Figure 2 | Procedure of the simple immersion protocol. (a) Preparation of reagent-2. This reagent
clearing performance surpasses that of some
contains a high concentration of urea (25 wt%) and sucrose (50 wt%; top). These can be completely
of the other clearing methods developed
dissolved by heat and stirring with a microwave and a hot stirrer (bottom). (b) A vacuum desiccator
for the degassing steps. (c) An incubator with a shaker (a hybridization incubator) that we use for the
for these imaging purposes19. Microscope
clearing procedure. Inset: five brain samples treated with reagent-1 in a single tube (day 5). (d) A table
vendors have released objective lenses
shaker used for the PBS washing step at room temperature. (e) Appearance of a brain sample in each step.
for deep tissue imaging (e.g., Olympus
A brain from C57BL/6 male mouse (8 weeks old) was used. Reagent-1–treated sample is temporally
XLPLN10XSVMP, 10×/0.6, working
swollen, but the size is recovered after immersion in reagent-2. Scale bar, 5 mm. All animal experiments
distance (WD) = 8 mm, ne (adjustable
were approved by the Animal Care and Use Committee of the RIKEN Kobe Institute and The University of
refractive index range) = 1.33–1.52 and
Tokyo, and all of the animals were cared for in accordance with institutional guidelines.
XLSLPLN25XGMP, 25×/1.0, WD = 8 mm,
ne = 1.41–1.52; Zeiss LD Plan-Aphochromat
filter for red to far-red fluorescence. The size of the acquired image 20×1.0 WD = 5.6 mm, ne = 1.43–1.47; Leica HC FLUOTAR
per pixel is dependent on the zoom of the microscope, such that L25×/1.00, WD = 6 mm, ne = 1.457).
one pixel is ~5.2 × 5.2 Mm at 2× zoom and one pixel is ~6.5 ×
a
b PFA perfusion CB-perfusion
6.5 Mm at 1.6× zoom. These pixel sizes are sufficient for detectPeristaltic pump
ing signal from single cells in regions such as the cerebral
cortex, or even from subcellular structures when they were
4% PFA
Butterfly
sparsely labeled, as discussed below (Fig. 4b–f and Supplementary
needle
Video 1). Users can also use higher magnified zoom values (~6.3×)
if finer resolution is needed. Z-step size is selected according to
PBS/heparin
the thickness of the laser sheet. We typically select the thinnest
1/2 reagent-1
PBS
sheet of the used LSFM and set 10 Mm as the z-step size. For
further image analysis, both signal images (e.g., transgenes) and
c Day 1 in reagent-1
Figure 3 | Procedure of the CB-perfusion protocol. (a) Surgical setup for
CB-perfusion. The transcardiac perfusion line is connected to heparin-PBS
for flushing the blood, 4% (wt/vol) PFA on ice with peristaltic pump for
fixation, PBS for flushing PFA and 1/2-diluted reagent-1 for accelerative
clearing through the vascular system. (b) Dissected organs just after
CB-perfusion. Organs such as the pancreas, spleen or kidney are macroscopically cleared and decolored at this point. The clearance of these organs
is indicative of how well researchers succeed at CB-perfusion. Scale bars,
5 mm. (c) Clearing performance of CB-perfused dissected organs at day 1 in
reagent-1 and at day 10 in reagent-2. CB-perfused organs from a successful
procedure are markedly transparent after 1 d of reagent-1 treatment.
Some of the organs such as pancreas are more transparent in reagent-1
rather than in reagent-2. Scale bars, 5 mm. (d) Clearing performance of a
CB-perfused whole body just after CB-perfusion and after 2 weeks of reagent-1
treatment. Organs such as pancreas, submaxillary gland and spleen quickly
tend to become transparent by CB-perfusion. Major abdominal organs except
bones and gastrointestinal content become sufficiently transparent after
2 weeks of reagent-1 treatment. C57BL/6 male mice (8 weeks old) were used
in b–d. All animal experiments were approved by the Animal Care and Use
Committee of the RIKEN Kobe Institute and The University of Tokyo, and all
of the animals were cared for in accordance with institutional guidelines.
1712 | VOL.10 NO.11 | 2015 | NATURE PROTOCOLS
Day 10 in reagent-2
Heart
d
Lung
Liver
Kidney
Pancreas
Spleen
After CB-perfusion
After 2 weeks of reagent-1 treatment
Head
Dorsum
Chest
Pelvis
a
b
Raw TIFF image (z = 3.5 mm)
Holder
Sample
Cover glass
10 pixels
Illumination
c
d
e
3D
f
3D
x-z
3D
d
e
f
1 mm
SYTO 16
mKate2
Merged
Heart
3D
2 mm
1 mm
2 mm
3D
Pancreas
g
500 µm
500 µm
200 µm
SYTO 16
mKate2
Merged
1 mm
2 mm
Lung
Spleen
1 mm
1 mm
2 mm
Stomach
2 mm
Kidney
2 mm 1 mm
1 mm
2 mm
Intestine
Figure 4 | Whole-organ imaging with LSFM.
(a) The microscope setup and the customized
sample holder (inset) used in this manuscript.
(b) Left, a raw TIFF image (2,560 × 2,160) from
a cleared Thy1-YFP-H Tg mouse brain34 (male,
23 weeks old) (imaging conditions: z = 10 Mm
step × 749 planes, zoom = 1.6×, expose = 50 ms ×
two illuminations from each side, total acquisition
time = ~30 min). The sample was cleared according
to the immersion protocol in this manuscript.
Right, a magnified image of the indicated area on
the left. These images were minimally processed
(sharpness, brightness and contrast) with ImageJ.
(c) The reconstituted 3D image of the acquired
data in b. A view from the dorsal side is shown.
(d) A magnified image of the indicated area in c.
(e) A magnified image of the right hippocampus,
viewed from the midline to lateral, as indicated in c.
(f) A magnified image of the reconstituted x-z
image of the right hippocampus, as indicated in c.
(g) The reconstituted 3D whole-organ images from
B-actin-nuc-3×mKate2 KI mouse19 (male, 8 weeks
old) (imaging condition: z = 20 Mm step × 350–500
planes, zoom = 0.8×−1.6×, expose = 100 ms to 2 s ×
two illuminations from each side, total acquisition
time = ~45 min). The samples were cleared and
stained with SYTO 16 according to the CB-perfusion
protocol. The magnified images of SYTO 16,
mKate2 and merged signals at the ~1 mm depth
of each organ (zoom = 5×). (h) The reconstituted
3D whole-organ images from a CAG-EGFP Tg35
(GFP, green) stained with PI (male, 8 weeks old)
(imaging condition: z = 20 Mm step × 350–500
planes, zoom = 0.8×−1.6×, expose = 100 ms to 2 s ×
two illuminations from each side, total acquisition
time = ~45 min). The samples were cleared and
stained with PI (red) according to the CB-perfusion
protocol. Brightness/contrast and minimal gammavalue of images in c–h were adjusted with Imaris.
All animal experiments here were approved by
the Animal Care and Use Committee of the RIKEN
Kobe Institute and The University of Tokyo, and all
of the animals were cared for in accordance with
institutional guidelines.
Liver
© 2015 Nature America, Inc. All rights reserved.
PROTOCOL
1 mm
h
Heart
Lung
2 mm
2 mm
All raw image data collected in an uncompressed TIFF format
(16-bit images for LSFM, Fig. 4b) are typically ~7 GB in total
per brain, color and direction and thus 25–30 GB for a single
brain data set (which contains structural and signal images
acquired in two directions, D-V and V-D). Therefore, a high-spec
PC with a large memory size and a good graphics board should
be used for 3D reconstitution (Fig. 4c–h). We use a Windows
PC with Intel(R) Core(TM) i7-3970X CPU at 3.50 GHz, 64 GB
of RAM and NVIDIA GeForce GTX 690, and with Imaris
software installed.
CUBIC informatics. Here we show the step-by-step procedures for
data processing (Steps 7–15). For preprocessing of raw data, each
TIFF stack (Fig. 5, ‘Collect raw images’) is first converted to a 3D
image in the NIfTI-1 data format (.nii extension) introduced by the
Neuroimaging Informatics Technology Initiative (NIfTI) (http://
nifti.nimh.nih.gov/nifti-1/). NIfTI-1 files are visualized using software such as ITK-SNAP25. Because of the memory limitations of
the current software tools, files need to be downscaled to 25% by
discarding three of every four images of the TIFF z-stack series and
Kidney
Liver
2 mm
Pancreas
1 mm
Spleen
2 mm 2 mm
2 mm
2 mm
Stomach
Intestine
2 mm
2 mm
by changing the resolution of these images from 2,560 × 2,160 to
640 × 540 (Fig. 5). This limitation should be overcome with the
future development of image informatics tools. The downscaling
procedure is done using ImageMagick (http://imagemagick.org/)
to create a temporary stack of 16-bit PNG files for each original
TIFF stack. Next, each PNG stack is converted to a NIfTI-1 file
using the Convert3D tool from ITK-SNAP (Fig. 5). In this step,
specification of the correct spacing (given the pixel number of
raw image data and downscaling parameters) and the orientation
(which depends on the acquisition direction) is needed.
In preparation for merging NIfTI-1 data in the same color
channel of the same brain acquired from opposite directions
(D-V and V-D), the files need to be aligned (Fig. 5). In this step,
a pair of NIfTI-1 data images, the structural and signal images,
is processed. The structural image is used to calculate the transformation parameter, which is needed to align the D-V image to
the V-D image. This is calculated by the ‘ANTS’ function of the
ANTS software. The transformation can allow deformation (for
example, Symmetric Normalization) or it can be restricted to affine operations only. Then, using the ‘WarpImageMultiTransform’
NATURE PROTOCOLS | VOL.10 NO.11 | 2015 | 1713
PROTOCOL
Collect raw images
Downscaling
Convert to NIfTI-1
Align D-V and V-D
D-V image
D-V image
Horizontal
Structure
25% reduction of
voxel resolution
Signal
Coronal
Apply to
signal and
structural
images
(alignment)
Signal
D-V image
TIFF PNG
z = 2.25 mm
Sagittal
Structure
Dorsal
V-D image
Horizontal
Sagittal
n11 n12 ...
n21
Coronal
V-D image
Signal
z = ~700 planes
540
Calculate
transformation
parameter
....
z = 2.25 mm
640
Structure
2,560
V-D image
Ventral
Register
structural
image
2,160
z = 5.5 mm
z = ~175 planes
Ventral
© 2015 Nature America, Inc. All rights reserved.
Dorsal
Structure
z = 5.5 mm
Combine D-V and V-D
Composite NIfTI-1 file
Aligned NIfTI-1 files (signal)
Dorsal
D-V image
V-D image
Last plane
Last plane
Signal
D-V + V-D NIfTI-1 file (signal)
Horizontal
Sagittal
Coronal
Use D-V
(m+1 ~ last)
m-th plane
m-th plane
n-th plane
n-th plane
Ventral
First plane
First plane
Use weighed
D-V + V-D
(n–m)
Use V-D
(First ~ n–1)
Figure 5 | Preprocessing of acquired 3D image for comparison analysis. Here we use the data set of the Thy1-YPF-H Tg mouse brain acquired in ref. 18
as an example. The ‘Collect raw images’ part of the figure shows the scheme of brain 3D imaging of two different directions (D-V and V-D). Raw TIFF images
(here only YFP channel shown) in the panel are z = 2.25 mm (sharp) and 5.5 mm (blurred) for top and bottom, respectively. The raw data sets are downscaled
to 25% and converted to NIfTI-1 files (shown as capture images of ITK-SNAP). Next, structural D-V data (shown as reconstituted 3D images) via nuclei
counterstaining are registered to the corresponding V-D data. This step is to calculate transformation parameters, which is applied to the signal D-V data in
the following step (alignment). We then merge the aligned images in order to ensure sharpness throughout the resulting 3D image (YFP channel is shown
as an example again). To do so, the ‘edge content’ based on the Prewitt operator is calculated for both the D-V and V-D images. This is used to define two
threshold values at the z-slice position n and m and to create the merged composite NIfTI-1 image (shown as capture images of ITK-SNAP) according to these
values. The reconstituted 3D images and plane images in the ‘Align’ and ‘Combine’ panels were prepared by using exported TIFF images from the corresponding
NIfTI-1 data. 3D reconstitutions were performed with Imaris software as in Figure 4, and they are shown as views from the dorsal and ventral side in D-V and
V-D images, respectively. All animal experiments here were approved by the Animal Care and Use Committee of the RIKEN Kobe Institute and The University of
Tokyo, and all animals were cared for in accordance with institutional guidelines. Adapted with permission from ref. 18.
function in ANTS, we apply the transformation parameters to
align both structural and signal images of the D-V image to the
V-D image.
Next, we merge the aligned images in order to ensure sharpness throughout the resulting 3D image (Fig. 5). To do so, we use
the Prewitt operator27 to calculate the ‘edge content’ of the two
images (as a proxy for image sharpness). This is used to define two
thresholds, n and m, so that slices below n only come from the V-D
image, slices above m only come from the D-V image and slices
in between the two are a linear combination of the two images.
We then take the image pairs and the thresholds, and we create
the merged NIfTI-1 image. In order to access the values of individual pixels in the images, we use the ‘fsl2ascii’ and ‘fslascii2img’
functions of FSL28. The steps above produce a pair of D-V + V-D
composite NIfTI-1 data for both structural and signal images.
1714 | VOL.10 NO.11 | 2015 | NATURE PROTOCOLS
Next, to facilitate analysis across different brain data sets, we
align these merged NIfTI-1 files, which permits subtraction of
signal images to be calculated. We show an example of Arc-dVenus
Tg mouse brains with or without light stimuli18, which express a
destabilized version of yellow fluorescence protein Venus under
control of the Arc gene promoter26 (Fig. 6a). First, the raw data
are preprocessed as in Figure 5. Then, the composite NIfTI-1
images from different samples are aligned (Fig. 6b). As before,
images are processed in pairs (structural and signal images), and
the process relies on ‘ANTS’ and ‘WarpImageMultiTransform’.
We first align all brain data sets from the same experiment to an
internal reference (i.e., one of the brain images among the samples
of that experiment). Next, the internal reference is registered to a
brain atlas such as the Allen Brain Atlas29. These calculations thus
provide the aligned 3D images to an atlas (Fig. 6c). Finally, the
signal channel images of different brains are compared. To do so,
we normalize these aligned images so that the median intensity
inside the brain is the same across all brains. Next, the ‘fslmaths’
function is used to compare pairs of brains by subtracting one
image from the other (Fig. 6d). We provide the set of scripts for
all steps and a brief user guide as Supplementary Data, with
up-to-date versions also available on a GitHub repository (https://
github.com/SystemsResearch/CUBIC_nprot). On our website
(http://cubic.riken.jp), we also share example raw TIFF data of
Arc-dVenus Tg mice shown in Figure 6 and NIfTI-1-converted
Allen Brain Atlas data.
Limitations of the current version of CUBIC. CUBIC was
developed and optimized for whole-organ and whole-body imaging, and the informatics analysis was developed to enable a comprehensive pipeline. Thus, there are several important advantages
to CUBIC compared with other clearing methods. The first is the
active tissue decoloring ability. This method is milder for proteins, as opposed to the simple flushing or harsh decolorization
methods with peroxidase or acetone used previously30,31. This
enables a wide range of applications for not only the brain but
also other organs inside (Table 1).
Because aqueous reagents are used to clear tissue, and fluorescence signals are preserved, tissues can be imaged by fluorescence
microscopy. To achieve efficient clearing for LSFM application,
the optimal CUBIC reagent comprises five chemicals (Table 2)
and takes several days to 2 weeks to process. However, the procedure can be modified for the user’s purpose: for example, if users
plan to image by two-photon microscopy, a one-step immersion
in reagent-1 for 1–3 d is sufficient18.
The actual scalability of CUBIC has not yet been fully investigated. So far, we have tested a hemisphere of infant marmoset,
and infant and adult whole mice18,19 (Table 1). Clearing of these
samples was efficient: for example, in the case of cleared infant
a
D-V image
V-D image
b
D-V + V-D
(Venus)
D-V + V-D
(PI)
Register
and align
to internal
reference
Light (+)
Light (+)
Align all images to atlas
1 mm
Light (–)
D-V + V-D
Aligned and normalized
1 mm
d
D-V + V-D
(Venus)
Register
to atlas
Venus, high in Light (+) /
Venus, high in Light (–) /PI
Subtraction, light (+/–)
c
ABA
NIfTI-1
D-V + V-D
(PI)
V-D image
Light (–)
D-V image
Light (+)
Figure 6 | Calculation of signal subtraction. (a) Here we use the data set
of the Arc-dVenus Tg mouse brains with or without light stimuli, acquired
in ref. 18, as an example (z = 10 Mm step × 625–675 planes, zoom = 2×,
expose = 3 s × two illuminations for Venus and 300 ms × two illuminations
for PI, respectively). The reconstituted 3D images from raw TIFF stacks
are shown as views from dorsal and ventral sides in D-V and V-D images,
respectively. Yellow, Venus.; blue, PI. (b) We first preprocess these data sets
as in Figure 5. Next, we align one data set, Light (+), to the other, Light (−),
by registering the first structural data to the second (internal reference).
The internal reference is also registered to a brain atlas such as the Allen
Brain Atlas. All images are then aligned to the atlas and normalized.
The 3D reconstituted images from NIfTI-1 data for structure and signal of
Light (+) or (−) samples, after alignment to the internal reference, are shown
in dorsal view. (c) Reconstituted 3D images of the aligned and normalized
Venus channel (yellow) and aligned PI channel (blue) images from the
corresponding NIfTI-1 data. Views from the dorsal side are shown.
(d) Results of subtraction, shown as 3D reconstituted images. Views from
the dorsal side (left and top right) and the dorsolateral side (lower right) are
shown. Signals observed in Light (+) or (−) conditions are shown in magenta
and light blue, respectively. Standardized PI signals of the Light (+) sample
are merged and indicated in blue. As seen in the magnified panel (top right),
single cells in the sparsely labeled regions can be detected even in the
downscaled images. The reconstituted 3D images in b–d were prepared by
using exported TIFF images from the corresponding NIfTI-1 data and with
Imaris software as in Figures 4 and 5. All animal experiments here were
approved by the Animal Care and Use Committee of the RIKEN Kobe Institute
and The University of Tokyo, and all animals were cared for in accordance
with institutional guidelines. Adapted with permission from ref. 18.
Light (–)
© 2015 Nature America, Inc. All rights reserved.
PROTOCOL
Magnified
0.5 mm
Rotated
1 mm
1 mm
mouse, internal structures of the brain could be imaged directly
even through the skull19. However, we have not yet tested adult
primate brains (e.g., marmoset), although we plan to investigate
this in future studies. Although clearing of larger adult primate
brains might be more difficult, longer incubation times and CBperfusion may address this issue. The passive clarity technique
(PACT)/perfusion-assisted agent release in situ (PARS) protocol
of CLARITY has apparently achieved a similar scalability, and it
may be considered as an alternative.
Whether a tissue-clearing method can be combined with particular dyes or stains is an important consideration when selecting
the clearing method. Whole-organ nuclei staining for anatomical annotation, registration and image analyses has been achieved
using the CUBIC clearing procedure18,19. Although we did not
test other variations of dyes, possible limitations on some labeling
methods, particularly lipophilic dye labeling (e.g., Dil and related
dyes), may exist given that the CUBIC clearing reagents massively
remove lipids. Dyes or proteins should be fixed by PFA before clearing. In this sense, fluorescent proteins fused to a membrane protein
can be observed in the cleared tissue, whereas lipophilic dyes such
as Dil may not be readily fixed by PFA because of their chemical
structures, and thus they may be removed during clearing. This
may be a drawback to CUBIC, in which case other clearing methods
(e.g., ClearT (ref. 11), SeeDB10, FRUIT14) should be considered.
Structural distortion has been carefully addressed in some
clearing methods, such as SeeDB, and it may need to be considered when clearing tissues with the other methods. Although
we did not observe obvious changes in brain tissue even in the
detailed subcellular structures including the axon and spine 18,
such structural distortion may happen given that CUBIC reagents
remove a large proportion of the lipids and cause transient
swelling during the procedure. In addition, CUBIC has not
yet been optimized to fully clear bone and melanin pigments
(Table 1). However, this issue remains unaddressed by other
NATURE PROTOCOLS | VOL.10 NO.11 | 2015 | 1715
PROTOCOL
TABLE 1 | Tissues to which CUBIC has been successfully applied.
© 2015 Nature America, Inc. All rights reserved.
Tissue
Clearing protocol
Reagent before imaging
Clearing efficiency
Mouse whole bodya,19
CB-perfusion
Reagent-1b
Good
Mouse brain18,19
Simple immersion or CB-perfusion
Reagent-2
Good
Marmoset hemispherec,18
Simple immersion
Reagent-2
Good
Mouse heart19
Simple immersion or CB-perfusion
Reagent-2
Good
Mouse lung19
Simple immersion or CB-perfusion
Reagent-2
Good
Mouse spleen19
Simple immersion or CB-perfusion
Reagent-2
Good
Mouse liver19
Simple immersion or CB-perfusion
Reagent-2
Good
Mouse stomach19
Simple immersion or CB-perfusion
Reagent-2
Good
Mouse intestine19
Simple immersion or CB-perfusion
Reagent-2
Good
Mouse kidney19
Simple immersion or CB-perfusion
Reagent-2
Good
Mouse pancreas19
Simple immersion or CB-perfusion
Reagent-2 or reagent-1
Good
Mouse lymph noded
Simple immersion or CB-perfusion
Reagent-2
Good
Mouse muscle19
Simple immersion or CB-perfusion
Reagent-2
Good
Mouse skine,19
Simple immersion or CB-perfusion
Reagent-2
Good
Mouse bone19
CB-perfusion
Reagent-1
Partially cleared but needs further
investigationf
Tissues with melanin
(eye and hair)19
CB-perfusion
aWe
cSo
—
Not cleared with the current reagents
performed imaging of postnatal day-1 samples19. Imaging of cleared adult mouse was not tested because of size limitation of current LSFM setup. bThe cleared body specimen can be stocked in the reagent.
far we tested a brain sample of postnatal day 3. Adult brain will be tested in future studies. dE.A.S., unpublished result. eNeeds hair removal. fBone clearing of infant mice was sufficient for imaging19.
clearing methods also, and thus we are unable to suggest a suitable alternative in this scenario. Thus, this issue needs further
investigation in future studies.
As discussed earlier, imaging resolution is also a point to be
considered. We have detected and counted signals from a single
cell, and in this manuscript we therefore define this as providing
‘single-cell resolution’. This criterion is roughly evaluated by
using spot analysis of Imaris software in the data set in Figure 4
(hippocampal cells of Thy1-YFP-H Tg; Supplementary Video 1).
In our opinion, this criterion is supported overall by the calculated
optical resolution. According to the vendor’s specifications, the
optical resolution of microscopy that we use in this manuscript
has 4.2 Mm and 3.7 Mm in x-y images with 1.6× and 2× zoom,
respectively. The thickness of the light sheet is <10 Mm at the
thinnest region, which is smaller than the typical step size (10 Mm).
In a typical nuclei-stained image, a 2.5 Mm in a half-diameter
sphere of a stained nucleus is detected as 2.5 + 3.7 Mm in the halfdiameter (actual half-diameter + optical blurring of lens according to the Rayleigh criterion) when 2× zoom (5.2 Mm × 5.2 Mm per
pixel) is used. To distinguish two neighboring cells, two parameters need to be considered: exclusion volume (EV), the average
voxel volume per cell calculated from cell density, and s, the voxel
1716 | VOL.10 NO.11 | 2015 | NATURE PROTOCOLS
volume of nuclei acquired by the camera (again considering the
cell or nuclei diameter plus the optical blurring, as above). Each
nucleus is detectable and separable if EV is sufficiently larger than s.
By our calculations, s = 3 × 3 × 2 voxels (= 15.6 × 15.6 × 20 Mm =
4,867.2 Mm3, enough to include a sphere with 2.5 + 3.7 Mm in
its half-diameter). The EV of mouse cerebral cortex neuron is
11,000 Mm3 per cell (ref. 32) = (22.2 Mm)3 = 5 × 5 × 3 voxels,
which is sufficiently larger than the voxel volume used in this
study, whereas EV of mouse hippocampal CA1 is 3,900 Mm3 per
cell (ref. 33) = (15.7 Mm)3 = 3 × 3 × 2 to 4 × 4 × 2 voxels, which is at
the limit of resolution of the examples in the current manuscript.
Thus, the acquired image at its best has single-cell resolution
in the cortex and regions with similar cell density, whereas the
voxel size is not sufficient for single-cell resolution in denser
regions such as the hippocampus or cerebellar granule layer.
A further consideration is that the thinnest area of the sheet is
limited and does not cover the entire imaged field. In addition,
the analysis software used in this manuscript does not support
large image data, and collected data must be downscaled.
Although CUBIC has the potential to detect all signals with singlecell resolution, these issues will need to be further addressed
in future studies.
PROTOCOL
TABLE 2 | Other clearing methods and their applications to imaging and analysis.
© 2015 Nature America, Inc. All rights reserved.
Methods
Chemical contents
Applied imaging methods
Applied computational analysis methods
CUBIC18,19
Quadrol, Triton X-100 and urea
for reagent-1, triethanolamine,
urea and sucrose for reagent-2
Ultramicroscope (LSFM),
two-photon microscopy,
confocal microscopy
Whole-brain signal comparison analysis,
extraction of anatomical and histological
structures, quantitative analysis of
pancreatic Langerhans islets
3DISCO4,5,7,36–38
Ethanol or methanol/xylene/benzyl
alcohol/benzyl benzoate for BABB
method
Tetrahydrofuran/dichloromethane/
dibenzyl ether for THF-DBE method
Ultramicroscope (LSFM),
two-photon microscopy,
confocal microscopy
3D reconstruction, visualization of
intensities, axon tracing, cell number
quantification, tumor volume calculation
Scale9
Urea/glycerol/Triton X-100
Two-photon microscopy,
confocal microscopy
3D reconstruction, distance measurement
SeeDB10
D(-)-Fructose/A-thioglycerol
Two-photon microscopy,
confocal microscopy
3D reconstruction, axon and dendrite
tracing, cell distribution analysis
FRUIT14
D(-)-Fructose/urea
Two-photon microscopy
ClearT (ref. 11)
Formamide for ClearT, formamide
and PEG for ClearT (ref. 2)
Optical sectioning microscopy, stereomicroscopy
2,2`-Thiodiethanol12,13
2,2`-Thiodiethanol
Two-photon microscopy,
confocal microscopy,
two-photon serial sectioning
tomography
CLARITYa,13,15,17
SDS in borate buffer and one of
FocusClear, RIMS (Histodenz or
Sorbitol) or 2,2`-thiodiethanol
Confocal microscopy, two3D reconstruction, neurite tracing
photon microscopy, COLM and
other custom LSFM setups
aUse
3D reconstruction
acrylamide-embedded specimens.
MATERIALS
REAGENTS
Animal samples used for imaging
• Animals expressing fluorescent proteins can be used. Strong expression
of a bright fluorescent protein gives the best imaging results. A bright red
fluorescent protein such as mKate2 is best. So far, we have confirmed good
imaging performance with Thy1-YFP-H Tg (The Jackson Laboratory),
R26-H2B-EGFP KI (RIKEN Center for Developmental Biology (CDB)),
R26-H2B-mCherry KI (RIKEN CDB), R26-CAG-nuc-3×mKate2 KI
(RIKEN CDB & QBiC), B-actin-CAG-nuc-3×mKate2 KI (RIKEN CDB & QBiC),
CAG-EGFP Tg (Japan SLC) and Arc-dVenus Tg (Gifu University). We also
usually use C57BL/6 (CLEA Japan) to prepare cleared organs and bodies
! CAUTION Animal experiments must be performed in accordance with
governmental and institutional regulations regarding the use of animals for
research purposes. All animal experiments and housing conditions in this
manuscript were approved by the Animal Care and Use Committee of the
RIKEN Kobe Institute, The University of Tokyo and the Gifu University,
and all animals were cared for and treated humanely in accordance with the
institutional guidelines for experiments using animals.
Fixative, perfusion and storage reagents
• PBS tablets (Takara, cat. no. T9181)
• Heparin sodium (Mochida Pharmaceutical, 10,000 U/10 ml)
• PFA (Nacalai Tesque, cat. no. 02890-45) ! CAUTION PFA is toxic. Perform all
procedures in a fume hood.
• HCl (Nacalai Tesque, cat. no. 18320-15 or 18321-05) ! CAUTION HCl is
toxic. Perform all procedures in a fume hood.
• NaOH (Nacalai Tesque, cat. no. 31511-05) ! CAUTION NaOH is toxic.
Perform all procedures in a fume hood.
• Sucrose (Nacalai Tesque, cat. no. 30403-55 or 30404-45)
• TISSUE TEK O.C.T. compound (Sakura Finetek, cat. no. 4583)
• 1/2-diluted ScaleCUBIC-1 (1/2-reagent-1). ScaleCUBIC-1 is mixed with
an equivalent volume of dH2O M CRITICAL ScaleCUBIC-1 should not be
diluted with PBS, because contamination with salt decreases the clearing
performance.
Clearing, nuclei-staining and imaging reagents
• N,N,N`,N`-Tetrakis(2-hydroxypropyl)ethylenediamine (Quadrol; Tokyo
Chemical Industry, cat. no. T0781)
• 2,2`,2p-Nitrilotriethanol (triethanolamine; Wako, cat. no. 145-05605)
• Urea (Nacalai Tesque, cat. no. 35904-45 or 35907-15)
• Polyethylene glycol (PEG) mono-p-isooctylphenyl ether (Triton X-100)
M CRITICAL In our first CUBIC paper18, we mentioned that the quality of
Triton X-100 product seems crucial for preserving fluorescence signals,
and we recommended a product from Nacalai Tesque (cat. no. 25987-85).
However, that product has been discontinued. We further checked the
same chemical from Nacalai Tesque (cat. no. 12967-45), Sigma-Aldrich
(cat. no. X100, T9284, T8787, T8532) or Tokyo Chemical Industry
(cat. no. P0873), and none of them caused fluorescence quenching in
the final reagent-1 recipe, at least in the short term, and thus they can be
used as substitutes.
• Sucrose (Nacalai Tesque, cat. no. 30403-55 or 30404-45)
• Sodium azide (Nacalai Tesque, cat. no. 31208-82) ! CAUTION Sodium azide
is highly toxic.
• SYTO 16 (Life Technologies, cat. no. S7578)
• Propidium iodide (PI; Life Techologies, cat. no. P21493)
• Silicon oil TSF4300 (Momentive, RI = 1.498)
NATURE PROTOCOLS | VOL.10 NO.11 | 2015 | 1717
© 2015 Nature America, Inc. All rights reserved.
PROTOCOL
• Mineral oil (RI = 1.467; Sigma-Aldrich, cat. no. M8410)
EQUIPMENT
• Tubes, 5 ml (Eppendorf, cat. no. 0030119.401SG)
• Conical tubes, 15 ml (Corning, cat. no. 352096 or 188271)
• Conical tubes, 30 ml (Sarstedt, cat. no. 60.544)
• Conical tubes, 50 ml (Corning, cat. no. 352070 or 227261)
• Peristaltic pump (EYELA, model no. MP-2000; Fig. 3a)
• Intravenous (i.v.) injection needle, 23-G, butterfly type (Terumo,
cat. no. SV-23CLK)
• 26-G, 1/2-inch injection needle (Terumo, cat. no. NN-2613S)
• T shape stopcock (Terumo, cat. no. TS-TL2K)
• Disposable syringes, 1, 10 and 20 ml (Terumo, cat. no. SS-01T, SS-1010SZ,
SS-20ESZ)
• Vacuum desiccator (AS ONE, cat. no. VXS 1-5943-01) with vacuum pump
(ULVAC, model no. DA-15D; Fig. 2b)
• Incubation devices. We use hybridization incubator (TAITEC,
model no. HB-80, Fig. 2c) or incubator (EYELA, model no. FMS-1000 or
MHS-2000) with a rotator (TAITEC, model no. RT-5)
• Shaker (TAITEC, model no. Wave-PR or MixerXR-36, Fig. 2d)
• Magnetic stirrer for preparing highly viscous reagents (ASH,
model no. AMG-S, or IKEDA Scientific, model no. IS-20PC)
• Hot stirrer for preparing highly viscous reagents (IKA, model no. C-MAG
HS10, or Advantec, model no. SRS710HA)
• pH meter (HORIBA scientific, model no. LAQUA twin)
• Positive-displacement pipettor (Gilson, model no. Microman M-1000)
M CRITICAL We highly recommend this pipettor for measuring the weight
of the viscous materials such as Triton X-100 and aminoalcohols.
• Microwave
• Fume hood
Imaging microscopy for whole mouse organs
• Light-sheet illumination device with a macrozoom microscope4,18
• In this study, we used the Ultramicroscopes from LaVision BioTec and the
MVX-ZB10 from Olympus, equipped with Olympus MVPLAPO0.63× lens
(numerical aperture (NA) = 0.15, working distance = 87 mm)
• Imaging reservoir (100% quartz; LaVision BioTec)
• Sample holder (LaVision BioTec or customized; Fig. 4a)
• Green fluorescent signal filter (Chroma ET525/50)
• Red fluorescent signal filter (Chroma ET650/60)
• Coherent sapphire laser 488LP-100
• Coherent sapphire laser 588LP-50
• Andor sCMOS charge-coupled device (CCD) camera Neo 5.5. The camera
and the MVX microscope are connected to a camera adapter (Olympus
MVX-TV1X), tube lens (Olympus MVX-TLU) and the Ultramicroscope
filter wheel unit (LaVision BioTec) with adapters (LaVision BioTec, LV AD
MVX-1 and LV AD MVX-2)18 (Fig. 4a)
• Customized sample holder
• Glass plate for specimen mounting stage
Image-analyzing software
• General image analysis tool
• ImageJ (freeware from the US National Institutes of Health (NIH))
• Visualization tool
• Imaris (Bitplane, http://www.bitplane.com/imaris/imaris) for 3D
reconstitution of TIFF image stacks
• ITK-SNAP (freeware from Paul Yushkevich, PhD at the University
of Pennsylvania and Guido Gerig, PhD at the University of Utah):
http://www.itksnap.org/pmwiki/pmwiki.php) for NlfTl-1 3D images
Analysis tools
• Python and a C++ compiler
• Code provided as Supplementary Data
• ImageMagick (http://imagemagick.org/) installed with TIFF support
• Convert3D (http://www.itksnap.org/download/snap/process.php?link=707
4&root=nitrc)
• ANTs 1.9-v4 (ANTs, freeware from stnava, DL URL: http://sourceforge.
net/projects/advants/files/ANTS/ANTS_Latest/)
• FSL (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FslInstallation)
REAGENT SETUP
PBS Prepare PBS according to the vendor’s manual. When using PBS
tablets (Takara, cat. no. T9181), the tablet is dissolved in 1 liter of dH2O.
When PBS/0.01% (wt/vol) sodium azide is prepared, directly dissolve
0.1 g of sodium azide in 1 liter of PBS. This solution can be stored at room
temperature (18–25 °C) for several months.
1718 | VOL.10 NO.11 | 2015 | NATURE PROTOCOLS
PFA solution To prepare 4% (wt/vol) PFA in PBS, dissolve 40 g of PFA in 1
liter (total) of PBS. Heat the PBS solution (avoid boiling) and add PFA powder and a 1/500–1/1,000 volume of 1 N NaOH to help dissolving PFA faster.
After complete dissolution, adjust the pH to 7.4 using HCl. PFA can be stored
at −20 °C until use for several months. ! CAUTION PFA is a very toxic reagent.
Avoid inhalation or contact with skin and eyes. Use a draft chamber, proper
gloves and a mask to handle PFA, HCl and NaOH. M CRITICAL The pH value
of PFA is a crucial factor for an efficient clearing with lower autofluorescence.
80 wt% Quadrol Quadrol is a highly viscous liquid, and it can be used as an
80 wt% working solution. In this case, add 125 g of dH2O to 500 g of
Quadrol reagent bottle and stir it for at least 30 min. Store the solution at
room temperature for up to 1 month. M CRITICAL Quadrol is highly viscous,
and we use wt% rather than % (wt/vol) or % (vol/vol) for convenience.
ScaleCUBIC-1 (reagent-1) Reagent-1 is a mixture of urea (25 wt% final
concentration), Quadrol (25 wt% final concentration), Triton X-100 (15 wt%
final concentration) and dH2O. For example, to prepare 500 g of reagent-1
solution, mix 125 g of urea and 156 g of 80 wt% Quadrol in 144 g of dH2O
with a hot stirrer. After complete dissolution, further stir the mixture at room
temperature and add 75 g of Triton X-100. Finally, degas the reagent with a
vacuum desiccator (~0.1 MPa, ~30 min; Fig. 2b). The reagent can be stored
at room temperature for up to 1 month. Prepare 1/2-reagent-1 by mixing 1:1
of reagent-1 and dH2O. M CRITICAL Quadrol and Triton X-100 are viscous,
and therefore we use wt% rather than % (wt/vol) or % (vol/vol) for
convenience. Reagent-1 should not be prepared with PBS, because salt
contamination decreases the clearing performance. We usually prepare
the stock of reagent-1 for convenience. However, longer storage may cause
quenching of fluorescent signals. Avoid boiling during the preparation
(see also TROUBLESHOOTING section).
ScaleCUBIC-2 (reagent-2) Reagent-2 is a mixture of urea (25 wt% final
concentration), sucrose (50 wt% final concentration), triethanolamine
(10 wt% final concentration) and dH2O. To prepare 50 g of reagent-2
solution, dissolve 12.5 g of urea and 25 g of sucrose in 7.5 g of dH2O in
a microwave using a hot stirrer (Fig. 2a). After complete dissolution
(typically it takes 10–15 min), cool the mixture at room temperature,
add 5 g of triethanolamine and stir it further. The 0.1% (vol/vol) of
Triton X-100 included in the original recipe18 is not necessary. Finally,
degas the reagent with a vacuum desiccator (~0.1 MPa, ~30 min; Fig. 2b).
Prepare 1/2-reagent-2 by mixing 1:1 of reagent-2 and PBS. Reagent-2 can
be stored at room temperature for up to 2 weeks. ! CAUTION The acrid
ammonia smell in these reagents indicates degradation of urea.
Generation of ammonia itself is not apparently a problem because the
reagents are buffered with aminoalcohol in an alkaline pH range (~pH 11;
ref. 19). We recommend that users avoid excess heating during preparation.
If users experience the smell during clearing, change to fresh medium.
(see also TROUBLESHOOTING section). M CRITICAL Reagent-2 becomes
highly viscous, and therefore we use wt% rather than % (wt/vol) or %
(vol/vol) for convenience. Because water evaporation will make it
difficult for highly concentrated chemicals to dissolve, the weight should
be monitored for the addition of evaporated water after completely
dissolving urea and sucrose. Avoid boiling during the preparation.
M CRITICAL Reagent-2 should not be prepared with PBS, because salt
contamination decreases the clearing performance. We use PBS only in
preparing 1/2-reagent-2/PBS, because tissues after reagent-1 treatment tend
to be easily swollen in 1/2-reagent-2/water, which might cause distortion
of the overall structure, and because the gradual exchange from PBS
through 1/2-reagent-2/PBS to salt-free reagent-2 does not affect the final
transparency. Before clearing, 1/2-reagent-1/water is not a problem.
M CRITICAL Make sure that there is no precipitation in the reagent-2
solution stock before use. The precipitation in the stock can be dissolved
again by mild heating in a microwave. Insufficient degassing may cause
bubbles around and inside the tissue during reagent-2 treatment.
Immersion oil mix Mix 1:1 of TSF4300 and mineral oil completely
with a stirrer and degas it with a vacuum desiccator (~0.1 MPa, ~30 min;
Fig. 2b) before use. The oil mix can be repeatedly used for imaging by
filtering contaminants (clearing reagents, etc.). Its RI is 1.48–1.49, a comparable RI of reagent-2. The oil can be wiped out with 70% (vol/vol) EtOH.
M CRITICAL The mix ratio can be optionally changed because the best RI
matching may be different between organs.
PROTOCOL
EQUIPMENT SETUP
Surgical setup for the CB-perfusion protocol Typical surgical setup for
CB-perfusion is depicted in Figure 3a. In the protocol, the adult mouse is
perfused with four solutions: 20–30 ml of cold heparin-PBS, 150 ml of
cold 4% (wt/vol) PFA, 20 ml of PBS, and 20–30 ml of 1/2-diluted reagent-1.
We recommend that PFA be perfused by peristaltic pump for successful and
reproducible surgeries. Thus, we devised a surgical instrument with a combination of T shape stopcocks, a peristaltic pump connected with silicon tube, an
i.v. injection needle and a disposable syringe, as shown in Figure 3a.
! CAUTION PFA is a very toxic reagent. Avoid inhalation or contact with skin
and eyes. Use a draft chamber, proper gloves and a mask to handle PFA. Great
care in handling the injection needle is needed to avoid accidental needlesticks.
© 2015 Nature America, Inc. All rights reserved.
PROCEDURE
Anesthesia L TIMING 5 min
1| At day 0, deeply anesthetize the mice using pentobarbital (~150 mg/kg of body weight, administer intraperitoneally
(i.p.) with a 1-ml syringe and a 26-G, 1/2-inch injection needle).
! CAUTION Every experiment must follow all relevant governmental and institutional guidelines for the use of experimental
animals.
Transcardial perfusion and tissue clearing L TIMING 4–14 d
M CRITICAL In this step, we particularly focus on the full clearing protocol for the purpose of LSFM imaging. However,
the immersion period and the final transparency of samples can be varied according to the user’s experimental purpose.
2| Start clearing organs by the simple immersion protocol (option A) or the CB-perfusion and immersion protocol
(option B; CB-perfusion clears better than the immersion protocol, particularly in heme-rich organs, but it tends to cause
decreased signal intensity because of the shorter fixation time). Alternatively, start whole-body clearing by using the
CB-perfusion protocol (option C).
M CRITICAL STEP Option A is for a single whole mouse brain (Fig. 2), and it may need some modifications when other organs
are cleared. Because handling of whole-body samples in the viscous reagent-2 become difficult, particularly by causing
bubbles, the cleared whole body with option C is kept in reagent-1 but not in reagent-2.
(A) Simple immersion protocol for dissected whole brain
(i) Day 0. Perfuse the mice with 10 ml of cold PBS (pH 7.4) containing 10 U/ml of heparin at ~10 ml/min to remove the
blood from the organ as much as possible. Next, perfuse ~25 ml of cold 4% (wt/vol) PFA (pH 7.4) at ~10 ml/min.
Dissect the brain and postfix it in 10 ml of 4% (wt/vol) PFA with shaking at 4 °C for 18–24 h.
M CRITICAL STEP Cooling of PBS and 4% (wt/vol) PFA on ice is important for successful perfusion. Muscle stiffness
after perfusion is a good indicator of successful perfusion. Residual blood in the mouse brain increases autofluorescence especially in the green-laser excitation. The pH value of PFA is also crucial for efficient clearing and lower
autofluorescence. Overfixation causes both lower clearing efficiency and autofluorescence.
? TROUBLESHOOTING
(ii) Day 1. Wash the tissue sample with 10 ml of PBS/0.01% (wt/vol) sodium azide for at least 2 h twice at room
temperature to remove the remaining PFA (Fig. 2e).
J PAUSE POINT The fixed organs can be stored. First, immerse them in 10 ml of 20–30% (wt/vol) sucrose in PBS
per organ with shaking at 4 °C for 1–2 d. When the samples sink to the bottom, put them into O.C.T. compound
and immediately transfer them to −80 °C. To continue the clearing protocol, thaw the samples gradually at room
temperature and wash them with PBS at least twice, with each wash for 1 h, to remove sucrose and O.C.T. compound.
The sample will now be ready for the next step; however, we find that the clearing efficiency is reduced in samples
that have been stored.
(iii) Immerse the sample in 8–10 ml of 1/2-water-diluted reagent-1 with shaking (~60 r.p.m. if you are using the orbital
shaker of a hybridization oven shown in Fig. 2c, and ~30 r.p.m. if you are using a seesaw shaker as shown in Fig. 2d)
or rotation (~5 r.p.m.) at 37 °C for 3–6 h. We recommend using a 30-ml conical tube for clearing a single brain rather
than a 15-ml tube in this and further clearing Step 2A(iv,v), because of sample swelling. Clearing effects can be
observed during this step (Fig. 2e). Note that a nuclear staining dye, such as SYTO 16 (1–2 MM) and PI (5–10 Mg/ml),
can also be added to 1/2-diluted reagent-1 at this step.
M CRITICAL STEP Inefficient mixing of the reagent and samples during clearing may affect the final clearing
performance. Pretreatment with 1/2-diluted reagent-1 gives a more effective final clearing efficiency than direct
immersion in reagent-1. However, this step can be skipped for other purposes such as two-photon imaging with a
partially cleared sample, for example18. This direct immersion procedure gives better clearing results than some other
clearing methods19, and it can be used for two-photon imaging18 and possibly as part of the sample preparation for
single-photon imaging.
NATURE PROTOCOLS | VOL.10 NO.11 | 2015 | 1719
© 2015 Nature America, Inc. All rights reserved.
PROTOCOL
(iv) Discard 1/2-diluted reagent-1. Immediately add 8–10 ml of reagent-1 and gently shake it or rotate the sample at
37 °C overnight. If desired, the same concentration of nuclear staining dye used in the previous step should be added
to reagent-1.
! CAUTION Reagent-1 can erase oily pen marks easily. Make sure that the tube is sealed by wrapping Parafilm around
the lid and the top of it. Sample labels should be written on both the body and lid of the tube to avoid loss of
information. An ammonia smell indicates degradation of urea, and that the reagent should be replaced by fresh medium.
? TROUBLESHOOTING
(v) Day 2. Replace 8–10 ml of reagent-1 and continue gentle shaking or rotation at 37 °C. Replace the reagent every
2 d (days 4 and 6). Also refresh any nuclear staining dye on day 4 and day 6. Typically, the brain will be sufficiently
cleared by days 7–8 (Fig. 2e).
! CAUTION As CUBIC-treated organs soften, we recommend using spoons instead of forceps for handling them in order
to avoid damage.
M CRITICAL STEP If the white matter has not substantially cleared by 8 d of immersion, try further immersion by
placing it into fresh reagent-1 for an additional 1–2 d.
? TROUBLESHOOTING
(vi) Days 7–10. To stop the clearing procedure, wash the sample with 20 ml of PBS/0.01% (wt/vol) sodium azide with
gentle shaking or rotation at room temperature three times for at least 2 h each time. We typically wash the sample
once for 2 h, once overnight and again once for 2 h. When a sample is stained with PI, further staining during
this step is needed: thus, incubate the washed sample in ~5 ml of PBS/0.01% (wt/vol) sodium azide containing
5–10 Mg/ml of PI for an additional 3 d (or more, if needed) at 37 °C with rotation18.
M CRITICAL STEP Complete removal of reagent-1 during the washing step is crucial for final clearing efficiency.
M CRITICAL STEP For cryoprotection at this step, we recommend using 30% (wt/vol) sucrose in PBS rather than
20% (wt/vol) sucrose solution to avoid any damage to the sample.
? TROUBLESHOOTING
J PAUSE POINT Organs can be stored. First, immerse them in 10 ml of 30% (wt/vol) sucrose in PBS/0.01% (wt/vol)
sodium azide per organ with shaking at room temperature overnight. When the samples sink to the bottom, put them
into O.C.T. compound and immediately store them at −80 °C. Thaw the samples as described in the PAUSE POINT
callout at Step 2A(ii).
(vii) Days 8–11. Degas the sample in a limited volume of PBS with a vacuum desiccator (Fig. 2b). To do this, immerse
the sample in ~5 ml of 1/2-PBS-diluted reagent-2 and shake it in a 5-ml tube for 6–24 h at 37 °C or at room
temperature (Fig. 2e). Check whether the sample sinks to the bottom (a sign of complete immersion).
M CRITICAL STEP Degassing of the sample prevents air bubbles from remaining in the ventricle.
? TROUBLESHOOTING
(viii) Days 8–11. Immerse the sample in ~5 ml of reagent-2 in a 5-ml tube and gently shake it at 37 °C overnight.
The next day, replace the reagent with fresh reagent and further incubate for ~24 h (Fig. 2e).
! CAUTION Do not rotate the tube to avoid making bubbles. Samples do not sink in the highly viscous reagent-2, and
it is difficult to take images in the reagent. The reagent-2–treated samples should be immersed in the low-viscosity
immersion oil mix at imaging steps. When structural distortion is apparent after reagent-2 treatment at 37 °C, try
incubation at room temperature for a longer time. Adjustment of PBS content in 1/2-diluted reagent-2 may also
mitigate shrinkage or swelling.
? TROUBLESHOOTING
J PAUSE POINT Organs can be left in reagent-2 for up to 1 week at room temperature. Further immersion increases
the final transparency, but it also causes swelling of the sample. After imaging, the sample can be washed with
PBS/0.01% (wt/vol) sodium azide, completely immersed in 30% (wt/vol) sucrose in PBS/0.01% (wt/vol) sodium azide
and stored in O.C.T. compound at −80 °C, as described in the PAUSE POINT callout in Step 2A(vi).
(B) CB-perfusion and immersion protocol for faster clearing of whole organs
(i) Prepare the surgical setup as shown in Figure 3a.
(ii) Day 0. Perfuse the mice with 20–30 ml of cold PBS (pH 7.4) containing 10 U/ml heparin at ~10 ml/min to remove the
blood from the tissues as much as possible.
M CRITICAL STEP Insufficient removal of blood inside the tissue prolongs the clearing period, and it may cause low
clearing performance.
(iii) Perfuse the mice with 150 ml of cold 4% (wt/vol) PFA in PBS (pH 7.4) at ~15 ml/min using a peristaltic pump.
! CAUTION PFA is a very toxic reagent. Perform all procedures in a fume hood with a safety glass to avoid inhalation
or contact with skin and eyes.
1720 | VOL.10 NO.11 | 2015 | NATURE PROTOCOLS
© 2015 Nature America, Inc. All rights reserved.
PROTOCOL
M CRITICAL STEP If the signal from a target reporter protein is weak, a prolonged perfusion period may be more
effective, or clearing as described in Step 2A.
M CRITICAL STEP Cooling of PBS and 4% (wt/vol) PFA on ice is important for successful perfusion.
? TROUBLESHOOTING
(iv) Perfuse the mice with 20 ml of PBS (pH 7.4) at ~10 ml/min to wash out PFA, followed by perfusion of 20–30 ml of
1/2-diluted reagent-1 at the same injection rate. Make sure that the organs become translucent by the end of the
perfusion (Fig. 3b). Note that a nuclear staining dye, such as SYTO 16 (1–2 MM) and PI (5–10 Mg/ml), can also be
added to 1/2-diluted reagent-1 at this CB-perfusion step.
M CRITICAL STEP Perfusion efficiency is crucial to the final clearing efficiency. Some organs such as the pancreas and
spleen are good indicators to evaluate perfusion efficiency (Fig. 3b).
(v) Dissect the organs of interest and immerse these in reagent-1. Several organs can be processed in a single tube, but
the stomach and intestine should be separated into different tubes. Gastrointestinal content in these organs should
be removed as much as possible in this step. Thus, immerse several organs such as heart, lung, kidney, spleen,
pancreas and a piece of liver in 40 ml of reagent-1 or immerse each small organ such as heart, lung, kidney, spleen
and pancreas in 5 ml of reagent-1. Incubate the samples with shaking (~60 r.p.m. if you are using an orbital shaker
in a hybridization oven as shown in Fig. 2c, or ~30 r.p.m. if you are using a seesaw shaker as shown in Fig. 2d) or
rotation (~5 r.p.m.) at 37 °C overnight. Add the same concentration of any nuclear staining dye used at Step 2B(iv) to
reagent-1.
! CAUTION Reagent-1 erases oily pen marks easily. Make sure that the tube is sealed by wrapping Parafilm around
the lid. Sample labels should be written on both the body and the lid of the tube to avoid loss of information.
An ammonia smell indicates degradation of urea, and in this scenario the reagent should be replaced with fresh medium.
M CRITICAL STEP Efficient mixing of the reagent and samples during clearing may affect the final clearing performance.
For efficient clearing, the samples of interest should be immersed in at least a fivefold volume of reagent-1.
? TROUBLESHOOTING
(vi) Day 1. Make sure that the organs are transparent and that the color of the supernatant has turned olive green. Replace
reagent-1 with the same volume of fresh reagent-1 and continue clearing by shaking at 37 °C. If appropriate, also
refresh the nuclear staining dye. Replace reagent-1 and any nuclear staining dye again at days 2 and 4. The total
incubation time for the complete clearing depends on the organ: 1 day of reagent-1 treatment is usually sufficient for
pancreas, spleen and intestine. However, note that we treated all indicated organs in Figure 3c with reagent-1 for 5 d.
M CRITICAL STEP Typically, successfully CB-perfused organs are turned almost transparent, with the exception of liver
and lung, by day 1 (Fig. 3c). Opacity in the lung occurs mainly from bubbles. Note that the color change of the
supernatant indicates decolorization of tissues as a result of heme elution.
? TROUBLESHOOTING
(vii) Days 2–6. To stop the clearing procedure, wash the samples with the same volume of PBS with gentle shaking or
rotation at room temperature three times for 2 h each time. After the PBS wash, move to the next step immediately.
M CRITICAL STEP CB-perfused samples are prone to overshrinking in the washing step. Do not wash the samples in
PBS more than three times for 2 h each time.
? TROUBLESHOOTING
(viii) Immerse the sample in the same volume of 1/2-PBS-diluted glycerol and shake it for 6 h to 24 h at room temperature.
Check whether the sample sinks to the bottom (a sign of complete immersion).
? TROUBLESHOOTING
(ix) Days 3–7. Immerse the samples in the same volume of reagent-2 and gently shake them at 37 °C overnight. The next
day, replace the reagent with fresh reagent and further incubate the samples for several days. Typically, an apparent
transparency plateau is reached after 2–3 d of reagent-2 treatment. At day 10, almost all organs should be transparent, as shown in Figure 3c. The gastrointestinal contents of the stomach and intestine should be removed as much as
possible before the following imaging step.
! CAUTION To avoid making bubbles, do not rotate the tube. Samples do not sink in the highly viscous reagent-2, and
it is difficult to take images in the reagent. The reagent-2–treated samples should be immersed in the low-viscosity
immersion oil mix at imaging steps. When structural distortion is apparent after reagent-2 treatment at 37 °C, try
incubating room temperature for a longer time or testing the simple immersion protocol (Step 2A).
? TROUBLESHOOTING
J PAUSE POINT Tissues can be left in reagent-2 for up to 1 week at room temperature. Further immersion increases
the final transparency, but it also causes swelling of the sample. After imaging, the sample can be washed with PBS,
completely immersed in 30% (wt/vol) sucrose in PBS and stored in O.C.T. compound at −80 °C, as in the PAUSE POINT
callout in Step 2A(vi).
NATURE PROTOCOLS | VOL.10 NO.11 | 2015 | 1721
© 2015 Nature America, Inc. All rights reserved.
PROTOCOL
(C) CB-perfusion protocol for whole-body clearing
M CRITICAL Here we describe clearing of whole adult mouse body only using reagent-1. This overcomes the difficulties of
handling whole-body samples in the viscous reagent-2, particularly because of bubble formation. For infant mouse wholebody imaging, perfusion and immersion of reagent-1 was sufficient (Table 1)19. Adult whole-body imaging is not applicable
to the microscope setup introduced in this manuscript because of stage size limitations.
(i) Perform CB-perfusion as described in Step 2C(i–iv).
(ii) Detach the skin from the body. Carefully remove as much pelage as possible. Make sure that the body is partially
transparent after the CB-perfusion. Typically, glands such as the pancreas and submaxillary gland are almost
transparent (Fig. 3d). Spleen is also as a good indicator of successful perfusion. Immerse the body in 200 ml of
reagent-1. Place the container on an orbital shaker set at ~60 r.p.m. or a seesaw shaker set at ~30 r.p.m. in the
incubator at 37 °C overnight. Use the same concentration of the nuclear staining dye used at Step 2B(iv), and add to
reagent-1 if desired.
M CRITICAL STEP Inefficient mixing of the reagent and samples during clearing may affect the final clearing
performance.
(iii) Replace the same volume of reagent-1 and continue gentle shaking at 37 °C. Refresh the nuclear staining dye if
necessary. Replace the reagent (and any nuclear staining dye) every day in the initial week, and every 2 or 3 d in
the second week. Continue the clearing with reagent-1 for at least 2 weeks. Typically, major abdominal organs except
bones and intestinal contents become sufficiently transparent after 2 weeks of treatment with reagent-1 (Fig. 3d).
M CRITICAL STEP An ammonia smell indicates the degradation of urea, and the reagent should be replaced with
fresh medium if this is smell is present.
? TROUBLESHOOTING
J PAUSE POINT The whole body can be kept in reagent-1 for up to several months at room temperature.
Imaging cleared tissues with macrozoom LSFM L TIMING 1–3 h per sample
M CRITICAL To perform a rapid image acquisition of whole organs, a light-sheet illumination unit combined with a
macrozoom microscope is suitable. Here we describe our setup using the Ultramicroscope combined with MVX-ZB10
(LaVision BioTec and Olympus). A confocal or a multiphoton microscope can also be used, but for more-limited regions.
3| Before imaging, wipe reagent-2–treated samples softly with a Kimwipe to remove excess reagent-2 on the surface, and
then immerse the sample into the oil mix for 10 min to 1 h. This process also helps remove bubbles around the tissue.
If bubbles attach on the surface of the sample, carefully remove them with a needle or tapered forceps.
4| Set the imaging reservoir filled with the immersion oil mix, and then set the sample holder. Put a glass slide on the
sample holder (Fig. 4a).
5| Put the sample on the glass slide (Fig. 4a). Acquire a live image with an appropriate laser and filter pair to adjust the
focus and the sample position to the center. A whole mouse brain image can be captured using 1.6× to 2× zoom on the
MVX-ZB10.
M CRITICAL STEP To avoid fluorescence quenching, laser power should be weakened during the adjustment of position
and focus.
6| Set the focus position of the illumination sheet, z-range (in the case of whole brain, typically ~7 mm in total), Z-step
size (typically 10 Mm per step), laser power (typically 70–100%) and exposure time (typically 50 ms–1 s for each side).
Each plane should be illuminated from both the right and left sides, and a merged image with maximum intensity should
be saved. The exposure times should be adjusted according to the fluorescent signal intensities of each sample. After all
parameters are adjusted, start image acquisition. When multicolor images are needed, repeat the image acquisition procedure
with the same z-range and readjust the laser power and exposure time. To collect both D-V and V-D data sets, manually flip
the sample upside down and acquire the image again for the opposite orientation.
M CRITICAL STEP To take images with a high signal-to-noise ratio, it is important to use bright fluorescent proteins and
chemicals. Because ~700 images (~11 MB per image, total ~7 GB) are acquired per color and direction, each image should be
saved to a secondary storage location (e.g., a hard drive) during acquisition. Each stack (one color, one direction) is saved
to a different folder. For further signal comparison steps (Steps 7–15), signal and structural images of both D-V and V-D
directions are needed. We recommend a simple naming rule for these folders with four fields separated by an underscore:
information about the experiment (including imaging date), a unique ID for this brain, information about the imaging
direction (‘VD’ or ‘DV’) and information about the channel (‘nuclear’ for the nuclear counterstaining and ‘geneExp’ for the
1722 | VOL.10 NO.11 | 2015 | NATURE PROTOCOLS
PROTOCOL
signal channel). For instance, 20131118LAdV_001_nuclear_DV. The code provided for the informatics section assumes
that the naming convention is respected. It is also important to note that white spaces and special characters must
be avoided.
? TROUBLESHOOTING
Informatics for signal comparison L TIMING 1–9 h for a two-brain data set, depending on the registration method
M CRITICAL We provide our source code and an additional user manual as Supplementary Data, as well as the NIfTI-1converted Allen Brain Atlas data and Arc-dVenus Tg mouse brain images used in Figure 6 on our website (http://cubic.
riken.jp). Note that timing is roughly proportional to the number of brains, and it can vary according to different computer
specifications.
7| If this is the first time the pipeline is being run, install all required software and copy the provided code to the
computer used for the analysis. Compile the C++ files for edge detection (g++ -O3 edge_detection_Prewitt.cpp -o
edge_detection_Prewitt) and file merging (g++ -O3 file_merging.cpp -o file_merging).
© 2015 Nature America, Inc. All rights reserved.
8| Construct a 3D NIfTI-1 file for each TIFF stack with the ‘convertTiffFiles.py’ script. This takes ~2 min per stack, and each
brain sample corresponds to four stacks (two channels, two acquisition directions).
?TROUBLESHOOTING
9| Align images of the same brain acquired from opposite directions, with the ‘sameBrainAlignment.py’ script. For each
brain, this includes the following: registration of the V-D–acquired nuclear counterstaining image to the D-V–acquired one,
alignment of the V-D–acquired nuclear counterstaining image and alignment of the V-D–acquired signal channel image.
The registration takes ~1 h 30 min using symmetric normalization or 3–5 min using affine transformations only, and both
alignments take under 1 min.
?TROUBLESHOOTING
10| Merge the V-D–acquired and D-V–acquired images. Use ‘edgeDetection.py ‘to calculate the n and m thresholds for each
brain and channel (4 min per brain), and use these results in ‘fileMerging.py’ to merge the files (7–8 min per brain).
?TROUBLESHOOTING
11| Choose one brain to be used as internal reference, and align all the other brains to this reference with ‘internalAlignment.py’
(for each brain, ~1 h 30 min for the registration if you are using symmetric normalization or 3–5 min if you are using
affine transformation only, and <1 min for the alignment of both channels).
?TROUBLESHOOTING
12| Use ‘atlasAlignment.py’ to register the internal reference to the brain atlas (~1 h 30 min), and align all brains to the
atlas (~1 min per brain).
?TROUBLESHOOTING
13| Calculate the normalization factors (7 min, plus 7 min per brain) with ‘median_brainOnly.py’.
?TROUBLESHOOTING
14| Normalize and compare brains (e.g., for subtraction) with ‘normalisation_comparison.py’ (exact timing is dependent on
comparison, but it typically takes 10–60 min).
?TROUBLESHOOTING
15| Export a TIFF stack for the resulting files with ‘exportTiffStack.py’ (about 4 min per NIfTI-1 file).
M CRITICAL STEP Although registration with Symmetric Normalization (SyN) is effective, in some cases it may cause
structural deformation or distortion. Users can choose whether to use SyN or affine-only registration. We recommend
checking the quality of final aligned data and, if necessary, trying SyN or affine-only instead. In this manuscript, we show
data aligned with SyN in Figure 5 and with affine-only transformation in Figure 6, for illustrative purposes.
?TROUBLESHOOTING
?TROUBLESHOOTING
Troubleshooting advice can be found in Table 3.
NATURE PROTOCOLS | VOL.10 NO.11 | 2015 | 1723
PROTOCOL
TABLE 3 | Troubleshooting table.
Step
Problem
Possible reason
Solution
Reagent Setup
(preparation)
Possible degradation of
chemicals in CUBIC
reagents
Too much heating (which
causes ammonia odor)
Milder and shorter heating during preparation
Precipitate in reagent-2
Lower room temperature, particularly
in winter
Prepare the reagent before use; heat it in a
microwave for 5–10 s (avoid boiling)
Poor PFA perfusion
Insufficient cooling of PBS and PFA
Keep PBS and PFA on ice just before perfusion
Wrong position of the tip of the needle
Make sure that the tip of the needle is in the
left ventricle of the heart
Insufficient pressure for perfusion
Make sure that the perfusion outlet only
occurs at the cut in the liver
Alkaline pH of PFA
Adjust the pH of PFA to ~7.4
Too much fixation time
Stop postfixation within 24 h
Insufficient incubation time,
reagent amount or mixing
Incubate it longer in reagent-1 (a few days
more); exchange reagent-1 more frequently
(every day rather than every 2 d); shake or
rotate appropriately
Use of an aged animal
Use a younger animal
Organ-dependent differences in clearing
Use reagent-1 rather than reagent-2 for the
final clearing reagent (e.g., pancreas becomes
clearer in reagent-1; see Fig. 3)
Inappropriate salt concentration during
treatment with 1/2 reagent-2
If the organ has shrunk too much, decrease
the concentration of PBS in the 1/2 reagent-2
(e.g., mix 1:1:2 of dH2O:PBS:reagent-2 rather
than 1:1 mix of PBS:reagent-2); use of 1/2
reagent-2/water instead of PBS causes
swelling
Incubation at 37 °C
Try to incubate samples at room
temperature during reagent-2 steps
(takes more time)
Insufficient replacement in 1/2
reagent-2
Sufficiently incubate the organ in the reagent
for complete replacement (~24 h)
Use of an infant or juvenile animal
Organs from an infant or juvenile animal
tend to shrink more in reagent-2 and
need less incubation time during
reagent-1 treatment
Use of samples prepared with
CB-perfusion
Try the simple immersion protocol
Incomplete dissolution or precipitate
in reagent-2
Make sure that no precipitate exists in the
prepared/stocked reagent-2
Insufficient incubation in reagent-1
Increase reagent-1 treatment time
Insufficient replacement into reagent-2
Use more reagent-2 and exchange the reagent
one or two more times
(continued)
© 2015 Nature America, Inc. All rights reserved.
2A(i), 2B(iii)
(fixation)
Step 2A(iv–viii),
2B(v–ix), 2C(iii)
(clearing)
Poor organ clearing
during reagent-1
treatment
Excess shrinkage or
deformation of cleared
organ after reagent-2
Poor organ clearing
during reagent-2 steps
1724 | VOL.10 NO.11 | 2015 | NATURE PROTOCOLS
PROTOCOL
TABLE 3 | Troubleshooting table (continued).
Step
Problem
Possible reason
Solution
Step 2A(iv–viii),
2B(v–ix), 2C(iii)
(clearing)
Ammonia smell during
clearing
Degradation of urea
Replace with fresh medium, and avoid too
much heating during preparation of reagents
Bubbles on inside
structures of the organs
(e.g., brain ventricles)
after reagent-2
Insufficient degassing
Degas the reagent-2 after preparation; degas
the sample in a limited volume of PBS before
the reagent-2 incubation
Rotation during reagent-2
Incubate in the reagent-2 with gentle
shaking rather than rotation
Keeping the sample at −20 °C to −30 °C
(which possibly causes growth of water
crystals inside the tissue)
Stock the sample in O.C.T. compound at
−80 °C; thaw the sample gradually at
room temperature
Insufficient replacement by sucrose
solution
Increase incubation time in the sucrose
solution (i.e., until organs sink)
Floating sample during
imaging
Use of reagent-2 rather than the
immersion oil mix
Immerse the sample in oil
Bubble on the surface
Insufficient removal of the reagent-2
Remove excess reagent-2 before immersion
into the oil mix; remove bubbles in the
oil with a needle or a tapered forceps
High autofluorescence
Insufficient fluorescent signal
Select a bright fluorescent protein (e.g., YFP,
Venus) with a strong expression promoter
(e.g., CAG); avoid green channel and use a
red fluorescent protein (e.g., mKate2)
Alkaline pH of PFA
Adjust the pH of PFA to ~7.4
Use of an aged animal
Use a younger animal
Insufficient fluorescence signals
Select a bright fluorescent protein with a
strong expression promoter as above
Signal decrease in CB-perfusion
Use more PFA for perfusion, and pause the
perfusion procedure after PFA perfusion to
increase fixation reaction (~3 h); try the
immersion protocol; select an animal strain
with a bright fluorescent signal
Temperature during clearing
Higher temperature during clearing may
decrease fluorescence signals8. Try to incubate at room temperature rather than 37 °C.
(Note that lower temperature also decreases
clearing efficiency)
Inappropriate setting of microscopy
Check the setting of laser power, filter and
exposure time
Inappropriate setting of the light-sheet
focus
Adjust the light-sheet focus to the region of
interest (for the LaVision Ultramicroscope,
the width of focused sheet is 1/3–1/4 of
an adult mouse hemisphere, and thus it is
impossible to take images with adjusted
focus throughout the brain)
© 2015 Nature America, Inc. All rights reserved.
Tissue damage when
freezing
6 (imaging)
Weak or nondetectable
fluorescence
Poor z-resolution
(continued)
NATURE PROTOCOLS | VOL.10 NO.11 | 2015 | 1725
PROTOCOL
© 2015 Nature America, Inc. All rights reserved.
TABLE 3 | Troubleshooting table (continued).
Step
Problem
Possible reason
Solution
8–15 (analysis)
Brain appears deformed
in the NIfTI-1 file
(Step 8)
Incorrect scaling
Check the parameter file and ensure that the
correct voxel dimensions are given (for the
raw TIFF image, in mm)
Incorrect brain
orientation (Step 8)
Different brain position during imaging
Reorient the NIfTI-1 file in ITK-SNAP until
it matches your raw TIFF stack. Note the
correct orientation (e.g., RPS) and edit
the convertTiffFiles.py script accordingly
(line 148 for DV-acquired files, line 150 for
VD-acquired files)
Error message: ‘No such
file or directory’
Wrong folder name or brain ID in the
parameter file
Check the parameter file for that step and
ensure that all brain IDs are valid and that all
folders exist
Error message: ‘Command
not found’
Missing software
Make sure that all the required tools are
installed, that they are accessible from the
command line and that our two C++ programs
are compiled
L TIMING
Step 1, anesthesia: 5 min
Step 2, transcardial perfusion and tissue clearing: 4–14 d
Steps 3–6, imaging of the cleared tissues with the macrozoom LSFM: 1–3 h per sample (depending on the number of samples,
color and direction, and the required exposure time)
Steps 7–15, informatics for signal comparison: 8–9 h for a two-brain data set (reduced to 1–2 h if you are using affine
registration instead of symmetric normalization)
ANTICIPATED RESULTS
The CUBIC pipeline can be used for whole-organ or whole-body clearing. It is simple, efficient and reproducible, and thus
it can be applied to simultaneous multisample clearing in a single tube (Fig. 2c) or a plastic container. The procedure can
be performed using equipment usually used in a typical biology laboratory (Fig. 2a–d). By simple immersion of sample in
reagent-1, clearing is obvious within several hours (Fig. 2e), and such partially cleared samples are even applicable to deep
region imaging with two-photon microscopy18. The CUBIC reagents also decolorize organs and the whole body by removing
heme, which is also apparent just after CB-perfusion (Fig. 3a,b), and this ability enables whole-body clearing within 2 weeks
(Fig. 3c,d). We summarize the organs we have used CUBIC on in Table 1. Removal of the CUBIC reagents by PBS washing
reverses the cleared state (Fig. 2e), but the tissue is clear again if it is re-immersed into CUBIC reagents.
Figure 4a–f shows typical imaging results for the Thy1-YFP-H Tg mouse brain at 1.6× zoom. Overall, soma and other
subcellular structures such as neurites in the mouse brain can be captured, provided that they are sparsely labeled
(Fig. 4b,d,e and Supplementary Video 1). Other organs were also subjected to rapid, multicolor 3D imaging in the same
platform (Fig. 4g,h). Of note, these data were collected for ~30–60 min for each direction/color.
According to the user’s experimental purpose, high- or low-resolution images may be acquired. We use relatively low
resolution because our primary purpose is to analyze cells within the context of a whole organ or body in a high-throughput
manner. However, subcellular structures can be observed in the cleared tissue (Fig. 4b–f)18, and thus CUBIC permits more
detailed observations by using a higher zoom on the LSFM instrument, or by using higher-NA objective lenses on other
confocal and two-photon microscopes.
We implemented an image informatics method originally used in fMRI analysis to compare different brains. In this
pipeline, the acquired data sets are preprocessed by merging D-V and V-D to make sure that the resulting images are clear
throughout the z-stack (Fig. 5). Then, signal subtraction between samples is calculated (Fig. 6). Here we show an example
of Arc-dVenus Tg mouse brains with or without light stimuli18,26 (Fig. 6a). The raw images were preprocessed, aligned and
normalized, and then subtraction of Venus signal was calculated to visualize light-responsive regions at the whole-brain scale
(Fig. 6b–d). These calculations could be achieved by whole-organ counterstaining in the CUBIC clearing protocol. Note that
the final resolution of images in this process is downscaled, and it is more difficult to achieve single-cell resolution
throughout the entire imaging field, because of the current software limitations. This will be addressed in future studies.
1726 | VOL.10 NO.11 | 2015 | NATURE PROTOCOLS
PROTOCOL
In summary, CUBIC provides a platform for comprehensive cell detection and analysis across whole organs and the body.
Possible applications of CUBIC will be for whole-organ samples from multiple conditions or time points, detection of aberrant
3D morphological changes in diseased tissues or a scalable observation of tissue-to-subcellular structures in a single cleared
organ. The method will therefore support organism-level systems biology and facilitate our understanding of complicated
biological phenomena in multicellular organisms.
© 2015 Nature America, Inc. All rights reserved.
Note: Any Supplementary Information and Source Data files are available in the
online version of the paper.
ACKNOWLEDGMENTS We thank the lab members at RIKEN QBiC and
The University of Tokyo, in particular S.I. Kubota for his kind help in
preparing the materials; A. Millius for his critical reading and editing of the
manuscript; and T. Mano for his kind contributions and suggestions to discuss
image resolution. This work was supported by the Program for Innovative
Cell Biology by Innovative Technology and the Brain Mapping by Integrated
Neurotechnologies for Disease Studies (Brain/MINDS) from the Ministry
of Education, Culture, Sports, Science and Technology (MEXT) of Japan;
a Grant-in-Aid for Scientific Research (S) (grant No. 25221004), for Scientific
Research on Innovative Areas (grant no. 23115006) and for Young Scientists (A)
(grant no. 15H05650) from MEXT/Japan Society for the Promotion of Science
(JSPS); by the strategic programs for R&D (President’s discretionary fund) of
RIKEN; by an intramural Grant-in-Aid from the RIKEN QBiC; by a grant from
AMED-CREST; by the RIKEN Special Postdoctoral Research Program; by the RIKEN
Foreign Postdoctoral Researcher Program; by a Grant-in-Aid from the Japan
Foundation for Applied Enzymology; by the Brain Sciences Project of the Center
for Novel Science Initiatives of the National Institutes of Natural Sciences
(grant nos. BS261004 and BS271005); by the Tokyo Society of Medical Science;
and by the Shimabara Science Promotion Foundation.
AUTHOR CONTRIBUTIONS H.R.U., E.A.S., K.T. and D.P. designed the study.
E.A.S., H.Y. and A.K. performed most of the immersion protocol. K.T. performed
most of the CB-perfusion protocol. D.P. performed most of the computational
image analysis. A.K. developed the improved immersion protocol. All authors
discussed the results and commented on the manuscript text.
COMPETING FINANCIAL INTERESTS The authors declare no competing financial
interests.
Reprints and permissions information is available online at http://www.nature.
com/reprints/index.html.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
Keller, P.J. & Ahrens, M.B. Visualizing whole-brain activity and
development at the single-cell level using light-sheet microscopy.
Neuron 85, 462–483 (2015).
Osten, P. & Margrie, T.W. Mapping brain circuitry with a light microscope.
Nat. Methods 10, 515–523 (2013).
Spalteholz, W. Über das Durchsichtigmachen von menschlichen und
tierischen Präparaten (S. Hirzel, 1914).
Dodt, H.U. et al. Ultramicroscopy: three-dimensional visualization of neuronal
networks in the whole mouse brain. Nat. Methods 4, 331–336 (2007).
Ertürk, A. et al. Three-dimensional imaging of solvent-cleared organs using
3DISCO. Nat. Protoc. 7, 1983–1995 (2012).
Becker, K., Jährling, N., Saghafi, S., Weiler, R. & Dodt, H.U. Chemical
clearing and dehydration of GFP expressing mouse brains. PLoS ONE 7,
e33916 (2012).
Renier, N. et al. iDISCO: a simple, rapid method to immunolabel large
tissue samples for volume imaging. Cell 159, 896–910 (2014).
Schwarz, M.K. et al. Fluorescent-protein stabilization and high-resolution
imaging of cleared, intact mouse brains. PLoS ONE 10, e0124650 (2015).
Hama, H. et al. Scale: a chemical approach for fluorescence imaging and
reconstruction of transparent mouse brain. Nat. Neurosci. 14, 1481–1488
(2011).
Ke, M.T., Fujimoto, S. & Imai, T. SeeDB: a simple and morphologypreserving optical clearing agent for neuronal circuit reconstruction.
Nat. Neurosci. 16, 1154–1161 (2013).
Kuwajima, T. et al. ClearT: a detergent- and solvent-free clearing method for
neuronal and non-neuronal tissue. Development 140, 1364–1368 (2013).
Aoyagi, Y., Kawakami, R., Osanai, H., Hibi, T. & Nemoto, T. A rapid
optical clearing protocol using 2,2`-thiodiethanol for microscopic
observation of fixed mouse brain. PLoS ONE 10, e0116280 (2015).
Costantini, I. et al. A versatile clearing agent for multi-modal brain
imaging. Sci. Rep. 5, 9808 (2015).
14. Hou, B. et al. Scalable and DiI-compatible optical clearance of the
mammalian brain. Front. Neuroanat. 9, 19 (2015).
15. Chung, K. et al. Structural and molecular interrogation of intact biological
systems. Nature 497, 332–337 (2013).
16. Tomer, R., Ye, L., Hsueh, B. & Deisseroth, K. Advanced CLARITY for
rapid and high-resolution imaging of intact tissues. Nat. Protoc. 9,
1682–1697 (2014).
17. Yang, B. et al. Single-cell phenotyping within transparent intact tissue
through whole-body clearing. Cell 158, 945–958 (2014).
18. Susaki, E.A. et al. Whole-brain imaging with single-cell resolution
using chemical cocktails and computational analysis. Cell 157,
726–739 (2014).
19. Tainaka, K. et al. Whole-body imaging with single-cell resolution
by tissue decolorization. Cell 159, 911–924 (2014).
20. Keller, P.J., Schmidt, A.D., Wittbrodt, J. & Stelzer, E.H.K. Reconstruction
of zebrafish early embryonic development by scanned light sheet
microscopy. Science 322, 1065–1069 (2008).
21. Keller, P.J. et al. Fast, high-contrast imaging of animal development
with scanned light sheet-based structured-illumination microscopy.
Nat. Methods 7, 637–642 (2010).
22. Ahrens, M.B., Orger, M.B., Robson, D.N., Li, J.M. & Keller, P.J. Whole-brain
functional imaging at cellular resolution using light-sheet microscopy.
Nat. Methods 10, 413–420 (2013).
23. Panier, T. et al. Fast functional imaging of multiple brain regions in
intact zebrafish larvae using selective plane illumination microscopy.
Front. Neural Circuits 7, 65 (2013).
24. Murphy, K. et al. Evaluation of registration methods on thoracic
CT: the EMPIRE10 challenge. IEEE Trans. Med. Imaging 30,
1901–1920 (2011).
25. Yushkevich, P.A. et al. User-guided 3D active contour segmentation of
anatomical structures: significantly improved efficiency and reliability.
Neuroimage 31, 1116–1128 (2006).
26. Eguchi, M. & Yamaguchi, S. In vivo and in vitro visualization of gene
expression dynamics over extensive areas of the brain. Neuroimage 44,
1274–1283 (2009).
27. Prewitt, J.M.S. Object enhancement and extraction. In Picture Processing
and Psychopictorics (eds. Lipkin, B.S. & Rosenfeld, A.) 75–149 (Academic
Press, 1970).
28. Jenkinson, M., Beckmann, C.F., Behrens, T.E., Woolrich, M.W. & Smith, S.M.
Fsl. Neuroimage 62, 782–790 (2012).
29. Lein, E.S. et al. Genome-wide atlas of gene expression in the adult mouse
brain. Nature 445, 168–176 (2007).
30. Alnuami, A.A., Zeedi, B., Qadri, S.M. & Ashraf, S.S. Oxyradical-induced GFP
damage and loss of fluorescence. Int. J. Biol. Macromol. 43, 182–186
(2008).
31. Steinke, H. & Wolff, W. A modified Spalteholz technique with preservation
of the histology. Ann. Anat. 183, 91–95 (2001).
32. Faisal, A.A., White, J.A. & Laughlin, S.B. Ion-channel noise places limits
on the miniaturization of the brain’s wiring. Curr. Biol. 15, 1143–1149
(2005).
33. Richards, K.L. et al. Hippocampal volume and cell density changes in a
mouse model of human genetic epilepsy. Neurology 80, 1240–1246 (2013).
34. Feng, G.P. et al. Imaging neuronal subsets in transgenic mice expressing
multiple spectral variants of GFP. Neuron 28, 41–51 (2000).
35. Okabe, M., Ikawa, M., Kominami, K., Nakanishi, T. & Nishimune, Y. ‘Green
mice’ as a source of ubiquitous green cells. FEBS Lett. 407, 313–319
(1997).
36. Ertürk, A. et al. Three-dimensional imaging of the unsectioned adult
spinal cord to assess axon regeneration and glial responses after injury.
Nat. Med. 18, 166–171 (2012).
37. Soderblom, C. et al. 3D imaging of axons in transparent spinal
cords from rodents and nonhuman primates. eNeuro 2 doi:10.1523/
ENEURO.0001-15.2015 (2015).
38. Weber, T.G. et al. Apoptosis imaging for monitoring DR5 antibody
accumulation and pharmacodynamics in brain tumors noninvasively.
Cancer Res. 74, 1913–1923 (2014).
NATURE PROTOCOLS | VOL.10 NO.11 | 2015 | 1727