scale nn.2928
Post on 25-Feb-2018
223 Views
Preview:
TRANSCRIPT
-
7/25/2019 Scale Nn.2928
1/10
NATURE NEUROSCIENCE VOLUME 14 |NUMBER 11 |NOVEMBER 2011 1481
T E C H N I C A L R E P O R T S
1Brain Science Institute, RIKEN, Wako-city, Saitama, Japan. 2School of Life Science, Tokyo University of Pharmacy and Life Science, Hachioji, Tokyo, Japan. 3Life
Function and Dynamics, Exploratory Research for Advanced Technology, Japan Science and Technology Agency, Wako-city, Saitama, Japan. 4Tokyo Institute of
Technology, Meguro-ku, Tokyo, Japan. Correspondence should be addressed to A.M. (matsushi@brain.riken.jp).
Received 21 March; accepted 12 August; published online 30 August 2011; doi:10.1038/nn.2928
imaging is prevented by tissue opacity and light scattering. At present,
standard laser-scanning confocal microscopy (LSCM) penetrates only toa depth of ~150 mm below the brain surface. Two-photon excitation fluo-
rescence microscopy (TPEFM) offers improved depth, but it cannot, even
under ideal conditions, penetrate more than 500800mm from the brainsurface59. Thus, optical sectioning of intact tissue is believed to be insuf-ficient to image and reconstruct large brain projections and cell popula-
tions that are often several millimeters in scale and deep below the surface.Light scattering can be reduced by optical clearing, which aims to
increase tissue transparency to achieve refractive uniformity through-out the specimen and allow greater depth of imaging. Although this
approach is only applied to fixed specimens, it can in principle facilitateoptical sectioning and enable three-dimensional imaging and recon-
struction. Several clearing solutions have been described. The water-soluble reagent FocusClear has been used to treat insect brains1012.
Unfortunately, this commercial reagent is prohibitively expensive forlarger samples. Furthermore, because its contents are proprietary,
FocusClear cannot be optimized for different biological samples. BABB(a mixture of benzyl-alcohol and benzyl-benzoate) is another clearing
solution that has been used with ultramicroscopy13,14and TPEFM15toperform three-dimensional reconstructions in whole organs of mice
and fruit flies. In this method, a fixed sample is incubated in BABB afterdehydration with ethanol and hexane. However, the extent to which
this organic chemical quenches fluorescent proteins inside specimensremains unknown. In addition, a technique was reported for clearing
thick slabs of mouse cortex for TPEFM with an index-matched solu-tion of 60% sucrose16. However, this sucrose-based method confers onlymodest transparency on tissue samples.
We developed a clearing reagent called Scale that alleviates the majorlimitations of previously reported solutions. We found that Scale ren-
ders mouse brain and embryos transparent while completely preserv-
ing fluorescent signals from labeled cells. This combination allows theimaging of intact brain at a depth of several millimeters and large-scalereconstructions of neuronal populations and projections at subcellular
resolution. We demonstrated proof of principle by reconstructing net-works involving cortical, callosal, hippocampal and neurogenic popula-
tions. We also developed additional Scale reagents and protocol variantsfor specific experimental applications, and discuss potential future
applications for quantitative three-dimensional brain reconstructions.
Optical methods for viewing neuronal populations and
projections in the intact mammalian brain are needed, but
light scattering prevents imaging deep into brain structures.
We imaged fixed brain tissue using Scale, an aqueous reagent
that renders biological samples optically transparent butcompletely preserves fluorescent signals in the clarified
structures. In Scale-treated mouse brain, neurons labeled with
genetically encoded fluorescent proteins were visualized at
an unprecedented depth in millimeter-scale networks and at
subcellular resolution. The improved depth and scale of imaging
permitted comprehensive three-dimensional reconstructions
of cortical, callosal and hippocampal projections whose extent
was limited only by the working distance of the objective lenses.
In the intact neurogenic niche of the dentate gyrus, Scale
allowed the quantitation of distances of neural stem cells to
blood vessels. Our findings suggest that the Scale method will
be useful for light microscopybased connectomics of cellular
networks in brain and other tissues.
An important challenge in biological research is the development of high-resolution optical methods to label and image cell populations in three
dimensions deep in intact tissue. For example, the ability to image andreconstruct intact neuronal networks would be valuable for understand-
ing structure-function relationships in the brain. Methods for labelingspecific cell types regardless of tissue depth and geometry have rapidly
progressed with the advent of genetically encoded fluorescent proteinsand transgenic marking methods such as Brainbow1. However, com-plementary optical techniques to image and reconstruct labeled three-
dimensional cell populations deep in intact tissue are also needed.Three-dimensional imaging of biological tissue typically involves
mechanical sectioning to improve axial resolution and access to deeper
structures. However, although this approach can have a high degree ofoptical resolution, promising current methods, such as array tomogra-phy and serial block-face scanning electron microscopy, are costly and
laborious, require sophisticated data reconstruction procedures and arecurrently limited to smaller tissue volumes24. In contrast, optical sec-
tioning provides a potentially fast, simple and inexpensive alternative forthree-dimensional reconstruction of fluorescently labeled structures at
subcellular resolution. However, the utility of optical sectioning for deep
Scale: a chemical approach for fluorescence imagingand reconstruction of transparent mouse brain
Hiroshi Hama1, Hiroshi Kurokawa1,2, Hiroyuki Kawano1,3, Ryoko Ando1, Tomomi Shimogori1,Hisayori Noda1,4, Kiyoko Fukami2, Asako Sakaue-Sawano1,3& Atsushi Miyawaki1,3
mailto:matsushi@brain.riken.jpmailto:matsushi@brain.riken.jphttp://www.nature.com/neuro -
7/25/2019 Scale Nn.2928
2/10
1482 VOLUME 14 |NUMBER 11 |NOVEMBER 2011 NATURE NEUROSCIENCE
T E C H N I C A L R E P O R T S
phosphate-buffered saline (PBS), 60% sucrose/PBS, FocusClear,
MountClear and ScaleA2. Transmission was measured in a spectro-photometer with water as a reference sample. The ScaleA2-treated slice was
substantially more permissive to visible and infrared light (350920 nm)than slices incubated in other reagents (Fig. 1b).
We next incubated intact fixed mouse brain in a Sca leA2 solution.Incubation for >2 weeks substantially cleared brain tissue; transparency
was evident against a patterned background (Fig. 1c) or by penetration
with a 532-nm laser light (Fig. 1d). The tissue-clearing effect of ScaleA2was also prominent on whole mouse embryos (Fig. 1e).
To estimate the extent of brain expansion, we calculated brain volume
by liquid displacement before and after ScaleA2 treatment by slowlylowering samples into a graduated cylinder containing water. The mean
sample volume was doubled (197 13%, n= 5) by ScaleA2 treatment.Taking the cube root of 1.97, we assume that ScaleA2 causes a 1.25-fold
linear expansion. To characterize the expansion of macroscopic struc-tures in two dimensions, we time-lapse imaged fixed brain slices (1 mm
thick) of a transgenic mouse line, thy1-YFPline H (YFP-H)17,18duringa 5-d incubation in ScaleA2 solution (Fig. 1fyand Supplementary
Fig. 2). We traced the outlines of the slices and several internal struc-tures in transmitted-light bright-field and fluorescence images. Overlay
RESULTS
Development and properties of the Scale reagent
The discovery of the Scale reagent was based on a serendipitous observa-
tion. We found that polyvinylidene fluoride membranes became transpar-ent when soaked in 4 M urea, which promotes the hydration of biological
samples (Supplementary Fig. 1). This result inspired us to search for anoptimal reagent to clear fixed biological samples for light microscopy.
We first treated mouse brain sections (60 mm thick) fixed with 4% para-
formaldehyde (PFA) with solutions containing 18 M urea. After 48 h,sections treated with 48 M urea became transparent along with someexpansion. To further optimize tissue clearance, we next combined urea-
containing solutions with other ingredients. The most effective solution,which we named ScaleA2, was composed of 4 M urea, 10% (wt/vol) glyc-
erol and 0.1% (wt/vol) Triton X-100. Glycerol was predicted to preventexcess hydration and minimize tissue expansion. ScaleA2 has a pH of 7.7
and refractive indices of 1.382, 1.387 and 1.380 at 589, 486 and 656 nm,respectively. ScaleA2 is colorless; the solution absorbs light at 276 nm but
is permissive to light greater than 300 nm (Fig. 1a).To examine tissue transparency quantitatively, we measured transmis-
sion in brain slices. We prepared 1.5-mm-thick slices from mouse brainsamples that had been fixed and incubated in water-soluble reagents:
Figure 1 Tissue clearing performance of ScaleA2. (a) Transmission curves of ScaleA2 (blue), 60% sucrose/PBS (green), FocusClear (yellow) and MountClear
(magenta). (b) Transmission curves of fixed brain slices (1.5 mm thick) in ScaleA2 (blue), 60% sucrose/PBS (green), Focus/MountClear (magenta, a slice treated
with FocusClear was placed in MountClear) and PBS (violet) after treatment with the respective solutions. (c,d) A whole fixed and cleared brain of a mouse
(P15) after treatment with ScaleA2 for 2 weeks. (c) A photo was taken with a black and white pattern as background. (d) The green light from a 1-mW, 532-nm
laser beam pointer traversed the cleared brain. (e) A photo of two embryos (E13.5) taken with a black and white pattern as background. Left,embryo placed in
PBS after fixation with 4% PFA. Right,embryo incubated in ScaleA2 solution for 2 weeks after fixation with 4% PFA. (fy) Characterization of the expansion of
macroscopic structures in fixed brain slices of a YFP-H mouse during ScaleA2 treatment. A coronal slice (1 mm thick) containing the hippocampus was prepared
from a 9-week-old mouse. The slice was split into two halves and the right half was incubated in ScaleA2 solution for 5 d while the left half was incubated in
PBS. Before (0 d, fi) and 1 d (jm), 2 d (nq) or 5 d (ru) after these incubations, the pair of slices on a coverslip with a patterned background were imaged
using a fluorescence stereomicroscope for transmission (f,i,j,m,n,q,rand u) and YFP fluorescence (g,h,k,l,o,p,sand t). The slice became transparent
and expanded after a 12-d incubation in ScaleA2 solution (l,m,p,q,tand u). The extent of the linear expansion was calculated as 1.28. Ag, amygdala; Cp,
cerebral peduncle (basal part); Cx, cortex; Dmn, dorsomedial nucleus; Hf, hippocampal formation; Pmc, posteromedial cortical amygdala nucleus. The outlines
of the slices and their internal structures at 0 d and 5 d were drawn with blue and orange, respectively. The outlines of the PBS-treated slice at 0 d and 5 d
overlapped substantially (v). Reduced drawings of the outlines of the ScaleA2-treated slice at 5 d also overlapped with the outlines at 0 d extensively (w). In
addition, the outlines of the ScaleA2-treated half (green) at 0 d were inverted and overlaid to the outlines of the PBS-treated half (magenta) at 0 d. As the brain
slice had been split slightly asymmetrically, the edges of each half were not precisely even, but proper alignment was achieved (x). A similar overlay was done
between the size-normalized outlines at 5 d (y). In xand y, the difference between green and red traces indicates the inherent baseline left/right asymmetry of
the slice. Notably, the degree and distribution of the asymmetry are almost identical between xand y. All scale bars represent 5 mm.
PBS
a
Wavelength (nm)
100
80
60
40
20
0Transmittance(%)
ScaleA2
ScaleA2
ScaleA2PBS
b
Before
1
d
2
d
5
d
Transmission TransmissionYFP YFP
Before5 d
Hf
Ag
Cp
Cx
Dmn
Pmc
f g h i
r s t u
n o p q
j k l m
v w yx
PBSScaleA2(inverted)
60
40
20
0Transmittance(%)
800700600500400300
Wavelength (nm)
60% sucrose/PBSFocusClearMountClear
ScaleA260% sucrose/PBSFocus/MountClearPBS
c d
e900800700600500400300
900
-
7/25/2019 Scale Nn.2928
3/10
NATURE NEUROSCIENCE VOLUME 14 |NUMBER 11 |NOVEMBER 2011 1483
T E C H N I C A L R E P O R T S
and several GFP-like proteins was resistant to 4 M urea at pH 7.7(Supplementary Fig. 4a,b).
To examine the in vivostability of EGFP fluorescence in ScaleA2, wetransfected the protein into cultured HeLa cells. After fixation, fluores-
cent cells were time-lapse imaged in a ScaleA2 solution and no substantialdecrease in fluorescence intensity was observed (Fig. 2a). In contrast,
when cells were incubated in BABB following dehydration with ethanol
and hexane13,14, EGFP fluorescence diminished over time (Fig. 2b).Next, we applied ScaleA2 to fixed brains of the YFP-H line17,18, in
which yellow fluorescent protein (YFP) is expressed in a subpopulation
of neurons confined to pyramidal neurons in the hippocampus andneurons with somata in layers V and VI in the cerebral cortex. The
whole fixed brain showed homogeneous fluorescence under blue light.
comparisons between Scaled and control samples revealed that the slice
outlines and relative positions of internal structures (hippocampus,amygdala and white matter) maintained their overall shape and pro-
portions, indicating that tissue expansion was isotropic and homog-enous. Similar experiments were performed using 50-mm-thick brain
slices of another transgenic mouse line, thy1-GFPline M (GFP-M)17
(Supplementary Fig. 3). Although ScaleA2-treated samples were typi-
cally soft and fragile, macroscopic structures that maintained their pro-portions in slice samples lacking full connective tension should showeven better preservation of topology in intact, unsectioned whole brains.
Fluorescence imaging with the Scale reagent
A critical question for brain clearing solutions is whether the capabil-ity for fluorescence imaging is retained. It was reported that wild-type
Aequorea green fluorescent protein (GFP)is sensitive to 8 M urea at acidic pH but not
at neutral or alkaline pH19. We verified thatthe fluorescence of enhanced GFP (EGFP)
60
ScaleA2
Time (min)50403020100 6050403020100
1.0
0.5
0.0
1.0
0.5
0.0
Normalized
fluorescenceintensity
Time (min)
50% ethanol
80% ethanol
100% ethanol
Hexane
BABB
thy1-YFPline H
BABB
a b c
ScaleA2
Figure 2 Comparison of ScaleA2 with BABB.
(a,b) Sensitivity of EGFP fluorescence to ScaleA2
solution and a conventional chemical clearing
reagent (BABB). Cultured HeLa cells expressing
EGFP were fixed with 4% PFA and were time-
lapse imaged while being exposed to ScaleA2
solution (a) or BABB following dehydration with
ethanol and hexane (b). Replacement of Hanks
Balanced Salt Solution with ScaleA2 resulted
in a change in focus and a slight decrease influorescence intensity. (c) Fluorescence images
comparing the preservation of YFP signals
between aqueous (left) and chemical (right) clearing agents. The brain of a YFP-H mouse (7 weeks old) was split into two halves. The left half was treated
with ScaleA2 for 3 d. The right half was treated with BABB after dehydration. Then slices (1 mm thick) were prepared and imaged for fluorescence with a
stereomicroscope. The original shape of the fixed brain is drawn with broken lines. Scale bar represents 5 mm.
2P 1P
2.0mm
(1.6mm)
d
e
f
i
j
k
g h
Pial surface
V
CA1White matter
VI
2.0 mm(1.6 mm)
2.0 mm
xy
z
920-nm laser
thy1-YFP line H
(3 weeks old)
a b c
Hippocampus
Cerebralcortex
Pial surface
VVI
CA1
DG
White matter
IV
thy1-YFPline H
(13 weeks old)
line
Objective
lens
xy
z
920-nm laser
4.0mm
l m
3.3mm
DG
CA3
CA1
White matter
Hilus
GCLML
on
4.0 mm(3.2 mm)
Excised hippocampusof thy1-YFP line H
(13 weeks old)
Objective
lens
920-nm laser
z
xy
Caudal
Anterior
Posterior
Dorsal
Rostral
(2.6mm)
Objective
lens
Cerebralcortex
Hippocampus
Figure 3 Three-dimensional reconstructions ofYFP-expressing neurons in ScaleA2-treated brain
samples of YFP-H mice. The actual imaging
depth is shown in parentheses. Unsectioned
brains (am) and an excised hippocampus (n,o)
were imaged. (ac) TPEFM imaging using a
25 objective (XLPLN25XWMP, numerical
aperture (NA) = 1.05, working distance =
2.0 mm). The experimental setup for TPEFM
imaging using the commercially available
objective is shown in a. A three-dimensional
reconstruction of YFP-expressing neurons in 16
(8 2) quadratic prisms located in the cerebral
cortex and hippocampus is shown in b.
A high-magnification xyimage at a depth of
0.9 mm (a yellow box in b) is shown in c.
(dk) Three-dimensional reconstruction ofYFP-expressing neurons in a quadratic prism
located in the cerebral cortex. The same brain
region was imaged using a 20 objective
(W-PlanApochromat, NA = 1.0, working distance
= 2.0 mm) and taking both two-photon (920-nm excitation, dg) and one-
photon (514-nm excitation, hk) approaches. For each volume rendering,
three xyimages at different z positions (dfand ik) are presented.
(lo) TPEFM imaging using a custom-designed objective with a working
distance of 4.0 mm. The experimental setup for TPEFM imaging using the
objective lens is shown in l and n. Three-dimensional reconstructions of
YFP-expressing neurons in a quadratic prism located in the cerebral cortex
and hippocampus (m) and in 24 (4 6) quadratic prisms located in the
excised hippocampus (o) are shown. DG, dentate gyrus; GCL, granule cell
layer; ML, molecular layer. All scale bars represent 50 mm.
-
7/25/2019 Scale Nn.2928
4/10
1484 VOLUME 14 |NUMBER 11 |NOVEMBER 2011 NATURE NEUROSCIENCE
T E C H N I C A L R E P O R T S
(Fig. 3df). The imaging depth was sufficient
to reach the dorsal tip of the hippocampus(Fig. 3g). We then imaged the same region with
a 514-nm excitation and an internal descannedpath detector through a confocal pinhole (~2
Airy disks; Supplementary Fig. 6b). This one-photon excitation imaging setup produced
sufficiently bright images at depths permittedby the objectives working distance (~2.0 mm;
Fig. 3hk), suggesting that very little scatteringoccurred inside the specimen. In deep regions,
however, TPEFM yielded a better signal-to-noise ratio than LSCM.
A similar three-dimensional reconstruc-tion was performed by TPEFM using an older
YFP-H mouse (13 weeks old; SupplementaryVideo 1). Axons that traveled horizontally
through the dendritic trees of the cortex wereidentified and individual axons tunneling
inside white matter was also discernable.
Beyond the current imaging depth limitWith brains cleared by ScaleA2, the imaging depth limit was deter-
mined by the working distances of currently available objective lenses.Among commercially available lenses, for example, a 40 objective
(LUMPLFLN40XW, NA = 0.80, working distance = 3.3 mm) has alonger working distance but a lower NA. Although the use of this lens
permitted an imaging depth of 3 mm (Supplementary Fig. 7a,b), theresolution was lower. We therefore asked the manufacturer (Olympus)
to develop a customized 25 objective lens, which has a longer work-ing distance (4 mm) and a sufficiently high NA (1.0). Using this lens,
we were able to generate very long quadratic prisms of the YFP-H linebrain, with reconstructions that extended from the brain surface to the
dentate gyrus (Fig. 3l,mand Supplementary Video 2).
In addition to reaching new depth limits for brain reconstruction offluorescent neurons, we were also able to optically reconstruct exten-sive neuronal networks. The hippocampal formation was excised from
a fixed and cleared YFP-H brain to permit a comprehensive three-dimensional reconstruction of the hippocampus containing the den-
tate gyrus and Ammons horn fields (Fig. 3n,o). Fine structures in thesame excised hippocampal preparation were visualized by increasing
the photomultiplier tube sensitivity (Supplementary Fig. 7ce).
Surveying commissural axons in the intact mouse brain
To determine whether the Scale system allows a comprehensive per-
spective of specific axon projections, we used a macro-confocal micros-copy system (AZ-C1, Nikon) and imaged optically cleared brains in
which specific neurons had been fluorescently labeled. We focused on
axons in the corpus callosum, which connects the left and right cere-bral hemispheres21,22. Tracing their axon bundles across the midlinerequires laborious procedures that produce a large number of sections.
To visualize callosal axon projections of layer II/III pyramidal neu-rons, we electroporated plasmids encoding EYFP in utero into the
dorsal ventricular zone of the mouse forebrain at embryonic day 15.5(E15.5)22,23. The embryos were raised until postnatal day 10 (P10), when
callosal connnections are almost fully established21,22, and their brainswere removed and fixed. After incubation in ScaleA2 for 1 week, the
entire EYFP signal in the brain sample was collected using a 1 objective(AZ-PlanApo, NA = 0.1, working distance = 35 mm). The z stack image
(Fig. 4a) mapped the labeled pyramidal neurons in layers II and III in theipsilateral (right) hemisphere and provided sweeping views of fluorescent
We then cut it at midplane and incubated the left half in Sca leA2 for3 d while dehydrating the right half with ethanol and treating it withBABB13,14. Yellow fluorescence was preserved in the ScaleA2-treated
half but not in the half exposed to ethanol and BABB (Fig. 2c). In addi-tion, the ScaleA2-treated half expanded, whereas the BABB-treated half
shrunk. A similar comparison with consistent results was made usinga fixed brain sample from the GFP-M line17(Supplementary Fig. 5).
Three-dimensional reconstruction of neuronal structures
We examined the three-dimensional architecture of neuronal networkscomprised of fluorescent neurons from the fixed and cleared intact brain
of a YFP-H mouse using TPEFM with 920-nm excitation (Fig. 3a). Witha TPEFM system (Olympus FV1000MPE) employing a 25 objective
(XLPLN25XWMP, numerical aperture (NA) = 1.05, working distance =2.0 mm) and correction collar, an imaging depth of 2 mm was achieved
(Fig. 3b). The three-dimensional reconstruction extended from the cere-bral cortex to the dorsal tip of the CA1 region through the white mat-
ter (corpus callosum). Cortical layer V/VI pyramidal neurons and theirdendritic networks were well resolved; individual dendritic spines were
discernable in an expanded view at a depth of 0.9 mm (Fig. 3c). Given theestimated 1.25-fold expansion along one axis, the imaging depth in real
tissue can be obtained by multiplying the measured depth value by 0.8.In these experiments, imaging at an xyposition produced a data
unit in the shape of a long quadratic prism and three-dimensionalreconstructions were extended in the xyplane. To reconstruct corti-
cal networks, multiple units of data were generated at neighboringxypositions and combined using the microscope systems tiling
software (Fig. 3b).
One- and two-photon microscopy of Scaled brain
One-photon excitation fluorescence microscopy should benefit from
tissue clearing to an even greater extent than TPEFM. To illustrate this,we employed LSCM to image a ScaleA2-treated brain sample. We exam-
ined the three-dimensional structures of a 3-week-old YFP-H mouseusing a ZEISS LSM710-NLO system equipped with a 20 objective
(W-PlanApochromat, NA = 1.0, working distance = 2.0 mm). With 920-nmexcitation and an external non-descanned detector (Supplementary
Fig. 6a), which is common to TPEFM20, three-dimensional reconstruc-
tion of the cortex was achieved to a depth of 2.0 mm. The high resolutionof the reconstruction was demonstrated by xyimages at various depths
R L
CC
CxII / III ML
LVLV
Cx
CPuCPu
aCx
HC
TM
LbCx
HC
Lc
Figure 4 Visualization of labeled callosal connections in the intact mouse brain. A population of
layer II/III pyramidal neurons was labeled by in uteroelectroporation of plasmids encoding EYFP into
the dorsal ventricular zone on the right side (R) of the mouse forebrain at E15.5, and their axonal
projections into the left side (L) were visualized at P10 using a macro-zoom confocal microscope
after fixation and a 7-d treatment with ScaleA2. CC, corpus callosum; CPu, caudate putamen; Cx,
cortex; HC, hippocampus; LV, lateral ventricle; ML, midline; TM, thalamus. (a) We acquired 18
confocal images (52-mm steps) using a 1 objective lens at scanner zoom 3, and z stacked them.
(b) We acquired 17 confocal images (43-mm steps) using a 2 objective lens at scanner zoom 2,
and z stacked them. (c) We acquired 34 confocal images (10.8 mm steps) using a 2 objective lens
at scanner zoom 4, and z stacked them. All scale bars represent 500 mm.
-
7/25/2019 Scale Nn.2928
5/10
NATURE NEUROSCIENCE VOLUME 14 |NUMBER 11 |NOVEMBER 2011 1485
T E C H N I C A L R E P O R T S
brain of an adult transgenic mouse (7 weeks old) expressing GFP under
the control of the Nestingene regulatory region26. To fluorescently labelblood vessels, we transcardially perfused the entire vasculature with
Texas Redlabeled lectin. After subsequent fixation, the hippocampuswas dissected out and treated with ScaleA2 for 2 d for TPEFM with
920-nm excitation (Fig. 5a). A relatively large number of NSCs werefound exclusively in the SGZ and a majority were found to possess radial
glia-like processes, whereas others had plump, short processes (Fig. 5b)
that together probably represent type-1 and type-2a cells, respectively.Both cell types made considerable contact with blood vessels throughtheir processes. We extended the three-dimensional reconstructions of
GFP-filled NSCs and Texas Redlabeled blood vessels inside the hip-pocampus (Fig. 5c). A perspective image from the front of the hippo-
campus revealed that the density of NSCs varied substantially alongthe longitudinal axis of the hippocampus. This observation serves as a
reminder that the number of NSCs in a single section does not neces-sarily reflect hippocampal neurogenic activity.
To observe directly proliferating NSCs (PNSCs), we employed theFucci (fluorescent ubiquitination-based cell cycle indicator) tech-
nique27. The Fucci probe consists of mKO2-hCdt1(30/120) and mAG-hGem(1/110), which label nuclei of G1/G0phase cells red and those in
callosal axons traveling into the contralateral (left) hemisphere. A series
of confocal images from anterior to posterior (Supplementary Video 3)revealed that the commissural axons labeled at E15.5 appeared more dor-
sally in the posterior images. Using a 2 objective (AZ-PlanFluor, NA =0.2, working distance = 45 mm), we then zoomed in on the projections of
the labeled callosal axons in the primary somatosensory cortex (Fig. 4b,c).
Neural stem cell association with blood vessels in the SGZ
We explored the utility of Scale for imaging of discrete cell populationsand quantitative measurement of their geometric properties. We chose tostudy the vascular niche for neural stem cells (NSCs) in the subgranular
zone (SGZ) of the dentate gyrus in the hippocampus. Although BrdUlabeling of mechanical z sections from the subventricular zone (SVZ)24,25
clearly shows that NSCs and blood vessels are proximal, the SGZ is lessamenable to the type of within-zsection measurements employed for the
SVZ as a result of its more complex vascular geometry with many ran-domly oriented branches. Thus, we felt that the Scale method, by clearing
the SGZ for optical rather than mechanical sectioning, could addressthis question, which has been difficult to study with current approaches.
To first gain a comprehensive three-dimensional perspective of howNSCs interface with blood vessels in the SGZ, we imaged the Scale-treated
d
f
Dentate gyrus
CA3GCLSGZ
SGZGCL
hilus
Objective
lens
920-nm laser
1.9 mm
(1.5 mm)
g500
400
300
200
100
0
0
10
20
30
40
PNSC nuclei
Total nuclei
PNSC nuclei
Total nuclei
Distance (m)
0 10(8)
20(16)
30(24)
40(32)
50(40)
Distance (m)
0 10(8)
20(16)
30(24)
40(32)
50(40)
h
Percentage
ofnuclei
PNSC nucleusBlood vessel
1.9 mm
(1.5 mm)
PNSC nucleus Blood vessel
Distance measurementby RINZO
Rostral
Caudal
Dentate gyrus
Excisedhippocampus
D
V
R
CF
NSC Blood vessel
D
C RF
V
GCLML
SGZ
Hilus
GCLML
SGZHilus
MLSGZ
GCL
GCL
ML
SGZ
Hilus
SGZ
GCL
NSC Blood vessela
b
c e
Numberofn
uclei
Figure 5 Quantitation of the distances between proliferating neural stem
cell (PNSC) nuclei and blood vessels in the subgranular zone (SGZ) of
adult mice. (ac) Visualization of GFP-labeled neural stem cells (NSCs)
and Texas Redlabeled blood vessels in the adult mouse hippocampus.
A schematic diagram showing the approach of TPEFM imaging (red arrow)
to a cleared excised hippocampus is presented in a. The imaged area isshown by six quadratic prisms. A high-magnification volume rendering of
NSCs (green) and blood vessels (red) in the SGZ is shown in b. Volume
renderings generated from a large region in the hippocampus are shown
in c. Five perspective views were created from different angles (D, dorsal;
V, ventral; C, caudal; R, rostral; F, front). GCL, granule cell layer; ML,
molecular layer. (dh) Hippocampi were excised from the fixed brains of
#504 mice (7 weeks old) and cleared with Sca leA2 for 2 d. An excised
hippocampus for TPEFM imaging is shown in d. The quadratic prism that
was approached from the surface (red arrow) is shown. The objective was
placed so that the z axis came into contact with the apex of the hilus.
A series of perspective images of PNSC nuclei (green) and blood vessels
(red) when tunneling into the hippocampus are shown in e. Backward
perspective images were created at different depths. After passing through
the SGZ, no PNSC nuclei were seen ahead. These images are animated
in Supplementary Video 4. RINZO automatically calculated the distance(white lines) from each PNSC nucleus (green) to the nearest blood vessel
(red) surface (f). Histograms show the distribution of distances to blood vessels for all SGZ cell nuclei (violet), and for PNSC nuclei (green). Cell numbers (g)
or their frequencies (h) were plotted. The real distance is shown in parentheses. Scale bars represent 500 mm (b,c) and 20 mm (e,f).
-
7/25/2019 Scale Nn.2928
6/10
1486 VOLUME 14 |NUMBER 11 |NOVEMBER 2011 NATURE NEUROSCIENCE
T E C H N I C A L R E P O R T S
Scale reagent variants for targeted applications
ScaleA2 is an optimal all-purpose solution for most brain clearing
experiments. Nevertheless, we optimized other Sca le variants that
could provide better control of certain properties of cleared tissue forspecific applications.
An ideal clearing agent will preserve the volume of tissue for more
accurate reconstruction. However, BABB and FocusClear make biologi-cal samples shrink, while ScaleA2 causes expansion. Thus, we devel-
oped a modified solution, ScaleU2, composed of 4 M urea, 30% glyceroland 0.1% Triton X-100. ScaleU2 requires longer incubation times to
achieve clearing, but it reduces tissue expansion and the fragility ofcleared brain samples and is therefore suitable for soft samples such
as mouse embryos. To assess the utility of ScaleU2, we analyzed cell-cycle profiles during mouse embryogenesis in a transgenic mouse line
(#596/#504) that expresses the Fucci probe in which almost every cellnucleus exhibits either red (G1/G0) or green (S/G2/M) fluorescence
27.
We incubated fixed embryos (E11.5 and E13.5) in ScaleU2 for 6 monthsand performed three-dimensional imaging of the left half of each body
using a Nikon AZ-C1 system (Fig. 6a,b). At both E11.5 and E13.5,the developing cerebral cortex contained more green nuclei than the
S/G2/M phase green, respectively. A transgenic mouse line that almostubiquitously expresses mAG-hGem(1/110), #504, was used previously
to obtain in vivoinformation about proliferation patterns at various
embryonic stages27.We labeled the vasculature of a #504 mouse with Texas Redlabeled
lectin. After fixation, the hippocampus was dissected out and treated
with ScaleA2 for 2 d. The cleared hippocampus was viewed with TPEFMat 920-nm excitation from its ventral surface (Fig. 5d) and a series of
perspective images aimed toward the dorsal side were taken (Fig. 5e).Green nuclei of PNSCs were localized exclusively in the SGZ in apparent
association with a network of blood vessels (Supplementary Video 4).We next examined the quantitative relationship between PNSCs of
the SGZ lineage and the vasculature. Multiple hippocampi were fixedand cleared by ScaleA2 and treated with 4,6-diamidino-2-phenylindole
(DAPI) to label nuclei. After performing TPEFM imaging on these prepa-rations, we made three-dimensional reconstructions that localized PNSC
nuclei and all nuclei in relation to blood vessels. The distances of the nucleito the nearest vessels were measured in three-dimensional space using the
RINZO algorithm (Fig. 5f), which we developed for this purpose.We analyzed 96 PNSC nuclei, out of 1,381 DAPI-positive nuclei in
the SGZ (Fig. 5g,h). Of the PNSC nuclei, 37% (35 of 96) were situatedwithin 10 mm of blood vessel surfaces. In contrast, only 22% of SGZ
nuclei (298 of 1,381) were within 10 m of blood vessels; these nucleimostly belonged to endothelial cells and pericytes, as well as PNSCs.
The average distance of PNSC nuclei to blood vessels (16.38 12.10 mm)was significantly closer (P< 0.05) than the average for all SGZ cell nuclei
(20.34 11.48 mm). Because these analyses involved automatic process-ing (except for manual tagging of PNSC nuclei) and were performed
comprehensively in three-dimensional space, we conclude that PNSCnuclei are closely associated with blood vessels in the mouse SGZ.
Figure 6 Three-dimensional reconstructions of Fucci transgenic mouse
embryos treated with ScaleU2. (ac) Green and red signals are derived from
the Fucci-S/G2/M marker and Fucci-G1/G0marker, respectively. Transgenic
mouse #596/#504 embryos (E11.5 and E13.5) were fixed with 4% PFA/
PBS and then incubated in ScaleU2 for 6 months. The right halves of their
bodies (heads) were imaged using macro-zoom LSCM (AZ-C1) equipped with
a 2 objective lens (AZ-PlanFluor, NA = 0.2, working distance = 45 mm).
The z step size was 5 mm. We used 488-nm and 561-nm laser diodes.
Shown are maximum intensity projection (MIP) images at E11.5 (a) and
E13.5 (b). A confocal image of the region indicated by a white box in the
MIP image (b) is shown in c. (di) Immunohistochemical localization of
Nestin (df) or PSA-NCAM (gi) on sections of the posterior end of the
diencephalon of an E13.5 #504 transgenic embryo producing mAG-
hGem(1/110). The immunostaining and mAG-hGem(1/110) signals are
shown in white and green, respectively. High-magnification images of
the regions indicated by yellow boxes in eand hare shown in f and i,
respectively. IC, inferior colliculus; V, ventricle. Scale bars represent
1 mm (ac) and 100 mm (di).
E13.5
baE11.5
c
Fucci-S/G2/M
Fucci-S/G2/M
Fucci-S/G2/M
Fucci-G1
NestinV
ICVIC SC
PSA-NCAMV
ICV
IC SC
E13.5
E13.5
d e
g h i
f
E13.5
GluR1 Synaptophysin Merge
YFP PSA-NCAM Merge
SGZ
GCL
Hilus
SGZ
GCL
Hilus
Fixed
Scaled
restored
Fixed
Scaled
restored
Fixed
a b c
d e f
g h i
j k l
Fixed
Figure 7 Immunohistochemistry on sections restored from ScaleA2.
(af) A brain sample from the thy1-YFPmouse line H (7 weeks old) was
used. Sections of the dentate gyrus were prepared from a fixed sample (ac)
and a sample restored from ScaleA2 (df). Samples were stained with a
mouse monoclonal antibody to PSA-NCAM. The YFP fluorescence and
immunoreactivity for PSA-NCAM (with a secondary antibody conjugated to
Alexa Fluor 546) were visualized. (gl) A brain sample from wild-type mouse
(7 weeks old) was used. Sections of the CA3 region were prepared from a
fixed sample (gi) and a sample restored from ScaleA2 (jl). Samples were
stained with a rabbit polyclonal antibody to GluR1 and a mouse monoclonal
antibody to synaptophysin. These primary antibodies were visualized with
secondary antibodies conjugated to Alexa Fluor 488 and 546, respectively
(Molecular Probes). Scale bars represent 20 mm.
-
7/25/2019 Scale Nn.2928
7/10
NATURE NEUROSCIENCE VOLUME 14 |NUMBER 11 |NOVEMBER 2011 1487
T E C H N I C A L R E P O R T S
Our results indicate that ScaleA2 is superior to other described
clearing agents in key properties necessary for performing detailedthree-dimensional brain reconstructions at both fine scale and broad
perspective (Supplementary Table 1). Unlike organic solventbasedreagents, Scale allows signal preservation of fluorescent proteins. Scale
is also superior to other aqueous reagents, including FocusClear, becauseit is inexpensive and its formula is simple and public, and to 60% sucrose,
which allows only modest transparency. Thus, a large quantity of Scalecan be used for clearing large tissue samples not only from rodent brains
and embryos, but, we expect, from primate28and human biopsy samples.Furthermore, researchers can modify the reagents composition accord-
ing to the nature of the samples to be cleared. For example, the concen-tration of glycerol should be increased to preserve tissue volume and
integrity, and the concentration of urea can be reduced for brain samplesfrom fish and flies (data not shown). The drawbacks of ScaleA2 include
the long time periods required for clearance and the fragility of clearedsamples, but these problems may be largely solved by modifying reagent
composition to ScaleB4 and ScaleU2, respectively.
Scale applications for brain development and function
Recent progress in gene targeting and genetic labeling methods has
revealed the roles of specific genes in regulating axon guidance duringthe formation of topographic projections. The Scale method will facili-tate the analysis of wild-type projection and screening of mutant mice
to detect those with aberrant axon projections. Moreover, the high reso-lution of Scale may allow mechanistic insights into the nature of such
defects. To follow up on our callosal findings, one could label axonswith different colors at different embryonic days to survey the develop-
mental structure of the corpus callosum29. The Scale technique can alsobe applied effectively to chemically stained samples, as we found with
transcardial perfusion of Texas Redlectin to label blood vessels beforefixation or in the future with in vivolabeling of astrocytes with Texas
Redhydrazide, a paraformaldehyde-fixable analog of SR101 (ref. 30),and neuronal tracing with NeuroVue dyes in fixed brain tissue31.
Further applications of the Scale system could involve studies of neu-ronal function. We labeled and reconstructed the proliferating NSC
niche in the SGZ, which was made possible by Scale clearance of thehippocampus. It will also be useful to obtain projection images of Scaled
brain samples using low-NA objective lenses to allow quantificationof the spatial expression of neural activityregulated immediate early
genes during behavior of transgenic mice in which the expression ofGFP is controlled by the promoter of c-fos32or Arc33. Scale will thusbe valuable for the high-resolution immediate early gene mapping of
behavior in intact large-scale brain networks.
Scale applications for connectomics in intact brain
Scale should be compatible with most light microscopy systems. In par-
ticular, we suggest the utility of Scale-treated samples for light sheet
ultramicroscopy, a technique that can gather three-dimensional imagedata quickly13,14,34,35. In addition, in combination with Scale, it may bepossible to observe larger animals via fluorescent protein tomography
(FPT)36using proteins that fluoresce at shorter wavelengths. Similarly, theScale technique should be applicable to all organs. Very few techniques
for three-dimensional reconstruction using light microscopy can pen-etrate tissue blocks thicker than 1 mm. In contrast, most tomographic
techniques, including optical projection tomography37, FPT36, computertomography and positron emission tomography, as well as magnetic
resonance imaging, are capable of analyzing structural and quantitativefeatures in a much larger mass of tissue, such as the whole body.
Among the emerging approaches for brain connectomics that aimat detailed three-dimensional reconstructions of brain structures, the
diencephalon, suggesting that cells in the diencephalon undergo cell-
cycle exit for differentiation before those in the cerebrum. Althoughneurogenesis in the diencephalon appeared to have reached completion
by E13.5, there was a region rich in green nuclei at the posterior endof the diencephalon (inferior colliculus) (Fig. 6b). Local cell prolifera-
tion was also evident in a sectional image along the midline (Fig. 6c).The green nucleuscontaining cells were immunostained for Nestin
(Fig. 6d
f), but not for PSA-NCAM (
Fig. 6g
i), which verified that theregion was still mitotic. Thus, the ScaleU2 technique applied to Fucci
transgenic mice provides a comprehensive perspective of proliferationversus differentiation in the developing brain.
ScaleA2 takes days or weeks to clear large specimens; ScaleU2 takesweeks or months. However, we found that the clearing process could
be accelerated by a transient tissue expansion induced by 8 M urea. Weexploited this observation to devise a speedy protocol that incorporates
a pulse of 8 M urea. In this protocol, fixed biological samples are incu-bated in ScaleA2 for 2 d, followed by ScaleB4 containing 8 M urea and
0.1% Triton X-100 for 2 d, and finally a ScaleA2 or ScaleU2 solution.As an added benefit, treating with ScaleB4 effectively depletes biologi-
cal samples of background signals. The pH of a ScaleB4 solution is 8.7;the stability of fluorescent proteins in 8 M urea (pH 8.7) was verified
(Supplementary Fig. 4a).It was not initially clear whether Scale treatment had an irreversible
effect on tissue architecture. We investigated whether a ScaleA2-treatedbrain sample could be restored to its original state by simply washing
with PBS. We split a fixed brain of the YFP-H mouse into two halves.The right half was kept in PBS while the left was cleared thoroughly with
ScaleA2 for 3 weeks and then washed in 20 volumes of PBS. After twosequential washes for 15 min each, the sample shrank to the original size
and became turbid. The two halves were embedded in optimal cuttingtemperature (OCT) compound, from which 30-mm-thick dentate gyrus
sections were prepared for histology. The patterns of immunostainingfor PSA-NCAM, as well as the cellular structures delineated with YFP
fluorescence, were very similar between the two samples ( Fig. 7af).
Next, we examined subcellular structures in the CA3 region using a wild-type mouse brain. When immunostained for the pre- and postsynapticmarkers Synaptophysin and GluR1, similar colocalization signals were
observed in sections from the two halves (Fig. 7gl). Similar results wereobtained with immunostaining for PSA-NCAM/BLBP (brain lipid bind-
ing protein) or VGLUT1 (vesicular glutamate transporter 1) using twoneighboring sections from a GFP-M mouse brain (Supplementary Figs. 8
and 9). It is therefore easy to restore a cleared specimen to its originalstate in a manner that is compatible with immunohistochemistry and
techniques such as array tomography2.
DISCUSSION
Comparison of Scale with other tissue-clearing techniques
With the Scale system, we present a simple, but effective, technique for
clearing mammalian brain tissue that has the potential to address ques-tions in brain structure and function at unprecedented spatial detail andscale. Our results demonstrate the utility of Scale for reconstruction of
neuronal populations and projections, in which labeled cells at subcel-lular detail can be viewed in three-dimensional networks at millimeter
scales. As Scale is part of an emerging area of chemical technology forimaging, we evaluated the strength and limitations of Scale, compared
its performance with that of other described clearing reagents and devel-oped solutions to extend its range of utility. Here we discuss its potential
roles among the emerging connectomics efforts to map the brain and itsconstituent structures. Given its simplicity and stability, we suggest that
Scale will popularize high-resolution three-dimensional reconstructionsin mammalian brain and other tissues, organs and animals.
-
7/25/2019 Scale Nn.2928
8/10
1488 VOLUME 14 |NUMBER 11 |NOVEMBER 2011 NATURE NEUROSCIENCE
T E C H N I C A L R E P O R T S
7. Zipfel, W.R., Williams, R.M. & Webb, W.W. Nonlinear magic: multiphoton microscopy
in the biosciences. Nat. Biotechnol.21, 13691377 (2003).
8. Helmchen, F. & Denk, W. Deep tissue two-photon microscopy. Nat. Methods2, 932
940 (2005).
9. Theer, P. & Denk, W. On the fundamental imaging-depth limit in two-photon micros-
copy. J. Opt. Soc. Am. A Opt. Image Sci. Vis. 23, 31393149 (2006).
10. Chiang, A.-S. et al.Insect NMDA receptors mediate juvenile hormone biosynthesis.
Proc. Natl. Acad. Sci. USA99, 3742 (2002).
11. Liu, Y.-C. & Chiang, A.-S. High-resolution confocal imaging and three-dimensional
rendering. Methods30, 8693 (2003).
12. Lin, H.-H., Lai, J.S.-Y., Chin, A.-L., Chen, Y.-C. & Chiang, A.-S. A map of olfactoryrepresentation in the Drosophilamushroom body. Cell128, 12051217 (2007).
13. Dodt, H.-U. et al.Ultramicroscopy: three-dimensional visualization of neuronal net-
works in the whole mouse brain. Nat. Methods4, 331336 (2007).
14. Jhrling, N., Becker, K. & Dodt, H.-U. 3D-reconstruction of blood vessels by ultrami-
croscopy. Organogenesis5, 145148 (2009).
15. Parra, S.G., Chia, T.H., Zinter, J.P. & Levene, M.J. Multiphoton microscopy of cleared
mouse organs. J. Biomed. Opt.15, 036017 (2010).
16. Tsai, P.S. et al.Correlations of neuronal and microvascular densities in murine cortex
revealed by direct counting and colocaliztion of nuclei and vessels. J. Neurosci.29,
1455314570 (2009).
17. Feng, G. et al. Imaging neuronal subsets in transgenic mice expressing multiple
spectral variants of GFP. Neuron28, 4151 (2000).
18. Porrero, C., Rubio-Garrido, P., Avendao, C. & Clasc, F. Mapping of fluorescent
proteinexpressing neurons and axon pathways in adult and developing Thy1-eYFP-H
transgenic mice. Brain Res.1345, 5972 (2010).
19. Alkaabi, K.M., Yafea, A. & Ashraf, S.S. Effect of pH on thermal- and chemical-induced
denaturation of GFP. Appl. Biochem. Biotechnol.126, 149156 (2005).
20. Denk, W., Piston, D.W. & Webb, W.W. Two-photon molecular excitation in laser scan-
ning microscopy. in The Handbook of Confocal Microscopy (ed.J. Pawley)445458(Plenum Press, New York, 1995).
21. Lindwall, C., Fothergill, T. & Richards, L.J. Commissure formation in the mammalian
forebrain. Curr. Opin. Neurobiol.17, 314 (2007).
22. Wang, C.-L. et al. Activity-dependent development of callosal projections in the
somatosensory cortex. J. Neurosci.27, 1133411342 (2007).
23. Shimogori, T. & Ogawa, M. Gene application with in uteroelectroporation in mouse
embryonic brain. Dev. Growth Differ.50, 499506 (2008).
24. Tavazoie, M. et al.A specialized vascular niche for adult neural stem cells. Cell Stem
Cell3, 279288 (2008).
25. Shen, Q. et al.Adult SVZ stem cells lie in a vascular niche: a quantitative analysis of
niche cell-cell interactions. Cell Stem Cell3, 289300 (2008).
26. Yamaguchi, M., Saito, H., Suzuki, M. & Mori, K. Visualization of neurogenesis in the
central nervous system using nestin promoterGFP transgenic mice. Neuroreport11,
19911996 (2000).
27. Sakaue-Sawano, A. et al.Visualizing spatiotemporal dynamics of multicellular cell-
cycle progression. Cell132, 487498 (2008).
28. Sasaki, E. et al.Generation of transgenic non-human primates with germline transmis-
sion. Nature459, 523527 (2009).29. Richards, L.J., Plachez, C. & Ren, T. Mechanisms regulating the development of the cor-
pus callosam and its agenesis in mouse and human. Clin. Genet.66, 276289 (2004).
30. Nimmerjahn, A., Kirchhoff, F., Kerr, J.N.D. & Helmchen, F. Sulforhodamine 101 as a
specific marker of astroglia in the neocortex in vivo.Nat. Methods1, 3137 (2004).
31. Fritzsch, B., Muirhead, K.A., Feng, F., Gray, B.D. & Ohlsson-Wilhelm, B.M. Diffusion
and imaging properties of three new lipophilic tracers, NeuroVue Maroon, NeuroVue
Red and NeuroVue Green and their use for double and triple labeling of neuronal
profile. Brain Res. Bull.66, 249258 (2005).
32. Barth, A.L., Gerkin, R.C. & Dean, K.L. Alteration of neuronal firing properties after in
vivoexperience in a FosGFP transgenic mouse. J. Neurosci.24, 64666475 (2004).
33. Wang, K.H. et al.In vivotwo-photon imaging reveals a role of Arc in enhancing orienta-
tion specificity in visual cortex. Cell126, 389402 (2006).
34. Voie, A.H., Burns, D.H. & Spelman, F.A. Orthogonal-plane fluorescence optical sec-
tioning: three-dimensional imaging of macroscopic biological specimens. J. Microsc.
170, 229236 (1993).
35. Huisken, J., Swoger, J., Del Bene, F., Wittbrodt, J. & Stelzer, E.H. Optical sectioning
deep inside live embryos by selective plane illumination microscopy. Science305,
10071009 (2004).
36. Zacharakis, G. et al.Volumetric tomography of fluorescent proteins through smallanimals in vivo. Proc. Natl. Acad. Sci. USA102, 1825218257 (2005).
37. Sharpe, J. et al.Optical projection tomography as a tool for 3D microscopy and gene
expression studies. Science296, 541545 (2002).
38. Sharpe, J. Optical projection tomography as a new tool for studying embryo anatomy.
J. Anat.202, 175181 (2003).
39. Lichtman, J.W. & Sanes, J.R. Ome sweet ome: what can the genome tell us about the
connectome? Curr. Opin. Neurobiol.18, 346353 (2008).
40. Lichtman, J.W. & Smith, S.J. Seeing circuits assemble. Neuron60, 441448 (2008).
41. Seung, H.S. Reading the book of memory: sparse sampling versus dense mapping of
connectomes. Neuron62, 1729 (2009).
Scale technique will likely provide a simple, inexpensive complement
to array tomography and serial section electron microscopy efforts byenlarging the volume and depth of three-dimensional reconstructions
that can be achieved using light microscopy data. In this way, the Scalesystem should help to bridge the imaging gap38between the size of
specimens that can be visualized with light microscopy versus otherbrain imaging techniques. The reversibility of Scale for retrospective
immunohistochemistry will enable zooming out to observe neuro-nal circuits in a global three-dimensional reconstruction, followed by
zooming in to image specific synaptic structures at both light andelectron microscopy scales. Together with advances in microscopy and
fluorescent proteins, Scale will contribute to the discovery of new bio-logical principles in connectomics and circuit genetics3941.
METHODS
Methods and any associated references are available in the online ver-
sion of the paper at http://www.nature.com/natureneuroscience/.
Note: Supplementary information is available on theNature Neurosciencewebsite.
ACKNOWLEDGMENTSWe thank H. Sakurai, H. Otsuka and M. Hirano for general assistance, F. Ishidate,
B. Zimmermann, R. Wolleschensky, Y. Watanabe, E. Nakasho, H. Kimura, T. Tajimaand S. Horie for help with acquiring and analyzing images, RIKEN BSI-OlympusCollaboration Center for technical support, Y. Yoshihara (RIKEN), M. Yamaguchiand K. Mori (The University of Tokyo) for the Nestin promoterGFP transgenicmice, J.R. Sanes (Harvard) for the YFP-H and GFP-M lines, E. Takahashi(RIKEN) for helpful advice on transgenic mice, S. J. Smith (Stanford) and J.W.Lichtman (Harvard) for helpful advice on tissue clearing, and D. Mou (Harvard),A. Govindarajan, K. Rockland and S. Tonegawa (Massachusetts Institute ofTechnology), A. Moore and C. Yokoyama (RIKEN) for critical comments. Thiswork was partly supported by grants from Japan Ministry of Education, Culture,Sports, Science and TechnologyGrant-in-Aid for Scientific Research on PriorityAreas and the Human Frontier Science Program.
AUTHOR CONTRIBUTIONSH.H. and A.M. conceived and designed the study. H.H. performed all theexperiments and analyzed the data. H. Kurokawa devised the algorithms andanalyzed the data. H. Kawano constructed the TPEFM system. R.A. performedin vitroexperiments using fluorescent proteins. T.S. designed and performedthe experiments that imaged callosal connections. H.N. refined the algorithms.K.F. contributed to data analysis. A.S.-S. performed the experiments using Fuccitransgenic mouse embryos. A.M. supervised the project and wrote the manuscriptwith the help of H.H.
COMPETING FINANCIAL INTERESTSThe authors declare competing financial interests: details accompany the full-textHTML version of the paper at http://www.nature.com/natureneuroscience/.
Published online at http://www.nature.com/natureneuroscience/.
Reprints and permissions information is available online at http://www.nature.com/
reprints/index.html.
1. Livet, J. et al.Transgenic strategies for combinatorial expression of fluorescent proteins
in the nervous system. Nature450, 5662 (2007).2. Micheva, K.D. & Smith, S.J. Array tomography: a new tool for imaging the molecular
architecture and ultrastructure of neural circuits. Neuron55, 2536 (2007).
3. Denk, W. & Horstmann, H. Serial block-face scanning electron microscopy to recon-
struct three-dimensional tissue nanostructure. PLoS Biol.2, e329 (2004).
4. Helmstaedter, M., Briggman, K.L. & Denk, W. 3D structural imaging of the brain with
photons and electrons. Curr. Opin. Neurobiol.18, 633641 (2008).
5. Denk, W. et al.Anatomical and functional imaging of neurons using 2-photon laser
scanning microscopy. J. Neurosci. Methods54, 151162 (1994).
6. Denk, W. & Svoboda, K. Photon upmanship: why multiphoton imaging is more than a
gimmick. Neuron18, 351357 (1997).
http://www.nature.com/neuro/index.htmlhttp://www.nature.com/natureneuroscience/http://www.nature.com/neuro/index.htmlhttp://www.nature.com/reprints/index.htmlhttp://www.nature.com/reprints/index.htmlhttp://www.nature.com/reprints/index.htmlhttp://www.nature.com/reprints/index.htmlhttp://www.nature.com/neuro/index.htmlhttp://www.nature.com/natureneuroscience/http://www.nature.com/neuro/index.html -
7/25/2019 Scale Nn.2928
9/10
NATURE NEUROSCIENCE doi:10.1038/nn.2928
Image acquisition using LSCM.In the experiment appearing in Figure 3hk,
brain samples of the YFP-H line were imaged using a ZEISS LSM710-NLO
system equipped with a 20 dipping objective (W-PlanApochromat, NA = 1.0,
working distance = 2.0 mm) and z drive. The z step size was 3.0 mm. YFP was
excited with a 514.5-nm Argon laser. The laser power was constant during image
acquisition; the excitation intensity at the level of the specimen was 2 mW. YFP
fluorescence was descanned and collected confocally. The size of the confocal
aperture was ~2 Airy disks (Supplementary Fig. 6b).
Image acquisition using macro zoom LSCM.Brain samples containing EYFP-
positive commissural connections were cleared with ScaleA2 and trimmed with
a coronal cut at the midbrain and positioned on a cover glass with the ante-
rior part uppermost. Transgenic mouse #596/#504 embryos (E11.5 and E13.5)
were cleared with ScaleU2. These samples were imaged using a Nikon macro
zoom confocal microscopy system (AZ-C1) equipped with a 1 objective lens
(AZ-PlanApo, NA = 0.1, working distance = 35 mm) or a 2 objective lens
(AZ-PlanFluor, NA = 0.2, working distance = 45 mm). EYFP and mAG were
excited with a 488-nm laser diode. mKO2 was excited with a 561-nm laser diode.
Pre-processing of images.Multiple xyimages were tiled using the stitch func-
tion of a commercial program (Multi-Area Time-Lapse View of FV10-ASW,
version 2.0c) at each zposition. The correlation function was used to compute
the xypositions in the overlapping regions. All tiled xyimages were processed
for shading correction by the custom-written filtering program JINARASHI.This program compensates signal intensity at the periphery of each image.
We developed JINARASHI using the Open CV Library (http://opencvlibrary.
sourceforge.net). JINARASHI is written in the C/C++ language and is available
on our website (http://cfds.brain.riken.jp/). JINARASHI corrects signal gradi-
ents on the basis of the background intensity distribution. Shading-corrected
images were stacked to achieve three-dimensional reconstruction using the
volume-rendering function of the commercially available Volocity version 5.3
(Improvision/PerkinElmer). Volume-rendered images were displayed using
fluorescence mode (three-dimensional opacity) and exported as AVI images.
Measurement of distances.Nuclei emitting mAG-hGem(1/110) green fluores-
cence or those stained with DAPI in the SGZ regions were tagged manually. In
contrast, blood vessels were automatically segmented using a custom-written
program, Frame Level Threshold, which converts the signal into binary. This
program is written in the C/C++ language and is available on our website (http://cfds.brain.riken.jp/). The distance from each nucleus to the nearest blood ves-
sel surface was measured using a custom-written program, RINZO. Different
parameters were set depending on the voxel shape, that is, the ratio of the length
of an xoryedge to that of a zedge. RINZO is written in Java and is available on
our website (http://cfds.brain.riken.jp/ ). Data were pooled from six SGZ regions.
Transmission measurement.Transmission curves were acquired using a U3310
spectrophotometer (Hitachi). Fixed brain slices (3 mm thick) were incubated
in PBS, a 60% sucrose solution, FocusClear or a Sca leA2 solution. The slices
equilibrated with sucrose or FocusClear shrank slightly, whereas the slices
treated with ScaleA2 expanded slightly. After incubation, slices with a thick-
ness of exactly 1.5 mm were generated from the equilibrated samples using
a tissue slicer. These 1.5-mm-thick slices each were mounted in a cuvette for
measurement of transmission.
Immunohistochemistry. Adult mice (7-9 weeks old) were deeply anesthetized
with pentobarbital (Nembutal) and transcardially perfused with 4% PFA/PBS.
After postfixation in 4% PFA/PBS and cryo-protection in 20% sucrose/PBS,
brains were embedded in OCT compound (Sakura). Restored brain samples
from ScaleA2 treatment were also embedded in OCT compound. Sagittal sec-
tions (35 mm thick) were cut with a cryostat. Sections were permeabilized/
blocked for 30 min in 0.1% Triton X-100 (wt/vol)/10% horse serum (vol/vol)/
PBS and then processed by free-floating immunohistochemistry. For double
staining, primary antibodies from different species were incubated simultane-
ously, followed by incubation with secondary antibodies.
For primary antibodies, we used mouse monoclonal antibody to PSA-NCAM
(Millipore), rabbit polyclonal antibody to GluR1 (Millipore) and mouse mono-
clonal antibody to synaptophysin (Sigma). For secondary antibodies, we used
ONLINE METHODS
Scale solutions.Stock solutions containing high concentrations of urea (Wako
Chemicals) or Triton X-100 (Wako Chemicals) were prepared. Scale solutions
were made by mixing the stock solutions and glycerol (Sigma). The final con-
centrations of urea, glycerol and Triton X-100 were adjusted by diluting the
mixed solutions with water.
Sample preparation.Neonatal and adult mice (2-13 weeks old) were deeply
anesthetized with pentobarbital (Nembutal) and killed by transcardial perfu-
sion with 4% PFA/PBS (wt/vol). The whole brains were taken out and subjected
to postfixation in 4% PFA/PBS at 4 C for 10 h and cryo-protection in 20%
sucrose/PBS (wt/vol) at 4 C for 24 h. For observation of the entire hippocampal
formation, the hippocampi were excised. Brains or hippocampal samples were
embedded in OCT compound and frozen. They were thawed in PBS and fixed
again with 4% PFA/PBS for 20 min at 25 C. Next, the samples were cleared in a
ScaleA2 solution (20 ml per 0.5 g tissue) at 4 C for >2 d. Mouse embryos (E13.5)
were transcardially perfused with 4% PFA/PBS, postfixed for 2 h and cleared
with ScaleA2 for 2 weeks. Alternatively, mouse embryos (E11.5 and E13.5) were
fixed with 4% PFA/PBS for 2 h and cleared with ScaleU2 for 6 months. For
speedy processing, fixed samples were initially cleared with ScaleA2 or ScaleU2
for 2 d, with ScaleB4 for 2 d, and then with ScaleA2 or ScaleU2 again. The experi-
mental procedures and housing conditions for animals were approved by the
Institutes Animal Experiments Committee of RIKENand all of the animals werecared for and treated humanely in accordance with the Institutional Guidelines
for Experiments using animals.
Imaging of fixed HeLa cells.HeLa cells grown on a 35-mm glass-bottom dish
were transiently transfected with cDNA for EGFP using Lipofectin (Invitrogen).
After fixation with 4% PFA for 10 min, cells were kept in Hanks Balanced Salt
Solution containing 15 mM HEPES-NaOH (pH 7.4) and time-lapse imaged
using an inverted microscope (IX81, Olympus) equipped with a standard 75-W
xenon lamp, a 40 objective lens (UplanFlN Oil, NA = 1.30) and a cooled CCD
camera (iXon EM+, Andor Technology). EGFP signals were obtained using
an excitation filter (470 10 nm), an emission filter (517.5 22.5 nm) and a
dichroic mirror DM485. The system was operated using MetaMorph 7.6 soft-
ware (Molecular Devices).
In utero
electroporation.After anesthesia with sodium pentobarbital, preg-nant mice at E15.5 were subjected to abdominal incision, and all the uterine
horns were exposed onto PBS-moistened cotton gauze. Embryos were visualized
through the uterine wall using a flexible fiber cable, and plasmid DNAs mixed
with the non-toxic dye Fast Green were injected into the lateral ventricle through
a pulled glass capillary. A pair of platinum electrodes was applied to the uterus,
and a series of square-wave current pulses (38 V, 50 ms) was delivered five times
at 1-s intervals using a pulse generator. Uterine horns were repositioned into the
abdominal cavity, and the abdominal wall and skin were sutured.
Image acquisition using TPEFM. The light source used for TPEFM was a
femtosecond-pulsed Ti:sapphire laser. When YFP- or GFP-expressing neurons
were imaged, an excitation wavelength of 920 nm was used. Emitted light was
collected by an external non-descanned detector (Supplementary Fig. 6a). In
the experiments shown in Figure 3ac,loand Supplementary Figure 7ae,
multiple neighboring regions in the brain were imaged using a TPEFM system(Olympus FV1000MPE + Coherent Chameleon Ultra-II + PreComp) equipped
with a motorized xystage module in addition to a motorized focus module
(Z-drive). Adjacent regions overlapped by 10% to allow precise alignment.
Lenses used included a 40 dipping objective (LUMPLFLN40XW, NA = 0.8,
working distance = 3.3 mm), a 25 dipping objective (XLPLN25XWMP, NA =
1.05, working distance = 2.0 mm) or a custom-designed 25 objective lens (NA =
1.0, working distance = 4.0 mm). The brightness compensation function in the
zdirection was used to change the detector sensitivity and laser power. In the
experiment represented in Figure 3dg, single regions were imaged using a
ZEISS LSM710-NLO system (+ Spectral Physics MaiTai HP DeepSee) equipped
with a 20 dipping objective (W-PlanApochromat, NA = 1.0, working distance =
2.0 mm) and z drive. The laser power was constant during image acquisition;
the excitation intensity at the level of the specimen was 6 mW.
http://cfds.brain.riken.jp/http://cfds.brain.riken.jp/http://cfds.brain.riken.jp/http://cfds.brain.riken.jp/http://cfds.brain.riken.jp/http://cfds.brain.riken.jp/ -
7/25/2019 Scale Nn.2928
10/10
doi:10.1038/nn.2928 NATURE NEUROSCIENCE
Transmitted-light bright-field (transmission) images were acquired with oblique
illumination and a reduced condenser aperture42, which can enhance struc-
tural differences in inherent absorption in the brain slices. Contrast was further
enhanced in the final images by electronic image processing. Fluorescence (YFP)
images were acquired using an excitation filter (480 20 nm) and an emission
filter (510-nm long pass). The outlines of the slices and several internal struc-
tures in the images were traced. The extent of the linear expansion (E) was calcu-
lated by comparing the distances across highlighted structures between 0-d and
5-d images. At 5 d, the image was reduced in size by 1 1/Efor superposition.
Statistical analysis.The statistical analysis for Figure 5g,hwas performed using
the Mann-Whitney Utest. Difference was considered to be significant when
P< 0.05. The data described in the text represent means s.d.
42. Keller, H. E. Contrast enhancement in light microscopy. Curr. Protoc. Cytom.2.1.1
2.1.11. (2001).
goat antibody to rabbit IgG conjugated to Alexa Fluor 546 (Molecular Probes),
goat antibody to mouse IgG conjugated to Alexa Fluor 546 (Molecular Probes), goat
antibody to rabbit IgG conjugated to Alexa Fluor 488 (Molecular Probes) and goat
antibody to rat IgG conjugated to Alexa Fluor 633 (Molecular Probes).
Observation of macroscopic structures in brain slices.Fixed brains were frozen
in OCT compound, thawed in PBS and fixed again with 4% PFA/PBS for 20 min
at 25 C. Then, coronal slices (1 mm thick) were cut with a razor blade from the
brain sample placed in the Mouse Brain Slicer (MB-A1-C, Muromachi Kikai).
Slices containing the hippocampus and the striatum were selected. These slices
were cut near the midline and the right halves were incubated in ScaleA2 solu-
tion at 4 C for 5 d, while the left halves were kept in PBS. Intermittently, these
slices were placed on a coverslip with a patterned background and imaged using
a Leica fluorescence stereomicroscope (MZ16F) equipped with a 1 objective
lens (PLANAPO, NA = 0.141) and a cooled CCD camera (DP30, Olympus).
top related