Neurogastroenterol Motil (2013) 25, e324e338

doi: 10.1111/nmo.12115

3-D imaging, illustration, and quantitation of enteric glial

network in transparent human colon mucosa
Y. A. LIU ,*,,

Y. C. CHUNG ,,
2

S. T. PAN , M. Y. SHEN , Y. C. HOU , S. J. PENG ,*,** P. J. PASRICHA ,

&

S. C. TANG *,,**,

*Connectomics Research Center, National Tsing Hua University, Hsinchu, Taiwan

Department of Chemical Engineering, National Tsing Hua University, Hsinchu, Taiwan
Department of Surgery, Cheng Ching General Hospital - Chung Kang Branch, Taichung, Taiwan
Department of Pathology, Miaoli General Hospital, Miaoli, Taiwan
Division of Colorectal Surgery, National Taiwan University Hospital - Hsinchu Branch, Hsinchu, Taiwan
**Institute of Biotechnology, National Tsing Hua University, Hsinchu, Taiwan
Division of Gastroenterology and Hepatology, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA

Abstract
Background Enteric glia form a network in the
intestinal mucosa and have been suggested to engage
in multidirectional interactions with the epithelium,
blood vessels, nerves, and immune system. However,
due to the dispersed nature of the glial network,
standard histology cannot provide a global view of the
network architecture. We prepared transparent human
colon mucosa for three-dimensional (3-D) confocal
microscopy with S100B immunostaining to reveal the
location-dependent glial network for qualitative and
quantitative analyses. Methods Full-thickness human
colons were acquired from colectomies performed for
colorectal cancer. We targeted the mucosa away from
the tumor site to characterize the glial network
morphology. Optical clearing (use of immersion solution to reduce scattering) was applied to generate
transparent specimens for deep-tissue microscopy. Key
Results Two features of the glial network were seen: (i)
A dense glial population resides at the crypt base/
mucosal boundary in contact with the lymphatic
vessels, and (ii) from the base, the glial network
elongates along the crypt axis with peri-cryptic and
peri-vascular connections toward the opening. We
quantified the mucosal glia as the S100B-positive cells
with at least two processes extending from the cell

body. Examples of the global and in-depth imaging of

adenoma were given to illustrate the morphological
correlation between the loss of glial fibers and the
aberrant crypts. Conclusions & Inferences We have
established a useful approach for 3-D imaging, panoramic illustration, and quantitation of the enteric
glia in the human colon mucosa to help characterize
their roles with mucosal components in health and
disease.
Keywords 3-D microscopy, adenoma, endothelia,
enteric glia, epithelia, human colon, lymphatic vessels, optical clearing, S100B.
Abbreviations: 2-D, two-dimensional; 3-D, threedimensional; a-SMA, a-smooth muscle actin; S100B,
S100 calcium binding protein B.

INTRODUCTION
Enteric glia form dispersed networks at different
layers of the intestine. The networks are most
noticeable at the myenteric and submucosal plexuses
as an essential component of the enteric nervous
system in association with neurons.15 In the
mucosa, glial cells have been suggested to engage
in multidirectional interactions with the epithelium,
blood vessels, nerves, and immune system.59 In
functional studies, conditional ablation of the enteric
glia using transgenic approaches resulted in gut
inflammation with compromised epithelial seal and
vascular integrity.10,11 Enteric glia also have been
reported to release soluble factors to regulate inflammation, promote mucosal wound healing, and protect
against pathogen invasion.1216 These results indicate

Address for Correspondence

Shiue-Cheng Tang, Institute of Biotechnology, National Tsing
Hua University, Hsinchu, Taiwan.
Tel: +886-3-574-2465; fax: +886-3-571-5934;
e-mail: sctang@life.nthu.edu.tw
1
Y. A. Liu and Y. C. Chung contributed equally to this work.
2
P. J. Pasricha and S. C. Tang contributed equally to this work.
Received: 9 October 2012
Accepted for publication: 15 February 2013

e324

2013 Blackwell Publishing Ltd

Volume 25, Number 5, May 2013

3-D glial network in human colon mucosa

Note that in clinical applications although robust

diagnostic criteria have been listed for examining the
qualitative alternations in the enteric nervous
system,25,38 these do not include quantitative values
of the neuronal and glial elements in the mucosal
samples due to lack of control data on studying the
normal human tissues.24 In particular, because the
glial cells/fibers are heterogeneously distributed in
the mucosa, a glial tissue map is critically needed to
illustrate the network orientation, which would help
specify the location-dependent variation in glial cell
quantitation. In this research, we took an engineering
approach to systematically build the labeling, imaging, and presentation system for microscopic visualization of the mucosal glia in a 3-D space continuum.
Our endeavor included: (i) optical clearing of the
human colon mucosa to prepare transparent specimens, (ii) paired immunostaining to identify the
association of glia with the mucosal elements, and
(iii) 3-D imaging and quantitation of the glial cells for
morphological illustration and quantitative analysis of
the glial network.
In this research, the glial cells are recognized by the
presence of glial-specific S100 calcium binding protein
B (S100B) together with nuclear staining to identify the
cellular processes and cell bodies.39,40 Paired immunostaining of S100B with the CD34 endothelial or D2-40
lymphatic vascular marker is used to reveal the glial
association with the blood and lymphatic vessels.41
Examples are given to develop an in-depth imaging
method for qualitative and quantitative analyses of the
enteric glia in the human colon mucosa. Illustration of
the glial network and the associated microstructure in
normal and adenoma tissues are presented and discussed in this report.

that enteric glia not only protect and interact with

neurons but also have a function in maintaining the
mucosal barrier.1619 In human bowel diseases, however, the roles of enteric glia and their changes in
response to pathological conditions have not been
well-characterized. In tissue samples from patients
and animals of intestinal inflammation, both reduced
glial content and up-regulation of glial proliferation
have been reported with limited understanding of the
mechanism (particularly in Crohns disease).5,11,2023
Although previous studies have suggested the functional interactions of the enteric glia with mucosal
components, few results have described the threedimensional (3-D) glial architecture in the human
intestine to associate the network morphology with
function or vice versa. Particularly, the glial network
orientation and cell density in the mucosa have not
been systematically examined,24 which limits our
understanding and assessment of the enteric glia in
health and disease.
To image the glial network, the current histological
protocol25 is constrained by using microtome sectioning to avoid scattering in optical microscopy. Biopsies
derived from endoscopic or surgical procedures are
routinely sectioned into thin slices, typically less than
5 lm, to allow for efficient light transmission in
microscopy. However, the long processes of glia (up
to 100 lm) and their connections in space cannot be
easily portrayed by the thin tissue slices. In addition,
the slender glial processes, often less than 1 lm,
demand substantial imaging sensitivity to detect and
determine their presence. Because of the intrinsic
limitation on the standard two-dimensional (2-D)
microscopy, a microtome-free, non-destructive imaging approach is preferable to provide an integral view of
the glial network.
To increase the imaging depth, we previously
developed a penetrative microscopic method, based on
preparation of transparent tissues by optical clearing,26,27 for 3-D imaging of the mouse gut tissue
networks, including the microstructure, vasculature,
and innervation,2832 and the human enteric nervous
system.33,34 Here, we employed the same penetrative
3-D microscopy with immunostaining to provide a
global and integral view of the enteric glia and the
associated tissue components in the human colon
mucosa. Optical clearing is performed by immersion of
the mucosal specimens in the clearing reagent with a
similar refractive index as the tissue constituents (~1.46)
to achieve refractive uniformity.3537 The clearing process reduces light scattering in the tissue, leading
to increased photon penetration in optical microscopy
to reveal the fluorescently labeled glial network.

2013 Blackwell Publishing Ltd

MATERIALS AND METHODS

Human specimens
Collection and use of human tissues were approved by the
Institutional Review Board of National Taiwan University
Hospital Hsinchu Branch with written consent from the
patients to use the obtained tissues. Colonic tissues were
obtained from colectomies carried out for carcinoma. Samples
of normal mucosa were obtained at least 5 cm apart from the
carcinoma. Freshly isolated tissues were first fixed with 4%
paraformaldehyde solution for 30 min at room temperature and
then washed extensively with phosphate-buffered saline (PBS).
Vibratome sections (~300 lm) were postfixed in 4% paraformaldehyde solution for two hours at 4 C before being used for
tissue labeling. Overall, 92 image stacks derived from four
patients were collected to illustrate the 3-D structure of the
glial network. Specimens from three of the four patients were
used in statistical analysis. Table 1 lists the gender, age, and
location of the sampled colon segments of the three subjects.

e325

Y. A. Liu et al.

Neurogastroenterology and Motility

Table 1 Mucosal glial cell density of human colon

Subject 1
Subject 2
Subject 3

Gender/Age

Segment

Mucosal glial cell

density in domain I
(crypt axis direction
in Fig. 6B) (cells mm 3)

Male/84
Female/59
Male/80

Sigmoid colon (n = 5)
Transverse colon (n = 5)
Ascending colon (n = 5)

4 073  448
4 603  486
4 068  1 084

The average densities derived from the three subjects are 4 248  727 cells mm 3 and 10 824  2 802 cells mm
Data are presented as means  SD.
*P < 0.005, P < 0.001, P < 0.05 vs domain I (crypt axis direction).
There was no statistical difference between any two subjects of their cell densities in the same domain.

Specimens from the fourth patient (man, age 80 years, sigmoid

colon) were used to present the glial network in Fig. 4. Similar
glial network morphologies were seen among the four subjects
in the samples of normal mucosa.

Mucosal glial cell

density in domain II
(longitudinal direction in
Fig. 6B) (cells mm 3)
10 365  2 555*
11 811  1 829
10 296  3 971
of domains I and II, respectively.

antibody (cat# 1184-1; Epitomics) was used to confirm the

morphology of the peri-cryptic myofibroblasts (Figure S5).
Finally, the labeling patterns of S100B, CD34, D2-40, and
a-SMA were consistent for the specimens derived from different
individuals.

Tissue labeling

Confocal microscopy

The fixed specimens were immersed in 2% Triton-X 100 solution

for 2 days at 15 C for permeabilization. Four different primary
antibodies were used to immunolabel the tissues following the
protocol outlined below. The antibodies used were a monoclonal
rabbit anti- S100B (Cat# 2017-1; Epitomics, Burlingame, CA,
USA), monoclonal mouse anti-CD34 (Cat# BSB 5230; Bio SB,
Santa Barbara, CA, USA), monoclonal mouse anti-a-smooth
muscle actin (a-SMA; Cat# MS-113-P0; Thermo Fisher Scientific,
Fremont, CA, USA), and mouse monoclonal D2-40 (Cat#
SIG-3730; Covance, Princeton, NJ, USA) antibodies. Before
applying the antibody, the tissue was rinsed in PBS. This was
followed by a blocking step, incubating the tissue with the
blocking buffer (2% Triton X-100, 10% normal goat serum, and
0.02% sodium azide in PBS). The primary antibody was then
diluted in the dilution buffer (1 : 100, 0.25% Triton X-100, 1%
normal goat serum, and 0.02% sodium azide in PBS) to replace the
blocking buffer and incubated for 1 day at 15 C.

Imaging of the tissue structure was performed with a Zeiss LSM

510 Meta confocal microscope equipped with the objective of 109
Fluar lenses to acquire the gross images under a regular (optical
section: 10 lm; Z-axis increment: 5 lm) or tile-scan mode (used
in Fig. 7). The 259 LD Plan-Apochromat glycerine immersion
lenses (working distance: 570 lm; optical section: 5 lm; Z-axis
increment: 2.5 lm) and 409 LD C-Apochromat water immersion
lenses (working distance: 620 lm; optical section: 2 lm; Z-axis
increment: 1 lm) were used to acquire the high-resolution
images. Each confocal micrograph consisted of 1 024
(X) 9 1 024 (Y) pixels (except Fig. 7). The laser-scanning process
was operated under the multi-track scanning mode to sequentially
acquire signals in multiple channels, including the transmitted
light channel. The Alexa Fluor 647-labeled tissue structures were
excited at 633 nm and the fluorescence was collected by the 650to 710-nm band-pass filter. The Alexa Fluor 546-labeled tissue
structures and propidium iodide-labeled nuclei were excited at
543 nm and the signals were collected by the 560- to 615-nm
band-pass filter. The SYTO 16 signals were excited at 488 nm and
the fluorescence was collected by the 500- to 550-nm band-pass
filter.

Alexa Fluor 647 conjugated goat anti-rabbit secondary antibody

and Alexa Fluor 546 conjugated goat anti-mouse secondary
antibody (1 : 200, Invitrogen, Carlsbad, CA, USA) were used to
reveal the immunostained structure. Afterward, nuclear staining
by propidium iodide (62.5 lg mL 1, Invitrogen) or SYTO 16
(4 lM, Invitrogen) was performed at room temperature for 1 h.
The labeled specimens were then immersed in the opticalclearing solution FocusClearTM (CelExplorer, Hsinchu, Taiwan)
before being imaged via confocal microscopy.37

Image processing, projection, and analysis

We performed the following control experiments to ensure

the immunostaining signals. Firstly, we used the goat anti-rabbit
or goat anti-mouse secondary antibody alone, or in combination
with the mismatched primary antibody, to rule out false positive
results due to non-specific secondary antibody binding and
species cross-reactivity of the antibodies (Figure S1). Secondly,
we used a second rabbit anti-S100 antibody (polyclonal, purchased from vender DAKO, Glostrup, Denmark; cat# Z0311) to
verify the enteric glial network morphology (Figure S2). Thirdly,
the signals derived from the CD34 immunostaining were
confirmed by the matched vascular information after overlaying
the confocal micrographs with the transmitted light micrographs
(Fig. 2A and Figure S3). The same approach was also used to
verify the D2-40+ lymphatic vessels (Figure S4 and Video S7).
Fourthly, coupled immunostaining using the mouse anti-a-SMA
antibody (Thermo Fisher Scientific) and rabbit anti-a-SMA

The Avizo 6.2 image reconstruction software (VSG, Burlington,

MA, USA), the Zen software (Carl Zeiss, Jena, Germany), and the
LSM 510 software (Carl Zeiss) were used for processing, projection, and analysis of the confocal image stacks. The Avizo
software was operated under a Dell T7500 workstation (96 GB
memory) with a Linux operating system. Avizos noise filtering
algorithms were applied for background noise reduction. In
Supplementary Videos S1, S3, S4, and S6S9 the image stacks
were displayed using the Ortho Slice function, and the video was
made via the Movie Maker function with the increase in display
time in association with the depth of the optical section. In
Figs 4E, 5CF, 6A and C, and Supplementary Figures 6G, 7BD,
and Videos S5S7 and S10, the Voltex module was used to
project the 3-D images. The Camera Rotation function was used
to create the 360-degree panoramic displays of the 3-D images.
Image segmentation and feature extraction were performed by the

e326

2013 Blackwell Publishing Ltd

Volume 25, Number 5, May 2013

3-D glial network in human colon mucosa

to qualitatively and quantitatively characterize the

glial network.

Label Field function of Avizo to present the glial nuclei (Figs 4E,
5D and E) and plexuses of different orientations (Fig. 6C). Figs 1E,
2DF, 3C and D, 4C and D, 6B, 7E and G, and Figures S1, S2B, C, E
and F and S6BF were derived from the 3-D projection module of
the LSM 510 software. The 3-D projection function of Avizo was
used to present the orthogonal view of the X/Y, Y/Z, and X/Z
planes of the image stack shown in Fig. 4E (middle panel) and
Figure 6G.

Transparent mucosa shows peri-vascular and

peri-cryptic morphologies of the glial network
Using the transparent specimen, we first zoomed in to
investigate the glial network at the top of the crypt
axis. Fig. 2A shows the close-up view of the optically
cleared mucosa with peri-vascular presence of the glial
cell. Under the same view, we used a series of highresolution confocal micrographs (Fig. 2B, C and
Video S1) to identify the glial cell bodies and processes
in space. Projections of the image stack (Fig. 2DF and
Video S2) show that the enteric glia form a continuous
network in the lamina propria, extending from the
lower part of the crypt toward the crypt openings with
peri-cryptic and peri-vascular extensions. From the
luminal point of view, Figure S6 and Video S3 confirm
the peri-cryptic and peri-vascular morphologies of the
mucosal glial network. The intimate association
among the enteric glia, endothelium, and epithelium
support their functional relationships in the colon
mucosa.
In addition, we used the peri-cryptic myofibroblasts
as the reference to define the proximity of the enteric
glia to the epithelium. Fig. 3 and Video S4 show the
result of paired staining of S100B and a-smooth muscle
actin (myofibroblast marker) revealing that the pericryptic glial processes are in close association with the
myofibroblasts and occasionally interposed between
the myofibroblasts and the epithelial basement membrane. Projections of the S100B and a-smooth muscle
actin signals illustrate the different architectures of the
two components.

Quantitation of glial cells in the normal tissue domain was

performed in the transverse sections of human colon mucosa
away from the tumor site. Fifteen 150-lm image stacks [with a
tissue volume of 369 (X) lm 9 369 (Y) lm 9 150 (Z) lm 
20 9 106 lm3 or 0.02 mm3 of each image stack] from three
subjects were used to quantify the glial cell density. In each
image stack, 61 consecutive optical sections were used in
quantitation. Overall, 305 images of each subject were used in
the test. Both the S100B and nuclear signals were used to define
a glial cell a S100B immunoractive cell body with at least two
processes. The Landmark and Material Statistics functions of
Avizo were applied to mark/count the glial cells in each
segmented image domain (such as the domains I and II in
Fig. 6B) and measure the segmented tissue volume for estimation of the cell density (Fig. 6C and Table 1). Data in Table 1
are presented as means  SD. In Figs 1 and 7, the Profile
Analysis module of the Zen software was used to reveal the
change in the signal intensity along the pixel line in the
micrograph.

RESULTS
Optical clearing facilitates photon penetration for
in-depth imaging of colon mucosa
The functional interaction between the enteric glia and
epithelia has been widely suggested to influence the
integrality of the mucosal barrier. However, because
the colonic mucosa strongly scatters light, particularly
at the mucosal surface around the crypt openings
(Fig. 1A and B), the intrinsic tissue opacity hinders indepth imaging of the glial network and the colonic
structure.
To visualize the mucosal crypts and glia, we
immersed the human colon specimens in the opticalclearing solution with high refractive index to reduce
scattering (Fig. 1C and D). Notably, the optically
cleared specimen allows a direct visualization of the
colonic crypts and the muscularis mucosae via transmitted light microscopy. In addition, the improved
light transmission also enables the use of confocal
microscopy to reveal the S100B-labeled glial cells deep
in the mucosal specimen (Fig. 1D). The sharp signal-tonoise ratios highlight the advantage of using optical
clearing to facilitate the detection of the fluorescently
labeled glial network.
Fig. 1E shows the projection of the acquired image
stack to provide a gross view of the glial network
morphology. In the next three sections we zoom in and
focus on the top and base of the human colon mucosa

2013 Blackwell Publishing Ltd

A dense glial population lining the crypt base/

mucosal boundary in contact with lymphatic
vessels
Fig. 4AD show the bottom half of the colonic crypts,
in which two features of the glial network were seen.
Firstly, a dense glial population was found at the crypt
base with longitudinal extensions following the contour of the mucosal boundary. Secondly, from the base
the glial network elongated along the crypt axis with
peri-cryptic and peri-vascular morphologies, similar to
those at the top of the mucosa (Fig. 2). Particularly, we
employed image segmentation to extract the nuclei
encircled by the S100B signals to identify the individual glial cells and traced their paths via 3-D image
projection (Fig. 4E). Rotation of the image stack (Videos S5 and S6) confirmed that the glial plexus lines the

e327

Y. A. Liu et al.

Neurogastroenterology and Motility

Control, tissue in saline

Relative signal intensity 100%

Confocal micrograph
Profile analysis

Mucosa

Mucosa

Distance (pixel)

Mucosa

Muscularis Mucosae

Muscularis Mucosae

Muscularis Mucosae

100 m

1024

50- m under the surface 1024

Optically cleared tissue

Mucosa

100 m

Mucosa

Muscularis Mucosae

Muscularis Mucosae

50- m under the surface 1024

Relative signal intensity 100%

Confocal micrograph

Distance (pixel)

Mucosa

Muscularis Mucosae

1024

Projection of S100B (white) and nuclear (green) signals; depth: 150 m

100 m

Figure 1 Optical clearing of human colon mucosa facilitates photon penetration for transmitted light and confocal imaging. (AD) Comparison
between the saline-immersed (control) and the optically cleared human colon specimens. The thickness of the sections was ~300 lm. Panels A and C
are transmitted light micrographs. Note that in panel A, the top of the mucosa is intrinsically more opaque than the area close to the muscularis
mucosae. Panels B and D are confocal micrographs of the S100B-labeled enteric glial signals and their profile analysis. Images were taken 50 lm under
the surface. The vertical pixel lines in the middle of the two micrographs correlate with the vertical axis of the intensity profile, the right panel that
consists of the signal intensities of the vertical pixel line. Particularly, in panel B, light scattering in the saline-immersed control leads to the
disappearance of signal peaks in the mucosa. (E) Gross view of the enteric glia in mucosa. Left: in-depth projection of the enteric glia. Middle: nuclear
signals. Right: merged projection. Panels CE were derived from the same image stack.

that the glial network at the crypt base was also in

contact with the lymphatic vessels at the mucosal
boundary (Fig. 5 and Video S7). To confirm that the
S100B+ cells at the crypt base are not the myofibroblasts, we used paired immunostaining of S100B and a-

interface of mucosa and muscularis mucosae and

followed the crypt curvature in extension along the
crypt axis.
In addition, using paired D2-40 staining of the
lymphatic endothelium with S100B, we discovered

e328

2013 Blackwell Publishing Ltd

Volume 25, Number 5, May 2013

Transmitted light micrograph

3-D glial network in human colon mucosa

Confocal micrographs of S100B (white), capillary (red) and nuclear (green) signals
Gallery display, increment: 2.5 m per section

50 m

50 m

CD34 (capillary, red) + S100B (white)

Figure 2 Transparent mucosa enables visualization of the enteric glia and the associated microstructures with high definition. (A) Close-up view of
the mucosa. Upper panel: optical clearing reveals the mucosal landmarks, including the capillary in lamina propria (arrows). Lower panel: CD34
staining (red) confirms the locations of capillaries. S100B staining shows the peri-vascular presence of the enteric glia (arrows). (B) Gallery display of
the S100B-labeled glial network and the CD34-labeled capillaries. Panels A and B were taken under the same view. (C) Enlarged micrograph shown in
panel B3 (the cyan box). Arrows indicate the glial cell bodies. (D and E) Projections of the image stack shown in panel B (depth: 20 lm). The box in
panel D is enlarged in panel E to reveal the glial fibers. Arrows in panels C and E indicate the same glial cells. (F) In-depth projections of the crypts
(nuclear signals), capillaries, and glial network. The peri-cryptic and peri-vascular glial fibers and connections are the hallmark of the network
morphologies in this domain. Signals were derived from the image stack shown in Video S1. Different viewing angles of the glial network (the last
panel) are shown in Video S2.

with a depth of 75 lm. Projection of the 75-lm S100B

and nuclear signals into a 2-D surface allowed quantitative analysis of the glial cells (Fig. 6B). The presence
of a glial cell was judged by at least two S100B-positive
processes (red) encircling the cell body (green, nuclear
staining). Three examples of the glial image stack are
presented in Video S9. Glial cell density can be
estimated by counting the cells at different domains,
e.g., along the crypt axis (axial direction, domain I) or at
the crypt base/mucosal boundary (longitudinal direction, domain II), and then dividing the number with the
respective tissue volumes.
In the second approach, we took advantage of the
voxel-based image data to quantify the glial cell
density with help from software algorithms. Fig. 6C
illustrates the two directions/domains of the glial
fibers along the crypt axis (yellow) and at the crypt
base/mucosal boundary (cyan). In the two domains, we

SMA and high-resolution microscopy to resolve the

enteric glial and myofibroblast populations beneath
and around the crypts. The result is presented in
Video S8. Overall, the tissue structures revealed in
Figs 4 and 5 indicate the glial network at the crypt base
associate with the epithelium, blood vessels, and
lymphatic vessels.

Quantitative analysis of the location-dependent

glial cell density
We have demonstrated the different glial network
morphologies along the crypt axis and at the crypt
base/mucosal boundary. In Fig. 6, we illustrate two
approaches to glial cell quantitation for analysis of the
location-dependent glial cell density.
In the first approach, Fig. 6A shows that the acquired
150-lm image stack is divided into two sections, each

2013 Blackwell Publishing Ltd

e329

Y. A. Liu et al.

Neurogastroenterology and Motility

Zoom-in examination of panel B3

Projection of S100B, capillary and nuclear signals

Depth: 20 m

Zoom-in examination of panel D

50 m

Projection depth: 200 m

50 m

50 m

S100B + CD34

Nuclei

S100B

50 m

Figure 2 Continued

crypts in comparison with those (white peaks) in the

normal domain (Fig. 7A and C).
To confirm the lack of glial fibers, we next zoomed
in to visualize the adenoma with high definition
(Fig. 7DG). In adenoma, the substantial change in
the glial signals was at the location close to the
mucosal surface, in which the abnormal crypt structure was seen. To emphasize the change, in Fig. 7DG
we compare the S100B staining of the adenoma with
that of the normal crypts, both at the top of the
mucosa. The lack of peri-cryptic glial fibers (Fig. 7E),
opposite from the rich glial presence in the normal
control (Fig. 7G), highlights the alternation of the
microenvironment around the irregular crypts in
adenoma.
Technically, the gross image and in-depth projections in Fig. 7 offer a side-by-side comparison
between the health and disease states of mucosa.
This allows us to examine the tissue structures and
verify the change in the glial network across the
normal and affected area by 3-D projections, which
cannot be easily portrayed by the standard 2-D tissue
analysis. Figure S7 and Video S10 provide additional
viewing styles and projection angles of the diseased
colon.

used software algorithms to automatically trace and

label the nuclei that were encircled by the S100B
signals in space for quantitation (the last panel of
Fig. 6C).
We used both approaches to calculate and verify the
location-dependent glial cell density. The result is
summarized in Table 1. The average glial cell density
at the crypt base is 10 824  2 802 cells mm 3 (average of three subjects), which is 2.5-fold higher than that
along the crypt axis.

Irregular crypts in adenoma lack peri-cryptic glial

fibers
In addition to the normal mucosa, Fig. 7AC illustrates the strategy of imaging the adenoma (left) with
the normal mucosa (right) by stitching the acquired
images laterally to compare the nuclear and S100B
signals across the normal and diseased domains. In the
adenoma, abnormal crypts and abundant nuclear signals (neoplasia; green peaks) are prominent in contrast
to the normal crypt organization and mucosal cell
density (Fig. 7A and B). Notably, contrary to the
increased nuclear signals in adenoma, a decrease in
glial signals was found associated with the irregular

e330

2013 Blackwell Publishing Ltd

Volume 25, Number 5, May 2013

3-D glial network in human colon mucosa

Transmitted light micrograph

Merged transmitted light & confocal images

Projection of S100B,

-SMA & nuclei

50 m

Nuclear (green), S100B (white), and -SMA (magenta) staining

S100B

S100B +

-SMA (myofibroblasts)

Depth: 81 m
S100B + nuclei

Figure 3 Close association among the peri-cryptic glia, myofibroblasts, and crypt epithelium. (A and B) Transmitted light and fluorescence
micrographs of the whole mount human colon mucosa (merged signals in panel B). S100B, a-smooth muscle actin (a-SMA), and nuclear staining were
used to reveal the locations of enteric glia, myofibroblasts, and crypts, respectively. (C and D) Merged and individual projections of the glial network
(S100B), myofibroblasts (a-SMA), and crypts (nuclear signals). The peri-cryptic glia and myofibroblasts are in contact with each other and close
proximity to the epithelial base membrane, suggesting their influence on the crypt epithelial cells (likely the proliferation/differentiation). Yellow
arrow: example of a peri-cryptic glial cell body. Cyan arrows: examples of the glial processes being interposed between the myofibroblasts and
epithelium. Panels AD were taken under the same view. An additional high-resolution image stack of the S100B, a-SMA, and crypt signals is shown
in Video S4.

Using the high-resolution 3-D images, we reveal the

peri-cryptic and peri-vascular glial plexuses in the
human colon mucosa (Figs 24). Our result simultaneously depicts the intimate glial-epithelial and glialendothelial associations in morphology, supporting
their functional interactions as shown previously by
conditional ablation of enteric glia resulting in intestinal inflammation with a breakdown of mucosal and
vascular barriers.10,11 In addition, in Fig. 5, we reveal a
dense glial population at the crypt base and mucosal
boundary in contact with the D2-40-labeled lymphatic
vessel. The unique glial cell population at the crypt
base implies functional influences of glial cells on the
crypt base cells, including the epithelial stem cells and
Paneth cells. The morphology also indicates the connection from crypt base cells to lymphatic vessels via
the glial network, suggesting potential information
relay or exchange among the three components.
From a technical point of view, the location-dependent network morphology and glial cell density (Fig. 6

DISCUSSION
Enteric glia are prominent in the human gut mucosa,
but their roles in health and disease have not been
extensively studied. The difficulty is, in part, due to
the lack of appropriate tools to perform qualitative and
quantitative analyses of the glial network. In this
research, we prepared transparent human colon
mucosa by optical clearing for penetrative imaging of
the S100B-labeled glial cells with high definition.
Unlike the standard microtome-based 2-D tissue analysis, our 3-D imaging approach provides continuous
anatomic information to trace the glial cell-to-cell
coupling in space. The 3-D histological scanning
generates panoramic views of the glial network and
its association with tissue components including the
crypts, capillaries, and lymphatic vessels, illustrating
the multidirectional interactions of the mucosal glia
with epithelia, endothelia, and the immune system in
the human colon.

2013 Blackwell Publishing Ltd

e331

Y. A. Liu et al.

Neurogastroenterology and Motility

Transmitted light micrograph

Fluorescent 2-D micrograph

Projection of nuclear signals

50 m

S100B + blood vessels (CD34)

Projection of S100B signals; depth: 80 m

3-D projection of enteric glia

Dimensions: 256 256 80 (Z, depth) m

Merged projection of enteric glia & blood vessels

with the orthogonal view of the image stack

S100B + nuclei

Merged projection of enteric glia, blood vessels,

and segmented crypt nuclei

*
*

Y
Z

Z
X

Nuclei are segmented from panel C

30 Rotation of the images tack along the X-axis

Figure 4 3-D illustration of the glial network at the crypt base. (AC) Close-up view of the colonic crypt base. Arrows in panel B indicate the glial cell
bodies. (D) In-depth projections of the glial network at the crypt base/mucosal boundary under the same view of panels AC. The peri-cryptic
and peri-vascular glial morphologies were revealed in the penetrative projections. (E) Individual and merged 3-D projections of the glial network
around the crypt bases and blood vessels. In the first panel, signals of glial cell nuclei were segmented from panel C to specify the locations of the glial
cell bodies. 360-degree projections of the first two panels are shown in Videos S5 and S6. In the last panel, signals of crypt cell nuclei were segmented
to outline the crypt structure. Overall, the merged signals simultaneously reveal: (i) the peri-cryptic glial fibers (arrows), (ii) the peri-vascular glial
cell extensions (asterisks), and (iii) a dense glial population residing at the crypt base and mucosal boundary (the lower part of the image).

the artifact and distortion inherent in microtome

slicing as well as the challenge of aligning series of
microtome slices,4245 the standard 2-D microscopy is
limited to provide an in-depth view of the glial
network in a space continuum.
Taking advantage of the optical clearing technique
(Fig. 1), in this research we overcome the depth

and Table 1) underline the challenge of using the 2-D

microscopy to study the mucosal glia. Specifically, the
dispersed glial network at the top of the colon mucosa
and the dense glial population at the crypt base require
a wide and in-depth view of the specimen with
substantial sensitivity and resolving power to capture
and distinguish the image signals. However, because of

e332

2013 Blackwell Publishing Ltd

Volume 25, Number 5, May 2013

3-D glial network in human colon mucosa

Transmitted light micrograph

Fluorescent 2-D micrograph

3-D projection of lymphatic vessels + panel B

Y
Z
50 m

50 m

Nuclear (green), S100B (white), and D2-40 (magenta) staining

3-D projection of lymphatic vessels and glia

Merged projection of panels C & D

Projection of lymphatic vessels, glia & crypts

Y
Z
X

Dimensions: 360 360 150 (Z, depth) m

Figure 5 Glial network at the crypt base links to lymphatic vessels. (A and B) Transmitted light and fluorescence micrographs of the crypt base and
mucosal boundary. In panel B, nuclear, S100B, and D2-40 (lymphatic endothelial marker) staining were used to reveal the locations of crypts, enteric
glia, and lymphatic vessels, respectively. (C) 3-D projection of the D2-40-labeled lymphatic vessels with panel B. Arrows in panels B and C indicate
the same peri-lymphatic glial cell bodies. (DF) Merged projections of enteric glia, lymphatic vessels, and nuclear signals at the crypt base and
mucosal boundary. The glial cell nuclei in panels D and E were segmented from the signals shown in panel F to reveal the locations of the glial cell
bodies. The three panels reveal that the glial network extends toward the lymphatic vessels from the peri-cryptic domain, constituting a crypt-gliallymphatic axis for potential information exchange. In-depth observation and 360-degree projection of panel D is shown in Video S7.

the diseased mucosa. We employed both the global and

in-depth tissue analysis to compare the adenoma with
the surrounding mucosa to establish the histological
association between the loss of S100B+ glial fibers and
the aberrant crypts in the diseased tissue. This observation supports the concept that the glial-epithelial interactions regulate epithelial cell growth in the normal
mucosa and implicates the lack of glial-epithelial interactions in colorectal carcinogenesis.8,46,47 However, we
do not rule out that the decrease in the immunohistochemical signals could be due to the release and
depletion of the glial S100B protein under the disease
condition. This is an important insight that needs to be
validated clinically with additional tissue specimens
but also illustrates the value of our microscopic technique in assessment of enteric glia in mucosal diseases.
In summary, we developed an optical approach
to systematically reveal the enteric glia in the
human colon mucosa. Light scattering in opaque colon
samples has previously hindered observation of the

limitation imposed by light scattering while imaging

the mucosal specimen. In-depth tissue microscopy is
feasible up to 200 lm with subcellular-level resolution to identify individual glial cell bodies and
processes with high definition. Importantly, in addition to the S100B-labeled glial cell signals, other
cellular and tissue information derived from the
nuclear, a-SMA, CD34, and D2-40 staining and transmitted light microscopy are simultaneously used to
generate a connected view of the glial network with the
associated microstructures. For example, in Fig. 6, we
take advantage of the nuclei encircled by the S100B
signals to specify the glial cell bodies for quantitation.
Importantly, the in-depth, multichannel displays of the
colon mucosa allowed us to verify the fidelity of the
image signals by comparing the sources of information
with each other, which is critical for future clinical
evaluation of the glial content in biopsies.
In Fig. 7, tumor specimens were used as an example to
demonstrate the feasibility of performing 3-D imaging of

2013 Blackwell Publishing Ltd

e333

Y. A. Liu et al.

Neurogastroenterology and Motility

B
Dimensions: 369 369 150 (Z, depth) m

I
Y
Z
X

Merged 3-D projection of S100B and

nuclear signals

Separately project two

75- m sections along
the Z-axis

II

Y
Z
X

100 m

Quantitation of enteric glia

Stereo projection of segmented S100B & nuclear signals

Dimensions: 369 369 150 (Z, depth) m

E.g., in the 1st section, 24 and 34 glial cells are found in domains I and II,
respectively; volume of the image section: 0.010 mm3 (domain I: 65%)
Projection of segmented S100B signals
Crypt axial direction: yellow; longitudinal direction: cyan

Digital identification of glial cell bodies in space

Y
Z
X

Figure 6 Quantitation of the location-dependent glial cell density in colon mucosa. (A) Example of a 150-lm image stack of S100B and nuclei
assigned for quantitative analysis. (B) 2-D features of the enteric glia (red) and nuclei (green) projected from the first section of the image stack shown
in panel A. Arrows indicate the glial cell bodies. We define a glial cell as S100B+ with at least two processes extending from the cell body. (C) 3-D
illustration of the software-based S100B and nuclear signal projections and quantitation. The S100B-labeled glial fibers were digitally segmented and
assigned with different colors to indicate their orientations. In the last panel, the overlap between the S100B and nuclear signals were presented as
dots in the scanned volume for quantitation.

ACKNOWLEDGMENTS

glial network morphology in space. We prepared

transparent mucosal specimens by optical clearing for
3-D imaging, panoramic illustration, and quantitation
of the glial content in the mucosa. Future study will
focus on mapping the enteric glia at different locations
of the digestive tract and establishing procedures to
analyze the endoscopic biopsies from patients with
mucosal diseases.

The authors thank the Brain Research Center in the National

Tsing Hua University for technical support in confocal imaging
and postrecording image processing. Present address for Y. C.
Hou: Division of Colorectal Surgery, Nan Men General Hospital,
Hsinchu, Taiwan.

e334

2013 Blackwell Publishing Ltd

Volume 25, Number 5, May 2013

3-D glial network in human colon mucosa

Adenoma

Relative signal intensity

100%

Normal domain

4080

Distance (pixel)

Adenoma

Normal domain
4080

200 m

Norm

Lack of glial fibers

Zoom-in examination of the dotted box in panel D (adenoma); projection depth: 150 m

Spreading/
progression

Adenoma

Adenoma

100 m
100 m

Control, away from the adenoma

Zoom-in examination of the solid box in panel F (unaffected mucosa); projection depth: 150 m

100 m
100 m

Figure 7 Signal and morphological analyses of tumor biopsies indicate irregular crypts in adenoma lack peri-cryptic glial fibers. (AC) Signal profile
and micrograph of the nuclear (green) and S100B (white) staining of the colonic tumor biopsy. Five images were stitched to provide a gross view of the
specimen across the adenoma and normal domains. The horizontal line in panel B (4 080 9 1 024 pixels) correlates to the horizontal axis of panel A,
which indicates the intensity profile of the pixel line. The increased nuclear signals in adenoma correlate to a higher frequency of the nuclear peaks
(green) at the left side in panel A than that of the normal domain (right). In comparison, the glial fibers and their signal peaks (white) are seen in the
normal domain (the right side of panels AC), but the glial fibers/signals are missing at the top of the mucosa in the adenoma region. (D and E)
Irregular crypts in adenoma lack the peri-cryptic S100B+ glial fibers. (D): gross view at the boundary of adenoma. The dotted box indicates the location
in which the high-resolution image stack was taken (E) to examine the crypt and glial morphologies. (E): projections of nuclear (left) and S100B (right)
signals reveal the loss of S100B+ glial fibers in adenoma: The cell bodies were seen scattered in this region, yet the S100B+ glial cell processes and their
connections were missing around the irregular crypts (upper part of the panel). (F and G) Normal mucosa away from the tumor site possesses
organized crypts and the peri-cryptic glial fibers. The solid box in panel F indicates the location in which the high-resolution image stack was taken
(G).

2013 Blackwell Publishing Ltd

e335

Y. A. Liu et al.

Neurogastroenterology and Motility

FUNDING

AUTHOR CONTRIBUTIONS

This study was supported by grants from the Taiwan National

Science Council (NSC-100-2628-E-007-021-MY2), National
Health Research Institutes (NHRI-EX102-10044EI), and Ministry
of Education (101N2060E1) to S.C.T.

YAL, YCC, STP, MYS, YCH, and SJP contributed to human tissue
collection/preparation, experimental design, and analysis/interpretation of data; YAL contributed to image processing and
analysis; PJP and SCT helped design the immunostaining study
and contributed to the writing of the article; SCT and YCC
directed the research project.

COMPETING INTEREST
None.

REFERENCES
1 Gershon MD. The enteric nervous
system. Annu Rev Neurosci 1981; 4:
22772.
2 Wedel T, Roblick U, Gleiss J et al.
Organization of the enteric nervous
system in the human colon demonstrated by whole mount immunohistochemistry with special reference to
the submucous plexus. Ann Anat
1999; 181: 32737.
3 Gershon MD, Rothman TP. Enteric
glia. Glia 1991; 4: 195204.
4 Ruhl A, Nasser Y, Sharkey KA.
Enteric
glia.
Neurogastroenterol
Motil 2004; 16(Suppl. 1): 449.
5 Ruhl A. Glial cells in the gut. Neurogastroenterol Motil 2005; 17: 77790.
6 Gulbransen BD, Sharkey KA. Novel
functional roles for enteric glia in the
gastrointestinal tract. Nat Rev
Gastroenterol Hepatol 2012; 9: 625
32.
7 De Giorgio R, Giancola F, Boschetti E,
Abdo H, Lardeux B, Neunlist M.
Enteric glia and neuroprotection:
basic and clinical aspects. Am J Physiol Gastrointest Liver Physiol 2012;
303: G88793.
8 Bach-Ngohou K, Mahe MM, Aubert
P et al. Enteric glia modulate
epithelial cell proliferation and differentiation through 15-deoxy-12,14prostaglandin J2. J Physiol 2010; 588:
253344.
9 Neunlist M, Van Landeghem L, Mahe
MM, Derkinderen P, des Varannes SB,
Rolli-Derkinderen M. The digestive
neuronal-glial-epithelial unit: a new
actor in gut health and disease. Nat
Rev Gastroenterol Hepatol 2013; 10:
90100.
10 Bush TG, Savidge TC, Freeman TC
et al.
Fulminant
jejuno-ileitis
following ablation of enteric glia in
adult transgenic mice. Cell 1998; 93:
189201.
11 Cornet A, Savidge TC, Cabarrocas J
et al. Enterocolitis induced by auto-

12

13

14

15

16

17

18

19

20

immune targeting of enteric glial

cells: a possible mechanism in Crohns disease? Proc Natl Acad Sci USA
2001; 98: 1330611.
Savidge TC, Newman P, Pothoulakis
C et al. Enteric glia regulate intestinal
barrier function and inflammation
via release of S-nitrosoglutathione.
Gastroenterology
2007;
132:
134458.
Steinkamp M, Geerling I, Seufferlein
T et al. Glial-derived neurotrophic
factor regulates apoptosis in colonic
epithelial cells. Gastroenterology
2003; 124: 174857.
Van Landeghem L, Chevalier J,
Mahe MM et al. Enteric glia promote intestinal mucosal healing via
activation of focal adhesion kinase
and release of proEGF. Am J Physiol
Gastrointest Liver Physiol 2011;
300: G97687.
Flamant M, Aubert P, Rolli-Derkinderen M et al. Enteric glia protect
against Shigella flexneri invasion in
intestinal epithelial cells: a role for
S-nitrosoglutathione. Gut 2011; 60:
47384.
Savidge TC, Sofroniew MV, Neunlist
M. Starring roles for astroglia in barrier pathologies of gut and brain. Lab
Invest 2007; 87: 7316.
Ben-Horin S, Chowers Y. Neuroimmunology of the gut: physiology,
pathology, and pharmacology. Curr
Opin Pharmacol 2008; 8: 4905.
Cabarrocas J, Savidge TC, Liblau RS.
Role of enteric glial cells in inflammatory bowel disease. Glia 2003; 41:
8193.
Costantini TW, Bansal V, Krzyzaniak M et al. Vagal nerve stimulation protects against burn-induced
intestinal injury through activation
of enteric glia cells. Am J Physiol
Gastrointest Liver Physiol 2010;
299: G130818.
Von Boyen GB, Schulte N, Pfluger C,
Spaniol U, Hartmann C, Steinkamp
M. Distribution of enteric glia and

e336

21

22

23

24

25

26

27

28

29

GDNF during gut inflammation.

BMC Gastroenterol 2011; 11: 3.
von Boyen GB, Steinkamp M,
Geerling I et al. Proinflammatory
cytokines induce neurotrophic factor
expression in enteric glia: a key to the
regulation of epithelial apoptosis in
Crohns disease. Inflamm Bowel Dis
2006; 12: 34654.
Geboes K, Collins S. Structural abnormalities of the nervous system in
Crohns disease and ulcerative colitis.
Neurogastroenterol Motil 1998; 10:
189202.
Esposito G, Cirillo C, Sarnelli G et al.
Enteric glial-derived S100B protein
stimulates nitric oxide production in
celiac
disease.
Gastroenterology
2007; 133: 91825.
Knowles CH, Veress B, Kapur RP et al.
Quantitation of cellular components
of the enteric nervous system in the
normal human gastrointestinal tract
report on behalf of the Gastro 2009
International
Working
Group.
Neurogastroenterol Motil 2011; 23:
11524.
Knowles CH, De Giorgio R, Kapur RP
et al. Gastrointestinal neuromuscular pathology: guidelines for histological techniques and reporting on
behalf of the Gastro 2009 International Working Group. Acta Neuropathol 2009; 118: 271301.
Fu YY, Tang SC. At the movies:
3-dimensional technology and gastrointestinal histology. Gastroenterology 2010; 139: 11005.
Smith K. Neurogastroenterology:
improving 3D imaging of the enteric
nervous system. Nat Rev Gastroenterol Hepatol 2011; 8: 600.
Fu YY, Lin CW, Enikolopov G et al.
Microtome-free 3-dimensional confocal imaging method for visualization
of mouse intestine with subcellularlevel resolution. Gastroenterology
2009; 137: 45365.
Fu YY, Tang SC. Optical clearing
facilitates integrated 3D visualization

2013 Blackwell Publishing Ltd

Volume 25, Number 5, May 2013

30

31

32

33

34

of mouse ileal microstructure and

vascular network with high definition. Microvasc Res 2010; 80: 51221.
Fu YY, Lu CH, Lin CW et al. Threedimensional optical method for integrated visualization of mouse islet
microstructure and vascular network
with subcellular-level resolution. J
Biomed Opt 2010; 15: 046018.
Fu YY, Peng SJ, Lin HY, Pasricha PJ,
Tang SC. 3-D imaging and illustration
of mouse intestinal neurovascular
complex. Am J Physiol Gastrointest
Liver Physiol 2013; 304: G111.
Chiu YC, Hua TE, Fu YY, Pasricha PJ,
Tang SC. 3-D imaging and illustration of the perfusive mouse islet
sympathetic innervation and its
remodelling in injury. Diabetologia
2012; 55: 325261.
Liu YA, Chen Y, Chiang AS et al.
Optical clearing improves the imaging depth and signal-to-noise ratio for
digital analysis and three-dimensional projection of the human
enteric nervous system. Neurogastroenterol Motil 2011; 23: e44657.
Liu YA, Chung YC, Pan ST et al. 3-D
illustration of network orientations
of interstitial cells of Cajal subgroups
in human colon as revealed by deeptissue imaging with optical clearing.
Am J Physiol Gastrointest Liver
Physiol 2012; 302: G1099110.

3-D glial network in human colon mucosa

35 Tuchin VV, Wang RK, Yeh AT. Optical clearing of tissues and cells. J Biomed Opt 2008; 13: 021101.
36 Hama H, Kurokawa H, Kawano H
et al. Scale: a chemical approach for
fluorescence imaging and reconstruction of transparent mouse brain. Nat
Neurosci 2011; 14: 14818.
37 Carl Zeiss Microimaging GmbH
(2009) Visualizing the architecture of
cells and tissues, Brochures for Laser
Scanning Microscopy. Available from
http://www.zeiss.de/micro
38 Knowles CH, De Giorgio R, Kapur RP
et al. The London Classification of
gastrointestinal
neuromuscular
pathology: report on behalf of the
Gastro 2009 International Working
Group. Gut 2010; 59: 8827.
39 Ferri GL, Probert L, Cocchia D et al.
Evidence for the presence of S-100
protein in the glial component of the
human enteric nervous system. Nature 1982; 297: 40910.
40 Hoff S, Zeller F, Von Weyhern CW
et al. Quantitative assessment of
glial cells in the human and guinea
pig enteric nervous system with an
anti-Sox8/9/10 antibody. J Comp
Neurol 2008; 509: 35671.
41 Pedica F, Ligorio C, Tonelli P et al.
Lymphangiogenesis in Crohns disease: an immunohistochemical study

42

43

44

45

46

47

using monoclonal antibody D2-40.

Virchows Arch 2008; 452: 5763.
Wu ML, Varga VS, Kamaras V et al.
Three-dimensional virtual microscopy of colorectal biopsies. Arch
Pathol Lab Med 2005; 129: 50710.
Denis B, Peters C, Chapelain C et al.
Diagnostic accuracy of community
pathologists in the interpretation of
colorectal polyps. Eur J Gastroenterol
Hepatol 2009; 21: 115360.
Calhoun BC, Gomes F, Robert ME
et al. Sampling error in the standard
evaluation of endoscopic colonic
biopsies. Am J Surg Pathol 2003; 27:
2547.
Parameswaran L, Prihoda TJ, Sharkey
FE. Diagnostic efficacy of additional
step-sections in colorectal biopsies
originally diagnosed as normal. Hum
Pathol 2008; 39: 57983.
Neunlist M, Van Landeghem L,
Bourreille A, Savidge T. Neuro-glial
crosstalk in inflammatory bowel disease. J Intern Med 2008; 263: 57783.
Neunlist M, Aubert P, Bonnaud S
et al. Enteric glia inhibit intestinal
epithelial cell proliferation partly
through a TGF-beta1-dependent pathway. Am J Physiol Gastrointest Liver
Physiol 2007; 292: G23141.

SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article:
Figure S1. Control experiment of secondary antibody staining for immunohistochemistry.
Figure S2. Control experiment of using a second S100 antibody to verify the mucosal glial network morphology
reported in the result section.
Figure S3. Overlay of the transmitted light and confocal micrographs confirms the fluorescent signals derived
from the CD34+ blood vessels.
Figure S4. Overlay of the transmitted light and confocal micrographs confirms the fluorescent signals derived
from the D2-40+ lymphatic vessels.
Figure S5. Use of coupled immunostaining of anti-a-SMA antibodies derived from different species to verify the
morphology of the peri-cryptic myofibroblasts.
Figure S6. Related to Fig. 2. Imaging of transparent whole mount confirms the peri-cryptic and peri-vascular
morphologies of the glial network in the colon mucosa.
Figure S7. Related to Fig. 7E. High-resolution 2-D and 3-D images of the crypt and S100B morphologies at the
adenoma and normal (apparent) tissue boundary.
Videos S1 and S2. (Related to Fig. 2) Deep-tissue microscopy reveals the peri-cryptic and peri-vascular
morphologies of the glial network. Video 1: in-depth imaging of the transverse mucosal section. Video 2: 360degree projection of the S100B signals. White: S100B. Red: vasculature. Green: nuclei. Depth: 200 lm.
Video S3. (Related to Figure S6) In-depth observation of the glial network in the whole mount mucosal specimen.
The confocal and transmitted light images are merged to associate the tissue microstructure, vasculature, and glial
network in the scanned volume. The result confirms the peri-cryptic and peri-vascular morphologies of the glial
network in colon mucosa. White: S100B. Red: vasculature. Green: nuclei. Depth: 150 lm.

2013 Blackwell Publishing Ltd

e337

Y. A. Liu et al.

Neurogastroenterology and Motility

Video S4. (Related to Fig. 3) In-depth observation and 360-degree projection of the peri-cryptic glia network and
the barrel staves structure of myofibroblasts. This high-resolution image stack reveals the intimate association
among the peri-cryptic glial network, the myofibroblasts, and the epithelium with the glial fibers being occasionally
interposed between the myofibroblasts and the epithelial basement membrane. Blue: nuclei. Magenta: a-SMA.
White: S100B. Transmitted light and confocal micrographs were merged in the first half of the video to highlight the
epithelial basement membrane. Dimensions of the image stack: 143 9 143 9 60 (depth) lm.
Videos S5 and S6. (Related to Fig. 4) 3-D illustration of the glial network at the crypt base. Video 5: 360-degree
projection of enteric glia. Video 6: Merged 3-D projection of enteric glia with blood vessels and the orthogonal slides
of the image stack. White: S100B. Red: vasculature. Green: nuclei. Dimensions of the image stack: 256 9 256 9 80
(depth) lm.
Video S7. (Related to Fig. 5) In depth observation and 360-degree projection of the glial network and its association
with lymphatic vessels at the crypt base and mucosal boundary. White: S100B. Magenta: lymphatic vessels (D2-40).
Green: nuclei. Dimensions of the image stack: 360 9 360 9 150 (depth) lm.
Video S8. (Related to Figs 4 and 5) The enteric glial and myofibroblast populations beneath and around the crypts
(two image stacks). Red: S100B. Magenta: a-SMA. Green: nuclei. Signals of S100B and a-SMA enclose different
nuclei, representing two distinct cellular populations.
Video S9. (Related to Fig. 6) High-resolution image stacks used for quantitation of glial cell density (three
examples). Red: S100B. Green: nuclei. The red and green colors were chosen to highlight their overlap in yellow to
spot the glial cell bodies.
Video S10. (Related to Fig. 7E) 360-degree projection of the S100B signals in adenoma. The irregular crypt
structure is shown in the 2-D micrograph, which is merged with the 3-D projection of S100B signals to indicate the
loss of S100B+ glial fibers in the adenoma. White: S100B. Green: nuclei. Dimensions of the image stack:
360 9 360 9 150 (depth) lm.

e338

2013 Blackwell Publishing Ltd