Download Pancreatic tissue formation from murine embryonic stem cells in vitro

Differentiation (2007) 75:1–11 DOI: 10.1111/j.1432-0436.2006.00109.x
r 2006, Copyright the Authors
Journal compilation r 2007, International Society of Differentiation
O RI G INA L AR T I C L E
Mio Nakanishi . Tatsuo S. Hamazaki . Shinji
Komazaki . Hitoshi Okochi . Makoto Asashima
Pancreatic tissue formation from murine embryonic stem cells in vitro
Received December 15, 2005; accepted in revised form July 6, 2006
Abstract
The in vitro formation of organs and/or tissues is a major goal for regenerative medicine that
would also provide a powerful tool for analyzing both
the mechanisms of development and disease processes
for each target organ. Here, we present a method
whereby pancreatic tissues can be formed in vitro from
mouse embryonic stem (ES) cells. Embryoid body-like
spheres (EBSs) induced from ES cell colonies were
treated with retinoic acid (RA) and activin, which are
candidate regulators of pancreatic development in vivo.
These induced tissues had decreased expression of the
sonic hedgehog (shh) gene and expressed several pancreatic marker genes. ES cell-derived pancreatic tissue
was composed of exocrine cells, endocrine cells, and
pancreatic duct-like structures. In addition, the ratio of
exocrine to endocrine cells in the induced tissue was
found to be sensitive to the concentrations of RA and
activin in the present experiment.
Mio Nakanishi Tatsuo S. Hamazaki Makoto Asashima
. )
(*
Department of Life Science (Biology)
Graduate School of Arts & Science
The University of Tokyo, Meguro
Tokyo 153-8902, Japan
E-mail: [email protected]
. ) Hitoshi Okochi
Tatsuo S. Hamazaki (*
Department of Tissue Regeneration
Research Institute, International Medical Center of Japan
Shinjuku, Tokyo 162-8655, Japan
E-mail: [email protected]
Shinji Komazaki
Department of Anatomy
Saitama Medical School
Iruma, Saitama 350-0495, Japan
Makoto Asashima
ICORP, Japan Science and Technology Agency (JST)
Kawaguchi, Saitama, Japan
U.S. Copyright Clearance Center Code Statement:
Key words mouse embryonic stem cells pancreas b-cells sonic hedgehog pancreatic and duodenal
homeobox 1 (Pdx-1) ptf1a/p48 activin, retinoic acid
Introduction
Mouse embryonic stem (ES) cells can differentiate into
various cell types in vitro (Loebel et al., 2003), including
the insulin-secreting b cells of the pancreas (Soria et al.,
2000; Lumelsky et al., 2001; Hori et al., 2002; Shiroi
et al., 2002; Blyszczuk et al., 2003; Kim et al., 2003;
Leon-Quinto et al., 2004) but not formed any organs or
tissues (Le Douarin, 2000). The promise of transplanting such cells into patients with diabetes has driven
extensive research into factors that induce the differentiation of endocrine cells. However, the pancreas is a
complex organ composed of exocrine cells, endocrine
cells, and pancreatic ducts, which are considered to differentiate from a common precursor cell (Slack, 1995;
Percival and Slack, 1999), making the in vitro formation
of whole pancreas a far more complex undertaking.
Retinoic acid (RA) and transforming growth factor-b
(TGF-b) play key roles in the regulation of pancreatic
organogenesis (Kim and Hebrok, 2001; Kumar and
Melton, 2003). RA signaling is involved in patterning
along the anteroposterior axis of the endoderm during
the late gastrula stage of zebrafish, African clawed frog
(Xenopus laevis), and quail. While inhibition of RA signaling in the late stages had little effect on the expression of anterior mesodermic markers, it abrogated
pancreatic marker expression. In addition, zebrafish
and Xenopus embryos treated with exogenous RA
showed enlargement of the pancreas, and in zebrafish
also of the liver, in the anterior direction (Stafford and
Prince, 2002; Chen et al., 2004; Stafford et al., 2004).
Another important mediator of pancreas differentiation is sonic hedgehog (Shh); its down-regulation by
signals from the notochord is essential for initiating
0301–4681/2007/7501–1 $ 15.00/0
2
differentiation of the dorsal pancreas in mouse and
chicken embryos (Kim et al., 1997). In tissue culture
experiments using endoderm isolated from chicken embryos, activin, a member of the TGF-b family, inhibited
Shh expression and induced transcription of pancreatic
marker genes, thereby mimicking the effects of the notochord signaling (Kim et al., 1997; Hebrok et al.,
1998). Some reports have also suggested that differentiation into pancreatic endocrine cells is regulated by
TGF-b signal transduction mediated via activin receptors (Sanvito et al., 1994; Ritvos et al., 1995; Miralles
et al., 1998; Yamaoka et al., 1998; Shiozaki et al., 1999;
Kim et al., 2000). Culture of the mouse embryonic
pancreas in vitro in the presence of activin or TGF-b1
led to enhancement of endocrine cell formation, particularly of b cells and pancreatic polypeptide cells
(Sanvito et al., 1994), while normal epithelial branching
and the development of exocrine cells were disturbed
(Ritvos et al., 1995). Furthermore, follistatin, an antagonist for TGF-b signaling (including that via activin), was shown to stimulate the pancreas and
differentiate into exocrine cells and reduce the differentiation of endocrine cells (Miralles et al., 1998). In addition, transgenic mice expressing a dominant-negative
form of the type II activin receptor showed hypoplasia
of the islets (Yamaoka et al., 1998; Shiozaki et al.,
1999).
Since both RA and activin have considered to regulate pancreas development, we examined whether treatment with both these factors could induce the
differentiation of ES cells into pancreatic cells.
Colony formation of ES cells (3 days)
on feeder cells in ES medium (15% FBS, +LIF)
to isolate the colonies treated
with collagenase/dispase
Formation of EBSs (4 days)
on non-treated dishes in DMEM (15% KSR, −LIF)
Treatment with RA and activin (2 days)
on non-treated dishes in DMEM (15% KSR) containing RA and activin
Expansion of EBSs
on gelatin coated tissue culture dishes in DMEM (10% KSR)
Fig. 1 Experimental protocol for the differentiation of embryonic
stem (ES) cells into pancreatic tissue with retinoic acid (RA) and
activin. Colonies of ES cells were maintained onto feeder cells and
incubated in ES medium containing LIF for 3 days. These ES colonies were detached from the feeder cells by collagenase/dispase
treatment. The ES cell clusters were incubated in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 15% KnockOut
Serum Replacement (KSR) in floating condition. Four days after
embryoid body-like spheres (EBSs) began to form, they were transferred in a low cell-binding multi-well plate and incubated for additional 2 days in DMEM supplemented with 15% KSR,
containing various concentrations of activin A and all-trans RA.
The EBSs were then attached to the gelatin-coated tissue culture
dish and continued to culture in DMEM supplemented with 10%
KSR until the examination.
Immunohistochemistry
Materials and methods
Maintenance and differentiation of ES cells
Mouse-derived ES cells (E14 and CMTI-1) were seeded onto mouse
embryonic fibroblasts pretreated with mitomycin C (10 ng/ml;
Sigma, St. Louis, MO) for 2.5 hr. The cells were then incubated in
Dulbecco’s modified Eagle’s medium (DMEM; high glucose with Lglutamine and pyruvate; Gibco 11995-065, GIBCO, Invitrogen,
Carlsbad, CA) supplemented with 15% ES cell qualified fetal bovine serum (FBS; Gibco), MEM non-essential amino acid solution
(Gibco), 0.001% b-mercaptoethanol (Sigma), and LIF ESGRO
(1,500 U/ml, Chemicon, Temecula, CA).
For the differentiation of ES cells, 3-day ES cell colonies were
detached from the feeder cells by the treatment of 1 mg/ml of collagenase/dispase (Roche Diagnostic, Indianapolis, IN) (Fig. 1). The
clusters of ES cells were transferred into a low cell-binding dish
(Nalge Nunc, Rochester, NY). They were cultured at floating condition in DMEM supplemented with 15% KnockOut Serum Replacement (KSR; Gibco) and the culture medium was renewed 2
days later. Four days after embryoid body-like spheres (EBSs)
began to form, they were transferred into a low cell-binding 96-well
plate (Nunc) and incubated for additional 2 days in DMEM supplemented with 15% KSR containing activin A (0, 10, 25, 50 ng/ml)
and all-trans RA (0, 0.001, 0.01, 0.1, or 1 mM; Sigma). The EBSs
were then attached to the wells of tissue culture plates or dishes
(TPP) that had been coated overnight with 0.1% gelatin (Sigma).
The cells were incubated in DMEM supplemented with 10% KSR,
and the medium was renewed every third day.
Thirteen days after the end of treatment (19 days after the beginning
of EBS formation), the EBSs were fixed in 4% paraformaldehyde in
0.1 M phosphate buffer (pH 7.4) for 40 min at room temperature. The
fixed cells were freed from the dish wall and embedded in an LR Gold
Resin System (structure probe, Electron Microscopy Sciences,
Hatfield, PA), an acrylic resin, and sectioned to yield 600 nm semithin sections. The sections were blocked for 40 min at room temperature in 3% bovine serum albumin (BSA) and phosphate-buffered
saline (PBS). They were then exposed to pancreas-specific primary
antibodies for 12 hr at 41C. After washing with PBS, the sections were
exposed to labeled secondary antibody for 8 hr at 41C. The primary
antibodies used were rabbit anti-a amylase antibody (1:1,000 dilution,
Sigma), anti-insulin monoclonal antibody (1:400 dilution, Sigma),
goat anti-C-peptide antibody (1:800 dilution, Linco Research,
Millipore, Billerica, MA), rabbit anti-Pdx-1 antibody (1:200 dilution, Chemicon); the secondary antibodies were Alexa-Fluor 488and Alexa-Fluor 594-conjugated (Molecular Probes, Invitrogen,
Carlsbad, CA). The sections were observed under a fluorescence
microscope and photographed with AquaCosmos (Hamamatsu
Photonics, Hamamatsu, Japan) connected to an ORCA-3CCD
camera. For the control, pancreas from 8-week-old mouse was fixed
and processed as described above. For smooth muscle detection,
EBSs were fixed as described above, blocked with 3% BSA in PBS
for 40 min at room temperature, and permeabilized with 0.1% Triton X-100 for 30 min. They were then exposed to fluorescein isothiocyanate (FITC)-conjugated monoclonal anti-a-smooth muscle
actin antibody (1:500 dillution; Sigma) for 12 hr at 41C.
To calculate the C-peptide-positive cells in the induced EBS, 1 mm
serial sections were prepared from each specimen, and examined
at every 20 mm thickness by immunostained with anti-C-peptide
3
antibody. Then they were calculated the percentage of the C-peptide-positive cells with image analyzing software (Luminavision).
Electron microscopy
Eleven days after the end of treatment (17 days after EBS formation), the EBSs were prefixed in a fixative containing 4% paraformaldehyde, 3% glutaraldehyde, and 0.1 M cacodylate buffer
(pH 7.4) for 2 hr at room temperature. Then, after washing with
0.1 M cacodylate buffer, they were post-fixed in 1% osmium tetroxide for 30 min at room temperature. The samples were washed
again in 0.1 M cacodylate buffer, then dehydrated through an
ethanol and acetone series, and embedded in Epoxy resin. Ultrathin
(80–90 nm) sections were prepared and stained with uranium acetate and lead citrate for observation under a transmission electron
microscope (JEM-1200CX, JEOL, Tokyo, Japan).
7700 Sequence Detector (Perkin Elmer). The forward and reverse
primers used and the length of their PCR products were as follows:
glucagon, GCA CAT TCA CCA GCG ACT ACA G and GGG
AAA GGT CCC TTC AGC ATG TCT (146 bp); somatostatin,
CGA GCC CAA CCA GAC AGA GA and CAT TGC TGG GTT
CGA GTT GG (115 bp); amylase 2, ATA CTC TGC TTG GGA
CTT TAA CGA and CAG AAG GCC AGT CAG ACG A
(100 bp); GAPDH, GCT ACA CTG AGG ACC AGG TTG TC
and AGC CGT ATT CAT TGT CAT ACC AGG (135 bp); insulin
II, AGA AGC GTG GCA TTG TAG ATC AGT and CAG AGG
GGT AGG CTG GGT AGT G (102 bp); Pdx-1, GAT GAA ATC
CAC CAA AGC TCA CGC and GGG TGT AGG CAG TAC
GGG TCC TC (101 bp); ptfla, ATG CAG TCC ATC AAC GAC
GCC TTC and GGC TTG CAC CAG CTC GCT GAG (138 bp);
Shh, GAC TGC GGG CAT CCA CTG GTA CTC and GTC GGG
CTT CAG CTG GAC TTG AC (114 bp). cDNA was amplified by
15 min initial denaturation at 951C, 45 cycles of heating at 941C
(30 sec), 551C (30 sec), and 721C (30 sec). This experiment was also
carried out in triplicate.
Reverse transcription-polymerase chain reaction (RT-PCR)
Total RNA was extracted from 16 EBSs to normalize the gene
expression among the EBSs, using an ISOGEN (Nippon Gene Co.,
Ltd., Tokyo, Japan). Complementary DNA (cDNA) was synthesized from 1 mg total RNA using a SuperScript First-Strand Synthesis System for RT-PCR (Invitrogen, Carlsbad, CA). The
forward and reverse primers used and the length of their PCR
products were as follows: albumin (Yamada et al., 2002), TGA ACT
GGC TGA CTG CTG TG and CAT CCT TGG CCT CAG CAT
AG (718 bp); a-phetoprotein (Afp), CCA CCC TTC CAG TTT
CCA G and GGG CTT TCC TCG TGT AAC C (609 bp); amylase
2, GCC AAG GAA TGT GAG CGA TAC TTA and CCA GAA
GGC CAG TCA GAC GA (418 bp); glucose-6-phosphatase (G6p)
(Ishii et al., 2005), TGC ATT CCT GTA TGG TAG TGG and
GAA TGA GAG CTC TTG GCT GG; glucagon, AAT GAA GAC
AAA CGC CAC T and AAT TCA TAT ACA ATC GTT GGG
TTA (554 bp); insulin I, TTA CAC ACC CAA GTC CCG CCG
TGA and AGG GGT GGG GCG GGT CGA G (207 bp); insulin
II, GGC TTC TTC TAC ACA CCC ATG TCC and TTT ATT
CAT TGC AGA GGG GTA GGC (234 bp); Pdx1, AGC AGT
CTG AGG GTG AGC GGG TCT and AAC CTC CAA CAG
CCG CCT TTC GT (412 bp); somatostatin, CCC AGA CTC CGT
CAG TTT CT and TCA ATT TCT AAT GCA GGG TCA AGT
(377 bp); Brachuyry (T), GAA TTC GTC CAC CCC CTG TCC
TAC and CAA GGG CAG AAC AGT TGA CGG TT (146 bp); b3
tubulin (Tubb3), GTC CTA GAT GTC GTG CGG AAA G and
GGA TGT CAC ACA CGG CTA CCT (724 bp); pancreatic polypeptide, GCC CAA CAC TCA CTA GCT CAG and AGA GGA
AAG AGC TGG ACC TGT ACT (419 bp); Shh, ACA TGT CCC
TTG TCC TGC GTT TCA and CTC GTG GGC TCG CTG CTA
GGT (440 bp); Sox17, GGC CAG AAG CAG TGT TAC ACA
and TTT GAT AAA AAT CGA TGC GAG AGA (336 bp); tyrosine amino transferase (Tat) TCC AGG AGT TCT GTG AAC
AGC and AGT ATA TGG TGC CTG CCT GC; GAPDH, TGA
AGG TCG GTG TGA ACG GAT TTG GC and CAT GTA GGC
CCA TGA GGT CCA CCA C. cDNA was amplified by a 5 min
initial denaturation at 951C, 40 cycles of heating at 941C (20 sec),
541C (20 sec), and 721C (40 sec) and finally, 7 min elongation at
721C. The experiment was carried out in triplicate. To determine
the inductive efficiency of pancreatic tissue from the EBSs, we
examined the expression of pancreatic marker gene Pdx-1 in each
EBS separately (n 5 15 for each treatment). For the examination of
gene expression, more than 0.7 mg of total RNA was obtained from
each EBS and equal amount (0.7 mg) was used for the RT-PCR.
Quantitative PCR
Real-time PCR was carried out using a QuantiTectt SYBR Green
PCR Master Mix (QIAGEN, Valencia, CA) and ABI PRISMt
Results
Differentiation of gut-like and pancreatic tissues from
ES cells
First, ES cell colonies were isolated after 3 days in culture by collagenase/dispase treatment and were then
cultured in DMEM supplemented with 15% KSR to
form EBSs, which can initiate differentiation. Four days
after the induction of EBS formation, the cells were
transferred into a medium containing RA (0, 0.001,
0.01, 0.1, or 1 mM) and activin (10 ng/ml) and incubated
for additional 2 days by floating culture, and then seeded on gelatin-coated plates. Six to 8 days after the end
of treatment with RA and activin (12–14 days after the
commencement of EBS formation), a proportion of the
treated EBSs showed tubular or gut-like structures and
a motility that resembled smooth-muscle peristalsis.
Immunostaining with anti-a-smooth muscle actin antibody demonstrated that abundant smooth muscle cells
surrounded these tubular or gut like structures (Figs.
2C–2E). Nine to 12 days after the end of treatment
(15–18 days after the commencement of EBS formation), over 20% of the EBSs treated with both RA and
activin formed a tissue that contained black spots,
which were rarely observed in the untreated EBSs or
EBSs treated with activin alone (Figs. 2A,2B).
Examination of expression of pancreatic marker genes
As an approach to examine differentiation state of the
EBSs at the onset of the treatment, we monitored the
gene expression patterns of b3-tubulin (Tubb3), Brachyury (T) and Sox17 as early development markers of
neuroectoderm, mesoderm, and endoderm, respectively
(Fig. 3A). From 3 days after the induction of EBSs
formation (a day before the onset of RA and activin
treatment), the expressions of Sox17 and T were detected (Fig. 3A and data not shown). In contrast, the
4
A
B
g
d
D
C
E
g
g
(bar = 50 µm)
expression of Tubb3 was not detected until 2 days after
the finish of the treatment. In the EBSs treated with RA
and activin, the expression of Sox17 was clearly observed. Whereas no differences was observed in the expressions of T and Tubb3 between treated EBSs and
untreated EBSs. Sox17 was also detected in the EBSs
which were treated with activin alone, but it did not
express when they were treated with RA alone. The
expression patterns of genes that regulate the development of the pancreas and several pancreatic marker
genes were also investigated using RT-PCR (Fig. 3A).
The EBSs treated with both RA and activin showed a
marked increase in the gene expression of insulin I, insulin II (markers of b cells), glucagons (a cells), pancreatic polypeptide (g cells), somatostatin (d cells), and
amylase 2 (a marker of pancreatic exocrine cells), as well
as of Pdx-1 and ptf1a/p48, which are genes that regulate
pancreatic development. In contrast, the untreated
EBSs showed almost no expression of these genes.
The expression of amylase and pancreatic polypeptide
increased over time from day 6 of treatment (12 days
after the commencement of EBS formation), while the
gene expression of insulin I, insulin II, glucagon, somatostatin, Pdx-1 and ptf1a/p48 commenced immediately after the end of the treatment. In untreated EBSs,
Shh was detected throughout the observation period,
while in the EBSs treated with both RA and activin, Shh
gene expression was suppressed following the treatment.
In the EBSs treated with 0.1 mM RA alone, the expression of glucagon, insulin I, insulin II, and pancreatic
polypeptide began to be detected but shh regressed similar to the EBSs treated with the same concentration of
RA and 10 ng/ml of activin A. The expression of other
Fig. 2 Phase contrast images of embryoid
body-like spheres (EBSs) containing differentiated pancreatic tissue. (A, B) EBS on
day 13 after treatment with 0.1 mM of
retinoic acid (RA) and 10 ng/ml of activin
showed a gut-like structure (g) and pancreatic duct-like structures (d) adjacent to
exocrine cells which were seen as many
black spots (B, arrowheads). (C–E)
Immunostaining of EBSs treated with
0.1 mM RA and 10 ng/ml activin prepared
11 days post-treatment demonstrated that
abundant smooth muscle surrounded the
gut-like structures (g) in the EBSs. Phase
contrast (C), anti-smooth muscle-a-actin
antibody staining (D), and merged image
of the EBS (E). (Bar 5 50 mm [A and B]).
pancreatic markers, amylase 2 and Pdx1, were also detected but their expression seemed to be weaker than
those in the EBSs treated with RA and activin, and the
expression of somatostatin was not detected. In the
EBSs treated with 10 ng/ml activin A alone, the expression of glucagon and Pdx1 was detected. However, they
did not express other pancreatic marker genes (Amylase2, insulin I, insulin II, pancreatic polypeptide, and somatostatin) and shh expression was detected 10 days
after the end of the treatment.
To examine whether the gene expression level analyzed by RT-PCR were reflected to the mRNA
amounts, we performed quantitative PCR using glucagon and somatostatin (Fig. 3B). Expression levels of
both genes elevated 2 days after the treatment and increased 4 days later. It was found that the RT-PCR
results were almost compatible with that of quantitative
PCR.
When the concentration of activin kept constant
(10 ng/ml) and the RA concentration was changed, the
gene expressions of pancreatic marker in EBSs appeared most efficiently when 0.1 mM RA was used with
10 ng/ml activin A (Fig. 3C). Pdx-1 expression was detected in 47.6% 7.3% of the EBSs treated with these
conditions. When the concentration of RA were 0.001,
0.01, and 1 mM, the percentage of EBSs expressing
Pdx-1 were 4.3% 2.5%, 12.8% 9.6%, and 25.1% 6.6%, respectively.
We attempted to determine whether hepatic differentiation was also induced in this system by the examination of expression patterns of Albumin, a-phetoprotein
(AFP), glucose-6-phosphatase (G6P), tyrosine amino
transferase (TAT) used as hepatic marker genes. As a
A
Control (untreated)
EBS 2d
4d 6d 8d 10d
0.1 µM RA
10 ng/ml
Activin
5
0.1 µM RA + 10 ng/ml Activin
10d 10d EBS 2d
4d
6d
8d 10d
Insulin II
Insulin I
Glucagon
Somatostatin
PP
Amylase 2
Pdx-1
Shh
GAPDH
Tubb3
T
Sox17
CK19
glucagon
200
C
150
100
50
0
EBS
2d
4d
6d
8d
10d
somatostatin
50
40
30
20
10
0
% of EBSs expressing
Pdx-1gene
Relative mRNA expression
B
60
50
40
30
20
10
0
0.001
0.01
0.1
1
Retinoic acid concentration (µM)
EBS
2d
4d
6d
8d
10d
result, no hepatic marker genes were detected by RTPCR in the EBSs treated with both RA and activin,
except the AFP expression was faintly detected 4 days
after the end of the treatment (data not shown).
Morphological identification of ES cell-derived
pancreas tissue
Histology of sections of EBSs treated with 0.1 mM RA
and 10 ng/ml activin 11 days after the end of treatment
(17 days after EBS formation) showed lobular tissue
that was intensely stained with toluidine blue in the area
adjacent to the tubular lumen (Fig. 4B). Pancreatic
duct-like structures were seen in the vicinity of the lobular tissue (Figs. 4C,4D). In contract, neither lobular
tissues nor duct-like structures were observed in the
sections of untreated EBSs (Fig. 4A). Electron microscopic observation revealed that the cells had a welldeveloped endoplasmic reticulum and contained large
amounts of zymogen granules (Figs. 5C,5D), a phenotype consistent with pancreatic exocrine cells (acinar
Fig. 3 The gene expression after induction
of differentiation. (A) Expression of pancreatic marker genes examined by reverse
transcription-polymerase chain reaction
(RT-PCR) in the untreated embryoid
body-like spheres (EBSs), and the EBSs
treated with either or both retinoic acid
(RA) and activin. The expression of insulin II, glucagon, somatostatin, and Pdx1
began to be detected on day 2 post-treatment, while amylase 2 and pancreatic
polypeptide expression was on days 6
and 8, respectively. Expression of sonic
hedgehog (Shh) diminished after treatment. In the EBSs treated with RA and
activin, the expression of Sox17 began to
be clearly detected after the treatment,
whereas no differences between the treated EBSs and the untreated EBSs were
observed in the expression of T and the
Tubb3. (B) Quantitative PCR confirmed
the RT-PCR results. The gene expression
of both glucagon and somatostatin were
up-regulated 2–4 days after treatment and
kept high level until 10 days of the treated
EBSs ( & ) compared with those of the
non-treated EBSs (&). (C) The most efficient gene expression of pancreatic
marker Pdx1 in EBSs was observed when
0.1 mM RA was used with 10 ng/ml activin
A. At these concentrations of RA and activin, 47.6% 7.3% of the treated EBSs
expressed Pdx1. Other concentrations did
not induce pancreatic gene expressions
effectively.
cells). The tissue formed from the treated EBSs resembled pancreatic acini also in that a narrow tubular area
was seen abutting the tissue (Fig. 5C). A structure surrounded by low columnar epithelium and resembling a
pancreatic duct-like structure was also present (Fig.
4D). Furthermore, cells morphologically resembling
pancreatic endocrine cells (Figs. 5E,5F) were observed,
although these were smaller in number than the cells
resembling pancreatic exocrine cells. These endocrinelike cells contained granules (Fig. 5F), although they
were morphologically different from the granules found
in the exocrine cells (Fig. 5D) and displayed a clear
space between the dark central area and the lateral
membrane, considered a feature of typical b cell granules in the pancreas. To confirm whether these granules
contained insulin or amylase, the cells were immunostained with anti-amylase and anti-insulin C-peptide
(proinsulin) antibodies. The EBSs treated with both RA
and activin contained cells with amylase-positive zymogen granules (Fig. 5A) and a relatively small number of
cell aggregates with C-peptide-positive granules (Fig.
5B). Immunohistochemical staining also showed that
6
A
B
d
d
C
D
n
d
d
Pdx-1-positive cells were localized around the duct-like
structures in the EBSs treated with RA and activin A
(Fig. 5G). Immunoreactivity of Pdx-1 was localized
mainly in cytoplasm and partially in nucleus (Fig. 5H).
These morphological findings suggested that pancreatic
tissue containing exocrine cells, endocrine cells, and
pancreatic ducts were formed from EBSs treated with
both RA and activin.
The concentration of activin alters the ratio of exocrine
to endocrine cells
We further examined whether the inductive effect of
activin and RA on the EBSs to form pancreatic tissues
was concentration-dependent. To address this, total
RNA was extracted from the EBSs 13 days after treatment with RA (0, 0.1, or 1 mM) and activin (0, 10, or
25 ng/ml) and real-time PCR was performed to measure
the transcriptional expression of amylase 2, insulin II,
ptf1a/p48, and Shh (Figs. 6A–6D) compared with that
expressed in untreated EBSs (normalized to a value of
1). Shh expression was reduced by treatment with RA
alone, activin alone, as well as by RA plus activin (Fig.
6D). The expression of insulin II and amylase 2 was
greater in EBSs treated with RA alone than in the untreated EBSs, and activin alone had no significant effect
on the level of insulin II expression but decreased amylase 2 expression compared with the untreated EBSs
(Figs. 6A,6B). Interesting results were obtained when
the concentration of RA was kept constant (0.1 mM)
n
Fig. 4 Morphology of embryoid
body-like spheres (EBSs) containing differentiated pancreatic tissue.
Sections of EBSs treated with
0.1 mM retinoic acid and 10 ng/ml
activin prepared 11 days post-treatment showed lobular tissue intensely stained with toluidine blue
adjacent to tubular lumina (B, arrowheads), and pancreatic ductlike structures (d) in the vicinity
of the lobular tissue (C, D). A
structure resembling a pancreatic
duct, surrounded by low columnar
epithelium, was also observed (D).
In the sections of untreated EBS,
neither lobular tissues nor duct-like
structures were observed (A). n,
nucleus. Scale bars: 50 mm (A–C)
and 5 mm (D).
and only the concentration of activin was changed. At
relatively low concentrations of activin (10 ng/ml), the
levels of amylase 2 and ptf1a/p48 expression (markers of
exocrine cells) were much higher in the RA plus activintreated groups than in the untreated EBSs, while the
expression of insulin II did not differ between the two
groups (Figs. 6A–6C). In contrast, at a higher activin
concentration (25 ng/ml), the expression of insulin was
markedly increased in the EBSs treated with RA plus
activin relative to the untreated group, whereas low the
expression levels of amylase 2 were observed in the two
groups.
The quantitative PCR results were supported by
double staining using anti-insulin C-peptide and antiamylase antibodies (Figs. 7A–7D). In EBSs treated with
0.1 mM RA and a low concentration (10 ng/ml) of activin, the number of insulin-positive cells were fewer
than amylase-positive cells (Fig. 7C). However, when a
high concentration of activin (25 ng/ml) was used with
0.1 mM RA, the percentage of insulin-positive cells
markedly increased (Fig. 7D). In the untreated EBSs,
insulin-positive cells and amylase-positive cells were not
observed (Fig. 7B). These results suggest that the ratio
of endocrine cells to exocrine cells in the induced tissue
from the EBSs was dependent on the activin concentration.
Immunocytochemical analysis also revealed that
some of the cells produced the a cell marker protein,
glucagon (Fig. 7E). We tried to estimate C-peptidepositive cells in each EBS. Serial sections (1 mm) were
cut and examined. As a result, three of 10 EBSs treated
7
A
B
Amylase/DAPI
C-peptide/ DAPI
C
D
er
n
z
n
n
z
z
n
n
z
E
F
m
m
n
n
G
H
with 0.1 mM RA and 25 ng/ml activin A were found to
contain C-peptide-positive cells. In these C-peptidepositive EBSs, 4.99% 0.59% of total cells contained
C-peptide (Table 1). By the evaluation of immunostaining with anti-C-peptide antibody, 30% of EBSs treated
with RA and activin showed pancreatic features.
Discussion
While a number of studies have isolated pancreatic b
cells from ES cells, here we have shown that treatment
Fig. 5 Immunohistochemical and ultrastructural analyses of embryoid body-like
spheres (EBSs) containing differentiated
pancreatic tissue. (A, B) Immunostaining
of EBSs treated with 0.1 mM retinoic acid
and 10 ng/ml activin revealed cells containing amylase-positive zymogen granules (A) and a few cells containing insulin
C peptide-positive granules (B). Cells
were counterstained with DAPI (4 0 ,6-diamidino-2-phenylindole) (A, B). Electron
microscopic observation revealed cells
containing large amounts of zymogen
granules (C, D), well-developed endoplasmic reticulum, and showed dark and
homogeneous staining, which are characteristic of acinar cells. A narrow tubular
area seen abutting the tissue was also
reminiscent of pancreatic acini (C). Cells
morphologically resembling pancreatic
endocrine cells were also noted (E, F), although they were smaller in number than
the cells resembling pancreatic exocrine
cells. These cells contained large amounts
of endocrine granules (F), which were
morphologically different from the granules contained in the exocrine cells (D). A
large space between the dark central area
and the lateral membrane, consistent with
the b cell granules of the pancreas, was
also observed (F). (G) Pdx-1-positive cells
were observed around the duct-like structures. The localization of Pdx-1 was
found mainly in cytoplasm and partially
in nucleus (H) (er, endoplasmic reticulum;
m, mitochondrion; n, nucleus; z, zymogen
granule). Scale bars: 20 mm (A, B, G, H),
5 mm (C–E), 1 mm (F).
of EBSs with RA and activin by floating culture can
induce differentiation of complex and functional pancreas that includes all endocrine (a, b, g, and d) cells,
acinar cells, and pancreatic duct-like structures. ES cells
that spontaneously differentiated into insulin-producing
cells in culture have been identified (Loebel et al., 2003;
Blyszczuk and Wobus, 2004), and isolated by the
gene-trap method (Soria et al., 2000) or by the use of
dithizone, a zinc chelator that specifically stains b cells
(Shiroi et al., 2002). However, these cells only had
minimal expression of a few pancreatic markers and no
functional activity, such as glucose concentrationdependent release of insulin (Shiroi et al., 2002).
8
Amylase 2
A
B
14
Activin A
0 ng/ ml
10 ng/ ml
10
25 ng/ ml
8
6
4
140
Relative mRNA expression
Relative mRNA expression
12
Insulin II
160
2
Activin A
0 ng/ ml
120
10 ng/ ml
25 ng/ ml
100
80
60
40
20
0
0
0
0.1
1.0
0
0.1
1.0
Concentration of retinoic acid (µM)
C
Ptf1a /p48
3
D
1.2
Shh
Activin A
10 ng /ml
25 ng /ml
2
1.5
1
0.5
0 ng/ ml
1
Relative mRNA expression
Relative mRNA expression
2.5
Activin A
0 ng / ml
10 ng/ ml
25 ng/ ml
0.8
0.6
0.4
0.2
0
0
0.1
0
1.0
0
Concentration of retinoic acid (µM)
Insulin-secreting cells were generated by selection of
cells expressing nestin from spontaneously differentiated ES cells and subsequent treatment of these
cells with factors such as nicotinamide (Lumelsky
et al., 2001; Hori et al., 2002; Blyszczuk et al., 2003;
Shi et al., 2005). Nestin, an intermediate filament
protein, is expressed in mesenchymal cells associated
with the pancreas, but this protein was not detected
in pancreatic endocrine precursor cells (Selander and
Edlund, 2002), nor is it expressed in endocrine
cells (Delacour et al., 2004). It is therefore unclear whether such induction steps would serve as reliable models for the differentiation of pancreatic cells
in vivo.
A recent study reported that EBs that formed in
the presence of serum could differentiate into b cells
by the sequential treatment of activin, RA, and nicotinamide (Shi et al., 2005). In the present study, we first
focused on the formation of pancreatic tissues containing all three components: endocrine cells, exocrine
0.1
1.0
Fig. 6 Expression of pancreatic marker genes
and sonic hedgehog (shh) in embryoid body-like
spheres (EBSs). Total RNA was collected from
the EBSs 13 days after the end of treatment with
retinoic acid (RA; 0, 0.1, or 1 mM) and activin (0,
10, or 25 ng/ml), and the expression levels of
amylase 2 (A), insulin II (B), ptf1a/p48 (C), and
Shh (D) were measured using real-time PCR.
Expression levels were normalized to those in the
untreated EBSs. Shh expression was reduced by
all combinations of the treatment with RA alone
and activin alone, as well as by RA plus activin
(C). Increased amounts of insulin II and amylase
2 were observed in EBSs treated with RA alone,
whereas activin alone had no such effect on insulin expression compared with untreated EBSs.
Activin alone, however, did induce a decrease in
amylase 2 expression (A, B). When the EBSs
were treated with a constant concentration of
RA (0.1 mM; middle of groups in A, B) and a
relatively low concentration of activin (10 ng/
ml), the expression of amylase 2 was much
greater in the RA plus activin-treated group
than in the untreated group, while the expression
of insulin II did not differ between the two
groups. In contrast, RA in combination with
high activin concentrations (25 ng/ml) induced
an approximately 150-fold increase in the expression of insulin II, whereas relatively low levels of amylase 2 expression were observed. The
expression patterns of ptf1a/p48 strongly correlated with those of amylase 2.
cells, and duct-like structures from ES cells in vitro by
floating cultures of the EBSs treated with both RA
and activin. There was a marked increase in the expression of both exocrine and endocrine pancreatic
cell marker genes and a decreased expression of the
Shh gene, a pattern consistent with the known features
of a developing pancreas in vivo. Furthermore, the
timing of expression of these marker genes was consistent with previous findings which demonstrated that
the increased expression of endocrine marker genes
precedes the up-regulation of exocrine marker genes
during pancreas development in vivo (Gittes and Rutter,
1992). Ultrastructural analysis identified two morphologically different types of zymogen granule in the
treated EBSs: amylase-positive granules that showed
features consistent with those found in pancreatic
exocrine cells (acinar cells); and the granules that
were immunoreactive to anti-C-peptide antibody that
are found in pancreatic endocrine cells. Furthermore,
a narrow tubular area was observed abutting the
9
A
B
C
D
Amylase/ C-peptide/ DAPI
E
Glucagon/
DAPI
Fig. 7 Effect of activin concentration on the appearance of exocrine
and endocrine cells. Pancreas of 8-week-old mouse (A), untreated
embryoid body-like spheres (EBSs) (B), and EBSs treated with
0.1 mM retinoic acid (RA) and either 10 ng/ml activin (C) or 25 ng/
ml activin (D, E) were stained with anti-insulin C peptide antibody,
anti-amylase antibody and DAPI (4 0 ,6-diamidino-2-phenylindole),
13 days after the end of treatment (19 days after the commencement
of EBS formation). No insulin-positive cells or amylase-positive
cells were observed in the untreated EBSs (B). In the EBSs treated
with 0.1 mM RA and 10 ng/ml activin, there were fewer insulinpositive cells than amylase-positive cells (C). In contrast, EBSs
treated with 0.1 mM RA and 25 ng/ml activin markedly increased
the insulin-producing cells (D). Moreover, anti-glucagon antibody
staining revealed a small but significant number of glucagon-producing cells (E). Scale bar: 50 mm (A–D) and 20 mm (E).
exocrine cells, a structure that resembled the pancreatic ducts surrounded by the interposed area and
low columnar epithelium. Immunohistochemical analysis revealed that Pdx-1-positive cells were localized
in these duct-like structures. Although Pdx-1 was considered to be localized in nucleus of mature pancreatic cells, it was observed mainly in the cytoplasm and partially in the nucleus of duct-like cells.
This result probably indicates that the immature
pancreatic ducts are formed in the present inductive
system.
RT-PCR and quantitative PCR identified different
contributions of RA and activin to the induction of
pancreas. Both factors could suppress Shh expression
alone or in combination. RA was considered to be essential for the induction of gene expression of pancreatic markers amylase 2, insulin II, glucagon, Pdx-1, and
Ppy. In contrast, activin may regulate differentiation
and/or the proliferation of pancreatic endocrine and
exocrine cells, since both insulin and amylase expression
levels changed in an activin concentration-dependent
manner in the EBSs treated with RA and activin. Recently, activin A has been reported to induce endoderm
differentiation from both human and mouse ES cells
(Kubo et al., 2004; D’Amour et al., 2005; Yasunaga
et al., 2005). In the present study, the expression level of
early endoderm development marker Sox17 was significantly elevated in the EBSs treated with activin A or
both activin A and RA than the untreated EBSs or
treated with RA alone. Thus, it also showed that activin
A has a role for endoderm induction in this experiment.
On the other hand, when the concentration of RA was
kept constant (0.1 mM) and only the concentration of
activin was changed, the low concentrations of activin
(10 ng/ml) treatment induced a much higher gene expression level of amylase 2 than that of untreated EBSs,
while the expression of insulin II was not different. In
contrast, at a higher activin concentration (25 ng/ml),
the expression of insulin was markedly increased in the
EBSs treated with both activin A and RA than the untreated group, whereas low expression of amylase 2
were observed in the two groups. These results suggested another possible role for activin in the differentiation
and/or the proliferation of pancreatic endocrine and
exocrine cells, concomitant with RA. It is true that the
pancreatic b cells induced by our method occupy a
small fraction of total induced tissues compared with
other studies (Blyszczuk et al., 2003; Kania et al., 2004;
Miyazaki et al., 2004), and that the efficiency of the
induction system needs to be improved in this aspect.
However, it is important to note that the differentiation
system presented here was not designed for the in vitro
production of a pure b cell population, but rather for
the formation of pancreas-like tissues containing most
of all components of intact pancreas. To this end, it was
an interesting finding that pancreas-like tissue could be
successfully induced from mouse ES cells using a similar
treatment regime to that reported by us previously for
the induction of pancreas from the presumptive endoderm of frog embryos (Moriya et al., 2000). In conclusion, we have developed a relatively simple approach
for inducing the differentiation of ES cells into pancreas-like tissue in vitro, which will provide a good
model for analyzing the mechanisms of pancreatic
development.
10
Table 1 The percentage of C-peptide-positive areas in the EBSs
RA1activin treated
Control
C-peptide-positive area in each
EBS (average of positive EBSs, %)
C-peptide-positive area in each
EBS (average of total EBSs, %)
Number of EBSs (containing
positive signal/total tested)
4.99 0.59
0
1.50 0.78
0
3/10
0/3
To estimate the C-peptide-positive cells in each EBS, serial sections (1 mm thick) of EBSs were made and each section of every 20 mm
thickness was stained with anti-C-peptide antibody. Then the C-peptide-positive cells were counted against whole cells in EBs. In the EBSs
treated with 0.1 mM RA and 25 ng/ml activin A, three of 10 EBSs contained C-peptide-positive cells. It was found that the C-peptidepositive EBSs, 4.99% 0.59% of total cells were C-peptide positive.
RA1activin treated: the EBSs treated with RA (0.1 mM) and activin (25 ng/ml).
EBSs, embryoid body-like spheres; RA, retinoic acid.
Acknowledgments We thank Dr. D.A. Melton, Howard Hughes
Medical Institute, Harvard University, for his kind comment and
suggestions throughout this work. This work was supported by a
Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Sport, Culture and Technology, and from the Ministry of Health, Labor and Welfare of Japan.
References
Blyszczuk, P., Czyz, J., Kania, G., Wagner, M., Roll, U., St-Onge,
L. and Wobus, A.M. (2003) Expression of Pax4 in embryonic
stem cells promotes differentiation of nestin-positive progenitor
and insulin-producing cells. Proc Natl Acad Sci USA 100:998–
1003.
Blyszczuk, P. and Wobus, A.M. (2004) Stem cells and pancreatic
differentiation in vitro. J Biotechnol 113:3–13.
Chen, Y., Pan, F.C., Brandes, N., Afelik, S., Solter, M. and Pieler,
T. (2004) Retinoic acid signaling is essential for pancreas development and promotes endocrine at the expense of exocrine cell
differentiation in Xenopus. Dev Biol 271:144–160.
D’Amour, K.A., Agulnick, A.D., Eliazer, S., Kelly, O.G., Kroon,
E. and Baetge, E.E. (2005) Efficient differentiation of human
embryonic stem cells to definitive endoderm. Nat Biotechnol
23:1534–1541.
Delacour, A., Nepote, V., Trumpp, A. and Herrera, P.L. (2004)
Nestin expression in pancreatic exocrine cell lineages. Mech Dev
121:3–14.
Gittes, G.K. and Rutter, W.J. (1992) Onset of cell-specific gene
expression in the developing mouse pancreas. Proc Natl Acad Sci
USA 89:1128–1132.
Hebrok, M., Kim, S.K. and Melton, D.A. (1998) Notochord repression of endodermal Sonic hedgehog permits pancreas development. Genes Dev 12:1705–1713.
Hori, Y., Rulifson, I.C., Tsai, B.C., Heit, J.J., Cahoy, J.D. and
Kim, S.K. (2002) Growth inhibitors promote differentiation of
insulin-producing tissue from embryonic stem cells. Proc Natl
Acad Sci USA 99:16105–16110.
Ishii, T., Yasuchika, K., Fujii, H., Hoppo, T., Baba, S., Naito, M.,
Machimoto, T., Kamo, N., Suemori, H., Nakatsuji, N. and Ikai,
I. (2005) In vitro differentiation and maturation of mouse
embryonic stem cells into hepatocytes. Exp Cell Res 309:68–77.
Kania, G., Blyszczuk, P. and Wobus, A.M. (2004) The generation
of insulin-producing cells from embryonic stem cells – a discussion of controversial findings. Int J Dev Biol 48:1061–1064.
Kim, D., Gu, Y., Ishii, M., Fujimiya, M., Qi, M., Nakamura, N.,
Yoshikawa, T., Sumi, S. and Inoue, K. (2003) In vivo functioning
and transplantable mature pancreatic islet-like cell clusters
differentiated from embryonic stem cell. Pancreas 27:e34–e41.
Kim, S.K. and Hebrok, M. (2001) Intercellular signals regulating pancreas development and function. Genes Dev 15:
111–127.
Kim, S.K., Hebrok, M., Li, E., Oh, S.P., Schrewe, H., Harmon,
E.B., Lee, J.S. and Melton, D.A. (2000) Activin receptor patterning of foregut organogenesis. Genes Dev 14:1866–1871.
Kim, S.K., Hebrok, M. and Melton, D.A. (1997) Notochord to
endoderm signaling is required for pancreas development.
Development 124:4243–4252.
Kubo, A., Shinozaki, K., Shannon, J.M., Kouskoff, V., Kennedy,
M., Woo, S., Fehling, H.J. and Keller, G. (2004) Development
of definitive endoderm from embryonic stem cells in culture.
Development 131:1651–1662.
Kumar, M. and Melton, D. (2003) Pancreas specification: a budding question. Curr Opin Genet Dev 13:401–407.
Le Douarin, N. (2000) Des Chimeras, des Clones et des Genes.
Odile Jacob, Paris.
Leon-Quinto, T., Jones, J., Skoudy, A., Burcin, M. and Soria, B.
(2004) In vitro directed differentiation of mouse embryonic stem
cells into insulin-producing cells. Diabetologia 47:1442–1451.
Loebel, D.A., Watson, C.M., De Young, R.A. and Tam, P.P.
(2003) Lineage choice and differentiation in mouse embryos and
embryonic stem cells. Dev Biol 264:1–14.
Lumelsky, N., Blondel, O., Laeng, P., Velasco, I., Ravin, R. and
McKay, R. (2001) Differentiation of embryonic stem cells to insulin-secreting structures similar to pancreatic islets. Science
292:1389–1394.
Miralles, F., Czernichow, P. and Scharfmann, R. (1998) Follistatin
regulates the relative proportions of endocrine versus exocrine
tissue during pancreatic development. Development 125:1017–
1024.
Miyazaki, S., Yamato, E. and Miyazaki, J. (2004) Regulated expression of pdx-1 promotes in vitro differentiation of insulinproducing cells from embryonic stem cells. Diabetes 53:
1030–1037.
Moriya, N., Komazaki, S., Takahashi, S., Yokota, C. and Asashima, M. (2000) In vitro pancreas formation from Xenopus ectoderm treated with activin and retinoic acid. Dev Growth Differ
42:593–602.
Percival, A.C. and Slack, J.M. (1999) Analysis of pancreatic
development using a cell lineage label. Exp Cell Res 247:
123–132.
Ritvos, O., Tuuri, T., Eramaa, M., Sainio, K., Hilden, K., Saxen, L.
and Gilbert, S.F. (1995) Activin disrupts epithelial branching
morphogenesis in developing glandular organs of the mouse.
Mech Dev 50:229–245.
Sanvito, F., Herrera, P.L., Huarte, J., Nichols, A., Montesano, R.,
Orci, L. and Vassalli, J.D. (1994) TGF-beta 1 influences the
relative development of the exocrine and endocrine pancreas in
vitro. Development 120:3451–3462.
Selander, L. and Edlund, H. (2002) Nestin is expressed in mesenchymal and not epithelial cells of the developing mouse pancreas.
113:189–192.
Shi, Y., Hou, L., Tang, F., Jiang, W., Wang, P., Ding, M. and
Deng, H. (2005) Inducing embryonic stem cells to differentiate
into pancreatic beta cells by a novel three-step approach with
activin A and all-trans retinoic acid. Stem Cells 23:656–662.
11
Shiozaki, S., Tajima, T., Zhang, Y.Q., Furukawa, M., Nakazato,
Y. and Kojima, I. (1999) Impaired differentiation of endocrine
and exocrine cells of the pancreas in transgenic mouse expressing
the truncated type II activin receptor. Biochim Biophys Acta
1450:1–11.
Shiroi, A., Yoshikawa, M., Yokota, H., Fukui, H., Ishizaka, S.,
Tatsumi, K. and Takahashi, Y. (2002) Identification of insulinproducing cells derived from embryonic stem cells by zinc-chelating dithizone. Stem Cells 20:284–292.
Slack, J.M. (1995) Developmental biology of the pancreas. Development 121:1569–1580.
Soria, B., Roche, E., Berna, G., Leon-Quinto, T., Reig, J.A. and
Martin, F. (2000) Insulin-secreting cells derived from embryonic
stem cells normalize glycemia in streptozotocin-induced diabetic
mice. Diabetes 49:157–162.
Stafford, D., Hornbruch, A., Mueller, P.R. and Prince, V.E. (2004)
A conserved role for retinoid signaling in vertebrate pancreas
development. Dev Genes Evol 214:432–441.
Stafford, D. and Prince, V.E. (2002) Retinoic acid signaling is required for a critical early step in zebrafish pancreatic development. Curr Biol 12:1215–1220.
Yamada, T., Yoshikawa, M., Kanda, S., Kato, Y., Nakajima, Y.,
Ishizaka, S. and Tsunoda, Y. (2002) In vitro differentiation of
embryonic stem cells into hepatocyte-like cells identified by cellular uptake of indocyanine green. Stem Cells 20:146–154.
Yamaoka, T., Idehara, C., Yano, M., Matsushita, T., Yamada, T.,
Ii, S., Moritani, M., Hata, J., Sugino, H., Noji, S. and Itakura,
M. (1998) Hypoplasia of pancreatic islets in transgenic mice
expressing activin receptor mutants. J Clin Invest 102:294–301.
Yasunaga, M., Tada, S., Torikai-Nishikawa, S., Nakano, Y., Okada, M., Jakt, L.M., Nishikawa, S., Chiba, T., Era, T. and
Nishikawa, S. (2005) Induction and monitoring of definitive and
visceral endoderm differentiation of mouse ES cells. Nat
Biotechnol 23:1542–1550.