Content uploaded by Laurent Meijer
Author content
All content in this area was uploaded by Laurent Meijer on Sep 04, 2015
Content may be subject to copyright.
ARTICLES
During early mammalian embryogenesis, the first critical fate
decision, occurring between the late morula and blastocyst stages,
divides the embryo into two major lineages: the pluripotent inner
cell mass, which generates all three germ-layer tissues, and the tro-
phoblast, which supports embryonic growth1,2.Embryonic stem
cells are self-renewing cells found in the inner cell mass of mouse
blastocysts, and are proven to maintain pluripotency1,3,4.The
recent establishment of HESC lines has led to investigations of the
molecular pathways that regulate early human embryogenesis.
HESCs could also be used to generate specific types of tissues for
cell replacement therapy as a promising approach to treatment of
devastating diseases such as Parkinson disease and diabetes melli-
tus5–8.Despite the importance of ESCs in developmental biology
and their potential impact on tissue engineering, little is known
about the signaling pathways that govern the unique properties of
ESCs. Leukemia inhibitory factor (LIF)/Stat-3 signaling is the only
pathway known to be involved in self-renewal of MESCs1.
However, this pathway is dispensable for embryonic development
before gastrulation, suggesting the existence of other signaling cas-
cades essential for ESC self-renewal1,9.
Using genetic and biochemical approaches, we have begun a
molecular dissection of signaling pathways functioning in MESCs
and HESCs. Large-scale gene expression profiling of HESCs
reveals that components of several signal transduction pathways
are transcriptionally enriched in the undifferentiated state, allow-
ing a prioritization of the pathways to be studied9.As the main
components of the canonical Wnt pathway are detected in undif-
ferentiated HESCs9,we evaluated the potential involvement of the
Wnt pathway in the self-renewal of MESCs and HESCs. We
recently reported the discovery of a new GSK-3 inhibitor, BIO,
which is derived from mollusk Tyrian purple10.This pharmacolog-
ical inhibitor inactivates GSK-3 function with higher specific
activity than LiCl, thus facilitating efficient Wnt pathway activa-
tion. We have used this unique GSK-3 inhibitor to uncover a criti-
cal role of the Wnt pathway in regulating self-renewal in MESCs
and HESCs.
RESULTS
LIF Stat-3 activation does not sustain HESC self-renewal
Although the LIF/Stat-3 pathway is currently the only pathway
known to be involved in the self-renewal of MESCs1,its role has not
been clearly demonstrated in HESCs5,6.To evaluate this, we used a
feeder-free culture system in which HESCs are physically free from
mouse embryonic fibroblasts (MEFs), thereby making their differ-
entiation state dependent only on the culture medium11.We used
three independent HESC lines, H1 (ref. 5),BGN1 and BGN2, which
1Laboratory of Molecular Vertebrate Embryology, The Rockefeller University, 1230 York Avenue, New York, NY 10021, USA. 2Laboratory of Molecular and Cellular
Neuroscience, The Rockefeller University, 1230 York Avenue, New York, NY 10021, USA. 3CNRS, Cell Cycle Group, Station Biologique, 29680-Roscoff, France.
4Department of Pharmacognosy, University of Athens, Greece. Correspondence should be addressed to A.H.B. (brvnlou@rockefeller.edu).
Published online 21 December 2003; doi:10.1038/nm979
Maintenance of pluripotency in human and mouse
embryonic stem cells through activation of Wnt signaling
by a pharmacological GSK-3-specific inhibitor
Noboru Sato1,Laurent Meijer2,3,Leandros Skaltsounis4,Paul Greengard2& Ali H Brivanlou1
Human and mouse embryonic stem cells (HESCs and MESCs, respectively) self-renew indefinitely while maintaining the
ability to generate all three germ-layer derivatives. Despite the importance of ESCs in developmental biology and their
potential impact on tissue replacement therapy, the molecular mechanism underlying ESC self-renewal is poorly understood.
Here we show that activation of the canonical Wnt pathway is sufficient to maintain self-renewal of both HESCs and MESCs.
Although Stat-3 signaling is involved in MESC self-renewal, stimulation of this pathway does not support self-renewal of
HESCs. Instead we find that Wnt pathway activation by 6-bromoindirubin-3′-oxime (BIO), a specific pharmacological inhibitor
of glycogen synthase kinase-3 (GSK-3), maintains the undifferentiated phenotype in both types of ESCs and sustains
expression of the pluripotent state-specific transcription factors Oct-3/4, Rex-1 and Nanog. Wnt signaling is endogenously
activated in undifferentiated MESCs and is downregulated upon differentiation. In addition, BIO-mediated Wnt activation is
functionally reversible, as withdrawal of the compound leads to normal multidifferentiation programs in both HESCs and
MESCs. These results suggest that the use of GSK-3-specific inhibitors such as BIO may have practical applications in
regenerative medicine.
NATURE MEDICINE VOLUME 10 |NUMBER 1 |JANUARY 2004 55
© 2004 Nature Publishing Group http://www.nature.com/naturemedicine
ARTICLES
can be successfully maintained in the undifferentiated state using
medium conditioned with MEFs. Normal karyotypes were con-
firmed after several passages (data not shown). Undifferentiated
H1 and BGN1 cells showed a typical compact morphology with a
high nuclear-cytoplasmic ratio, comparable to what is seen in
HESCs grown on MEFs5(Fig. 1a). The HESCs started to flatten as
early as 24 h after replacing the conditioned medium with uncon-
ditioned medium, achieving a completely differentiated cell mor-
phology after 5 to 7 d incubation (Fig. 1b).
To confirm and quantify the differentiation process, we used
immunocytochemical analysis to monitor the expression of Oct-
3/4, a key transcription factor for pluripotency that is restricted to
the inner cell mass of blastocysts12–14.HESCs grown in condi-
tioned medium showed unambiguous nuclear Oct-3/4 staining in
each cell after 7 d in culture, whereas a marked reduction in
Oct-3/4 expression was observed in flattened differentiated cells
cultured in unconditioned medium for 7 d (Fig. 1c). This differen-
tiation program was not prevented by the addition of LIF, even at
higher concentrations (2,000–3,000 U/ml), suggesting that LIF
alone is not sufficient to maintain HESCs in an undifferentiated
state (Fig. 1b,c). Quantitative evaluation of Oct-3/4 expression by
image analysis confirmed these observations (Fig. 1d). Human-
derived LIF, or a Stat-3-activating combination of interleukin-6
and soluble interleukin-6 receptor, also failed to prevent differen-
tiation (data not shown).
To evaluate whether HESCs are capable of responding to a LIF
signal, we used western blot analysis to determine the level of Stat-
3 (Tyr705) phosphorylation that can be used as a measure of the
activation level of the Stat-3 signaling pathway15.H1 and BGN1
cells treated with LIF for 20 min showed a weak increase in
Tyr705-phosphorylated Stat-3 (Fig. 1e) that was not further
enhanced at later time points (data not shown). In contrast,strong
activation of the ERK pathway was evident upon LIF stimulation
in both HESCs and MESCs. The weak Stat-3 activation observed in
HESCs contrasts with previous reports that MESCs show a marked
increase in Tyr705-phosphorylated Stat-3 upon LIF treatment15.
These results indicate that although the Stat-3 signaling pathway
can be stimulated by LIF in HESCs, the level of activation is far less
than in MESCs, and does not effect self-renewal in HESCs.
Because other Stat transcription factors might be involved in
HESCs, we also tested the phosphorylation status of Stat-1 and
Stat-5. Neither of these molecules were activated in HESCs or
MESCs upon LIF treatment (D. Besser, N.S. and A.H.B., unpub-
lished observations), eliminating Stat signaling as being causal for
self-renewal of HESCs.
We therefore investigated the contribution of other pathways to
ESC self-renewal. We conducted global expression screens using
microarrays for HESCs9,16,and tested the biochemical state of the
components of the major signaling pathways17.The main signal
transducers of the canonical Wnt pathway were detected in undif-
ferentiated HESCs in our microarray experiments (Supple-
mentary Table 1 online)9.This result prompted us to evaluate the
Wnt pathway.
MESCs and HESCs can transduce Wnt signaling
The canonical Wnt pathway is initiated upon Wnt protein binding
to the Frizzled receptor at the cell surface.Signaling downstream of
the receptor leads to inactivation of GSK-3, resulting in the nuclear
accumulation of β-catenin, which activates the transcription of
Wnt target genes in collaboration with T-cell-specific factors
56 VOLUME 10 |NUMBER 1 |JANUARY 2004 NATURE MEDICINE
CM
H1BGN1
Non-CM
H1BGN1
LIF
Non-CM
LIF CM
CM Non-CM LIF
6,000
5,000
4,000
3,000
2,000
1,000
0
Relative intensity
H1 BGN1 CJ7
LIF
Phospho-Stat-3
Stat-3
Phospho-ERK-1/2
ERK-1/2
ab c
d
e
Figure 1 LIF-induced Stat-3 activation is not sufficient to maintain the undifferentiated state of HESCs. (a) Phase contrast images of H1 and BGN1 cells
grown in a feeder-free system with conditioned medium (CM). Scale bars, 300 µm (left) or 100 µm (right). (b) HESCs cultured in nonconditioned medium
(non-CM) or LIF-containing nonconditioned medium show flattened, differentiated morphology. Insets show high-power fields. Scale bars, 300 µm or
100 µm (insets). (c) Immunofluorescent (left) and phase-contrast (right) images of BGN1 cells grown under indicated conditions and stained with
Oct-3/4-specific antibody. Scale bars, 100 µm. (d) Intensity of Oct-3/4 expression under each condition, as quantified by an image-analyzing system.
(e) Western blot analysis using the indicated antibodies of H1 and BGN1 HESCs and CJ7 MESCs treated with or without LIF.
© 2004 Nature Publishing Group http://www.nature.com/naturemedicine
ARTICLES
(TCFs)18,19.Wnt signaling can also be activated by direct, intracel-
lular inhibition of GSK-3 function using specific inhibitors20.
We recently discovered that 6-bromoindirubins, initially derived
from Tyrian purple, are selective and potent inhibitors of GSK-3 (ref.
10). The indirubins BIO and its kinase-inactive analog, 1-methyl-6-
bromoindirubin-3′-oxime (MeBIO; Fig. 2a), are particularly con-
venient tools for modulating GSK-3 activity10.We first assessed the
ability of BIO to activate the Wnt pathway in 293 human kidney
epithelial cells. We used a luciferase reporter system in which the
promoter module contained TCF binding sites (TopFlash) or nonre-
sponsive, mutated binding sites (FopFlash)21.BIO, but not MeBIO,
robustly upregulated reporter activity at micromolar concentrations
far below those required for LiCl-mediated activation, indicating
efficient activation of the canonical Wnt pathway by BIO (data not
shown). Based on this evidence, we used BIO as a positive regulator
of Wnt signaling in subsequent experiments.
We first examined whether mouse and human ESCs are capable
of transducing Wnt signaling when treated with BIO. MESCs (CJ7
cells) treated with BIO showed a substantial, dose-dependent
increase in TopFlash reporter activity, whereas MeBIO-treated
cells did not show any change in activity (Fig. 2b). As expected, no
substantial change in FopFlash reporter
activity was observed under similar condi-
tions (Fig. 2b). Similar results were
obtained using the E14 MESC line derived
from mice with a 129 background (data not
shown). We then evaluated expression of
β-catenin at the cellular level, as an indica-
tor of the activation status of the canonical
Wnt pathway in HESCs. Like HESCs grown
in conditioned medium, BIO-treated cells
showed nuclear accumulation of β-catenin,
whereas cells treated with nonconditioned
medium did not (Fig. 2c). MeBIO-treated
cells also did not show nuclear accumula-
tion of β-catenin (data not shown). These
results indicate that BIO can activate Wnt
signaling in HESCs. This result is further
supported by the observation that cyclin
D1, a Wnt target gene22,was upregulated in
HESCs treated with BIO (Supplementary
Fig. 1). Because MESCs and HESCs remain
undifferentiated in the presence of MEFs,
and because MEFs express multiple Wnt
ligands (Supplementary Fig. 2 online), these results raise the pos-
sibility that Wnt proteins secreted from MEFs might activate Wnt
signaling in undifferentiated HESCs and MESCs.
Active Wnt signaling in undifferentiated MESCs
To monitor Wnt signaling in MESCs over a longer period of time, we
generated a reporter MESC line (CJ-TY) in which a modified version
of yellow fluorescent protein (YFP)23 was regulated by the TopFlash
promoter module. Similar to the luciferase reporter data (Fig. 2b),
there was no appreciable difference in YFP expression between LIF-
treated and untreated cells on day 2 (data not shown). After 5 d of
incubation, however, the LIF-treated reporter cells maintained a
strong level of promoter activity and had an undifferentiated cell
morphology, whereas untreated cells had a differentiated cell mor-
phology with a notable decrease in YFP expression (Fig. 2d,e). These
data suggest that Wnt signaling is endogenously active in undifferen-
tiated MESCs and is downregulated upon their differentiation.
Wnt signaling induces Rex-1 expression in MESCs
On the basis of these observations, we reasoned that active Wnt
signaling might be instrumental in maintaining the molecular
NATURE MEDICINE VOLUME 10 |NUMBER 1 |JANUARY 2004 57
BIO: R = H
MeBIO: R = CH3
250,000
200,000
150,000
100,000
50,000
0
LIF(+) MeBIO MeBIO MeBIO BIO BIO BIO
1 µM 2 µM 5 µM 1 µM 2 µM 5 µM
LIF(+) MeBIO MeBIO MeBIO BIO BIO BIO
1 µM 2 µM 5 µM 1 µM 2 µM 5 µM
LIF (+) LIF ( )
_
40,000
35,000
30,000
25,000
20,000
15,000
10,000
5,000
0
LIF (+) LIF ( )
_
CM
Non-CM
BIO (2 µM)
BIO (5 µM)
Relative intensity
Relative luciferase activity
ab
c
d
e
Figure 2 MESCs and HESCs can transduce Wnt
signaling when treated with BIO. (a) Chemical
structure of BIO and MeBIO. (b) Luciferase
reporter assay of CJ7 cells transfected with
reporter constructs (TopFlash or FopFlash)
and treated with BIO or MeBIO. (c) Confocal
images of H1 cells grown under the indicated
conditions and incubated with β-catenin-specific
antibody. CM, conditioned medium; non-CM,
nonconditioned medium. HESCs cultured with
BIO showed nuclear accumulation of β-catenin.
Cells were counterstained with DAPI (bottom
right). Scale bars, 20 µm (left) or 10 µm (right).
(d) Immunofluorescent images of CJ-TY cells
grown with or without LIF. Insets show phase-
contrast images. Scale bars, all 100 µm.
(e) Quantitation of YFP expression by an image-
analyzing system.
© 2004 Nature Publishing Group http://www.nature.com/naturemedicine
ARTICLES
machinery responsible for the undifferentiated state. The loss of
Wnt signaling might trigger deactivation of this machinery, allow-
ing initiation of the differentiation program. To test this hypothe-
sis, we monitored the expression of Rex-1, another molecular
marker of pluripotency, in MESCs using a luciferase reporter con-
struct in which the luciferase gene was regulated by the Rex-1 min-
imal enhancer element24.Compared with LIF-treated CJ7 cells,
Rex-1 promoter activity was substantially upregulated in BIO-
treated cells, and was slightly reduced in cells treated with MeBIO
or grown in the absence of LIF (Fig. 3a). Similar results were
obtained with the E14 cell line, whereas P19 embryonal carcinoma
cells24 and nonpluripotent stem cell lines (including 293, NIH3T3
and mesenchymal stem cells) did not show notable Rex-1 tran-
scriptional activity (data not shown). When cells were transfected
with a dominant-negative TCF-3 construct that specifically blocks
expression of the canonical Wnt target genes25,26,BIO-mediated
transcriptional activation was largely abolished, confirming that
BIO functions through the canonical Wnt pathway. Given that
GSK-3 regulates multiple pathways,including those mediating sig-
naling by insulin and growth factors, we cannot not rule out the
possibility that other BIO-activated signaling pathways might
affect the pluripotency of MESCs. To further substantiate the role
of Wnt signaling in Rex-1 transcriptional regulation, we cotrans-
fected constitutively active forms of β-catenin and TCF-4, which
assembles a pivotal transcription factor in the canonical Wnt
pathway, into MESCs, and observed efficient upregulation of Rex-
1 promoter activity (Fig. 3a).
Although Rex-1 reporter activity in non-LIF-treated MESCs did
not decline markedly at 48 h (Fig. 3a), this time period may have
been too short to allow cells to differentiate completely, as there
were no obvious morphological changes. To monitor transcription
from the Rex-1 reporter over a longer period of time at the cellular
level, we generated a stable MESC reporter line (CJRex-Y) express-
ing a mutant form of YFP regulated by the Rex-1 minimal
enhancer24.BIO- and LIF-treated CJRex-Y cells showed strong YFP
signals. BIO-treated cells formed colonies that were somewhat
more compact than those observed in LIF-treated cells after 5 d of
incubation (Fig. 3b). Non-LIF-treated and MeBIO-treated cells
showed substantially lower activity and a flattened cell shape. These
data showed that activation of the canonical Wnt pathway by BIO
is sufficient to maintain an undifferentiated cell morphology and
Rex-1 promoter activity in the absence of LIF. To further examine
the role of Wnt signaling in the undifferentiated state, we used
Wnt3a-conditioned medium instead of BIO to stimulate the Wnt
pathway (Supplementary Methods online). CJRex-Y cells grown in
medium conditioned from L cells that stably express Wnt3a main-
tained a high level of Rex-1 transcriptional activity comparable to
that of LIF-treated cells. Cells grown in medium conditioned from
wild-type L cells or non-LIF-treated cells showed reduced reporter
activity, as determined quantitatively by fluorescence-activated cell
sorting analysis (Supplementary Fig. 3 online).
Pluripotency-specific gene expression maintained by Wnt
We next asked whether the effect of Wnt pathway activation
observed in MESCs is conserved in HESCs. We examined whether
the undifferentiated state of HESCs in the feeder-free system could
be maintained by exogenous activation of the Wnt pathway using
GSK-3 inhibitors. MeBIO-treated H1 and BGN1 cells showed fully
flattened morphology after 7 d of incubation (Fig. 4a), similar to
cells treated with nonconditioned medium (Fig. 1b). Another
GSK-3 inhibitor, LiCl, did not maintain the undifferentiated phe-
notype at 5 mM and showed substantial toxicity at 10 mM
(Fig. 4a). In contrast, HESCs treated with 2 µM BIO largely
retained an undifferentiated morphology (Fig. 4a) comparable to
that of conditioned medium–treated cells (Fig. 1a). This observa-
tion was further supported by monitoring Oct-3/4 expression at
the cellular level. Notably, sustained Oct-3/4 expression was
observed in the majority of BIO-treated HESCs, in contrast to a
substantial reduction of Oct-3/4 expression in MeBIO-treated,
differentiated cells (Fig. 4a). Quantitative imaging analysis showed
a comparable level of Oct-3/4 expression in cells treated with BIO
or conditioned medium (data not shown). We also used recombi-
nant Wnt3a protein to confirm that the BIO-mediated effect was
caused by Wnt activation. Wnt3a-treated cells maintained com-
pact, undifferentiated colonies with high Oct-3/4 expression
(Fig. 4b), whereas cells cultured in nonconditioned medium with
PBS (used for reconstitution of Wnt3a protein) showed differenti-
58 VOLUME 10 |NUMBER 1 |JANUARY 2004 NATURE MEDICINE
70,000
60,000
50,000
40,000
30,000
20,000
10,000
0
Relative luciferase activity
LIF
dnTCF-3
MeBIO (5 µM)
β-catenin
TCF-4
BIO (5 µM)
a
b
Figure 3 Activation of Wnt signaling by BIO maintains the
undifferentiated state of MESCs. (a) Luciferase reporter assay of CJ7
cells transfected with Rex-1 reporter plasmid and effector constructs,
and treated with test compounds. dnTCF-3, dominant-negative TCF-3.
b) Immunofluorescent images of CJRex-Y cells cultured under the
indicated conditions. Cells incubated with BIO showed robust YFP
expression and, to some extent, more compact colonies (inset) than
LIF-treated cells. Scale bars, all 100 µm.
© 2004 Nature Publishing Group http://www.nature.com/naturemedicine
ARTICLES
ated morphology (Fig. 4b) with low Oct-3/4 expression (data not
shown). To determine whether Oct-3/4 expression is regulated at
the transcriptional level, northern blot analysis was done (Fig. 4c).
A substantial amount of Oct-3/4 transcript was expressed in both
BIO- and conditioned medium–treated cells. In contrast, there
was much less Oct-3/4 transcript in cells under other conditions
tested, indicating that Oct-3/4 expression is maintained through
activation of Wnt signaling. We also used the same Rex-1 reporter
assay system used in MESCs to test HESC lines. Because the trans-
fection efficiency of H1 cells was extremely low, we evaluated
BGN1 and BGN2 cells, which had higher transfection efficiencies.
Both lines showed a pattern of Rex-1 reporter activity identical to
that observed in MESCs (Fig. 4d), suggesting transcriptional regu-
lation of Rex-1 expression by Wnt signaling in HESCs.
Nanog is a recently discovered homeodomain transcription fac-
tor that is both sufficient and necessary for maintenance of
pluripotency in MESCs and mouse epiblasts, independent of Stat-
3 signaling27,28.Our northern blot analysis revealed that a sub-
stantial amount of Nanog transcript was expressed in BIO-treated
HESCs, whereas a marked reduction in its expression level was
observed under differentiation conditions (Fig. 4c). Taken
together, these results indicate that activation of the canonical
Wnt pathway by BIO facilitates maintenance of the undifferenti-
ated phenotype and the sustained expression of pluripotent state-
specific transcription factors in MESCs and HESCs, implying a
conserved role for Wnt signaling in human and mouse ESCs.
Wnt signaling preserves HESC multidifferentiation programs
Because a unique property of ESCs is their ability to generate cells
with functional diversity,we examined the differentiation potential
of BIO-treated ESCs using established ESC differentiation systems.
We first generated embryoid bodies consisting of three germ-layer
derivatives from undifferentiated HESCs1,29.We observed em-
bryoid body formation from BIO-treated H1 cells at a level compa-
rable to that from conditioned medium–treated cells, whereas no
embryoid bodies were formed under other conditions (Fig. 5a).
Similar results were obtained with BGN1 and BGN2 cells using the
same system (data not shown). Lineage-specific marker analysis by
NATURE MEDICINE VOLUME 10 |NUMBER 1 |JANUARY 2004 59
MeBIO (2 µM)
BIO (2 µM)
BIO (2 µM)
LiCl
Oct-3/4
Nanog
28S rRNA
CM
Non-CM
Relative luciferase activity
14,000
12,000
10,000
8,000
6,000
4,000
2,000
0
CM
dnXTCF3
dnhTCF4
MeBIO (2 µM)
BIO (2 µM)
β-catenin
hTCF4
MeBIO (2 µM)
BIO (2 µM)
ab
cd
Figure 4 Activation of Wnt signaling by BIO maintains the undifferentiated state of HESCs. (a) Immunocytochemical analysis of Oct-3/4 expression in
H1 and BGN1 cells cultured in nonconditioned medium under the indicated conditions. Most cells treated with BIO had strong Oct-3/4 expression and
compact, undifferentiated morphology. First and third rows show fluorescent images; insets show phase-contrast images. Scale bars, all 100 µm. (b) BGN2
cells cultured in nonconditioned medium with Wnt3a, or in conditioned medium (CM) for 5 d (phase-contrast images, except bottom right). Oct-3/4
expression in Wnt3a-treated cells was detected by immunocytochemistry (bottom right). Scale bars, 50 µm (bottom right) or 100 µm (all others).
(c) Northern blot analysis of H1 cells cultured under indicated conditions for 7 d. 28S rRNA was used as loading control. (d) Rex-1 reporter assay of
BGN2 cells treated with test compounds. Similar results were obtained using BGN1 cells (data not shown). dn XTCF3, dominant-negative Xenopus TCF-3;
dnhTCF4, dominant-negative human TCF-4 (hTCF4).
© 2004 Nature Publishing Group http://www.nature.com/naturemedicine
ARTICLES
RT-PCR showed that embryoid bodies derived from conditioned
medium– or BIO-treated cells similarly developed into ectoderm
(NeuroD and keratin), mesoderm (T gene) and endoderm (α-feto-
protein and GATA-4) derivatives (Fig. 5b). To further determine
the differentiation phenotype of BIO-treated HESCs at the cellular
level, embryoid bodies were grown under adherent conditions and
evaluated by immunocytochemistry. HESCs initially treated with
BIO showed a wide variety of morphologies and expression of line-
age-specific molecules, including markers of ectoderm (cytokeratin
and glial fibrillary acidic protein (GFAP)), mesoderm (smooth-
muscle actin), endoderm (α-fetoprotein) and trophectoderm
(Tromo-1; Fig. 5c), at levels comparable to those seen in condi-
tioned medium–treated cells (data not shown).
A growing body of evidence indicates that ESCs can be manipu-
lated to undergo lineage-restricted differentiation programs, includ-
ing differentiation into neurons by specific culture techniques. They
can subsequently be grafted into host tissues, suggesting possible
applications for tissue engineering1,7,8.We therefore tested whether
BIO-treated HESCs retain the capacity to differentiate exclusively
into neurons in a stromal coculture system in which stromal-derived
factors promote a high level of neurogenesis9,30.We found that BIO-
treated H1 cells underwent robust neurogenesis on stromal feeders,
comparable to that seen in conditioned medium–treated cells
(Fig. 5d), whereas much lower efficiencies were observed in cells
grown under other conditions (Fig. 5e). Similar results were
obtained with BGN1 and BGN2 cells (data not shown).
Wnt signaling maintains MESCs in the pluripotent state
We next addressed whether BIO-treated MESCs maintain the abil-
ity to form three germ-layer derivatives, as evaluated by subcuta-
neous injection of MESCs into syngeneic mice. BIO-treated
MESCs generated teratomas consisting of all three germ
layer–derived tissues, including neuroepithelium (ectoderm), car-
tilage (mesoderm) and ciliated epithelium (endoderm; Fig. 6a).
Finally, given that a unique functional property of ESCs is their
capacity to synchronize with the surrounding embryonic microen-
vironment when transferred into the blastocyst, we used chimeric
mice to investigate whether BIO-treated MESCs retain the poten-
tial to be integrated into early embryonic tissues. We found that
BIO-treated CJ-GFP cells that constitutively express GFP were
incorporated into several embryonic tissues at the mid-gestation
stage, as determined by immunohistochemistry (Fig. 6b). More
than 60% of the injected embryos contained colonized GFP-
positive cells in two experiments (11 of 14 and 8 of 12 embryos).
An initial assessment of coat-color chimerism of live offspring
found that two of five newborn animals were chimeric.
60 VOLUME 10 |NUMBER 1 |JANUARY 2004 NATURE MEDICINE
12
10
8
6
4
2
0CM Non-CM BIO
(2 µM)
MeBIO
(2 µM)
NeuroD
Keratin
T gene
α-FP
GATA-4
GAPDH
RT
(GAPDH)
_
Number of Tuj-1
20
18
16
14
12
10
8
6
4
2
0CM Non-CM MeBIO
(2 µM) BIO
(2 µM)
+
Number of EBs
CM BIO (2 µM)
abc
d
e
wells
Figure 5 Wnt activation of HESCs by BIO preserves normal multidifferentiation potential. (a) H1 cells cultured in conditioned medium (CM), nonconditioned
medium (non-CM), MeBIO or BIO were induced to form embryoid bodies (EBs). Numbers of embryoid bodies per condition were counted in triplicate.
(b) RT-PCR analysis of embryoid bodies derived from conditioned medium– or BIO-treated cells. α-FP, α-fetoprotein; –RT, no reverse transcriptase.
(c) Immunocytochemical analysis of further differentiated embryoid bodies . Top left, phase-contrast image of embryoid bodies derived from BIO-treated
cells and grown for 7 d. Cells were stained with antibodies to the indicated protein. Cells stained with antibodies to GFAP or α-fetoprotein were
counterstained with DAPI. Scale bars, 50 µm (top right and bottom left) or 100 µm (all others). (d) H1 cells were grown under different conditions and
differentiated on PA6 feeder cells. Robust generation of neurons (Tuj-1+cells) was observed in BIO-treated cells. Scale bars, 300 µm (top) or 100 µm
(bottom). (e) Number of wells containing Tuj-1+cells were counted for each condition, in three experiments.
© 2004 Nature Publishing Group http://www.nature.com/naturemedicine
ARTICLES
Together, these results indicate that although activation of Wnt
signaling by BIO allows ESCs to remain undifferentiated, normal
differentiation programs can be properly reactivated by withdraw-
ing the compound, highlighting the preservation of the essential
features of ESCs.
DISCUSSION
We show here that despite the ability of LIF/Stat-3 signaling to
support self-renewal of MESCs, it does not prevent the differentia-
tion of three independent HESC lines, suggesting that this path-
way is not essential for self-renewal in HESCs. The LIF/Stat-3
pathway, however, is dispensable for pregastrulation mouse
embryos1.In addition, an as-yet-unidentified soluble factor
secreted from a differentiated cell line supports germline trans-
mission of MESCs independently of Stat-3 signaling31.We pro-
pose that Wnt signaling is a possible common pathway that
maintains the undifferentiated state of human and mouse ESCs.
Oct-3/4 and Rex-1 have been studied as representative tran-
scription factors involved in controlling the pluripotent state of
MESCs, although little is known about the upstream signals that
regulate these molecules32.Our results provide new insight into
the regulatory cascades underlying the unique molecular program
in ESCs by showing that Wnt signaling can positively regulate
expression of these key molecules in human and mouse cells. This
finding was further expanded by the observation that expression
of Nanog, a homeoprotein both sufficient and necessary for the
maintenance of pluripotency27,28,is also sustained by Wnt activa-
tion . Because preservation of Oct-3/4 expression is not sufficient
to prevent differentiation33,self-renewal of ESCs by Wnt activa-
tion might be mediated by transcriptional regulation of Nanog.
Further studies are required to identify the molecular interactions
between these transcription factors and Wnt signaling compo-
nents in ESCs.
Loss-of-function mutations in Wnt
pathway components have been generated
by gene targeting in mice. β-catenin
mutant mice are defective in anteroposte-
rior axis formation, but not in maintaining
the pluripotent state34.However, because
plakoglobin, another member of the
armadillo family, redistributes to compen-
sate for the adhesion function of β-catenin,
it might also transduce Wnt signaling in
the mutant embryos35.Given that pluripo-
tency is a fundamental biological function
in multicellular organisms, it is likely to be
evolutionarily secured by multiple genetic
backup systems, as suggested by the expres-
sion of several Wnt ligands in preimplanta-
tion embryos36.Combined gene targeting
approaches will be required to uncover the
phenotypes that are defective in the
pluripotent program.
Several other lines of evidence have indi-
cated possible links between Wnt signaling
and multipotent self-renewal of stem
cells37–41.Purified, biologically active Wnt
ligands have recently been shown to sustain
self-renewal of mouse hematopoietic stem
cells, possibly through upregulation of
HoxB4 and Notch-1, which are involved in
hematopoietic stem cell proliferation42.Together with our results,
these observations raise the possibility that Wnt signaling main-
tains the undifferentiated state in each class of stem cells by regu-
lating individual stem cell–specific self-renewal machineries.
Elucidation of how Wnt signaling governs the regulation of spe-
cific programs that predetermine different levels of ‘stemness’ will
be essential for further understanding of stem cell biology.
During early vertebrate embryogenesis, the canonical Wnt path-
way has early and important embryological functions, including
its role in the induction of the dorsal organizer (node)19 through
the formation of the Nieuwkoop center19.It is,therefore, tempting
to speculate that in addition to mediating the pathway underlying
stemness, the sustained activation of this pathway might lead to
the formation of the embryonic node (or organizer) in vitro.
Aberrant activation of Wnt signaling has been implicated in
cancer formation in numerous basic and clinical studies19,43.A
recent study used mutant MESC lines in which Wnt signaling was
constitutively activated by mutations in genes encoding the adeno-
matous polyposis coli protein or β-catenin. The cells showed sus-
tained undifferentiated morphology and impaired differentiation
capacities, reminiscent of the uncontrolled growth of immature
cells in tumors44.Although this report is basically consistent with
our results, the simple morphological evaluation of the undiffer-
entiated phenotype, without monitoring pluripotent-
specific molecular markers, precludes a precise characterization of
the state of stemness in MESCs. Our system, in which Wnt signal-
ing is transiently activated in ESCs by a GSK-3 inhibitor, indicates
clearly that the undifferentiated state is readily reversible upon
withdrawal of the inhibitor, as illustrated by a series of functional
differentiation assays using MESCs and HESCs. Another recent
report has shown that inhibition of Wnt signaling in MESCs
accelerates neural differentiation; conversely, activation of Wnt
signaling by Wnt-1 overexpression or LiCl treatment results in
NATURE MEDICINE VOLUME 10 |NUMBER 1 |JANUARY 2004 61
ab
Figure 6 MESCs maintain pluripotency through BIO-mediated Wnt activation. (a) H&E staining of
teratomas from MESCs grown in medium containing BIO 1 µM and subcutaneously injected into
syngeneic mice. All three germ layer–derived tissues, including neuroepithelium (ectoderm; top left),
cartilage (mesoderm; top right), ciliated epithelium (endoderm; bottom left) and mucus-producing
epithelium (endoderm; bottom right) were observed. Scale bars, 100 µm (top left and top right),
10 µm (bottom left) or 20 µm (bottom right). (b) CJ-GFP cells were grown in medium containing
1 µM BIO, microinjected into mouse blastocysts and subjected to immunohistochemistry at E10.5.
Representative fluorescent images of injected embryos show colonization of GFP-positive cells in
several tissues (head, left side of top panel; dorsal trunk, left side of bottom panel).
© 2004 Nature Publishing Group http://www.nature.com/naturemedicine
ARTICLES
inhibition of neural differentiation45.Although these studies are
in line with our observations, they used culture conditions that
specifically induced neural lineages and evaluated cell progeny
using mainly neuron-specific markers, so the influence of Wnt sig-
naling on other germ-layer derivatives was not addressed.A recent
study using another pharmacological GSK-3 inhibitor showed that
P19 embryonal carcinoma cells and MESCs can be differentiated
into neurons, probably through activation of Wnt signaling46.This
result contrasts with our observations and previous reports44,45.
Although the reason for this discrepancy is unknown, it may
involve different culture conditions used or differences in unchar-
acterized functions of the compounds used.
The finding that HESCs can be reversibly maintained in an
undifferentiated state by the addition of a synthetic pharmacolog-
ical compound might open new avenues in the practical applica-
tions of HESCs in regenerative medicine. Large-scale cultivation
of a homogenous population of undifferentiated HESCs would be
an inevitable and fundamental first step in providing an unlimited
source of cells for tissue transplants. A well-defined, stable com-
pound might therefore be useful for standardizing the quality of
HESCs, rather than the undefined, feeder cell–derived factor(s)
used in the current culture protocol. In addition, indirubins such
as BIO might eliminate the requirement for all mouse-derived
materials from culture conditions, including the derivation
process. These compounds may also be tested for effects on the
expansion of many other types of adult stem or progenitor cell
populations, as the growth of these cells seems to be highly
dependent on Wnt signaling42,47.
Finally, we showed that Wnt signaling is a common pathway for
maintenance of the undifferentiated state in both mouse and
human ESCs, whereas LIF signaling is mainly involved in mouse
cells, providing an example of genetic pathways that are function-
ally different between the two species. The recent derivation and
availability of HESCs allows us to begin to access the molecular
mechanisms that regulate early human embryogenesis5,6.HESCs
represent a powerful biological tool to explore not only human-
specific genetic programs, but also the essential mechanisms
underlying embryogenesis.
METHODS
Chemicals. BIO and its kinase-inactive analog MeBIO were prepared as
described10.LiCl was purchased from Sigma.
Cell culture. HESC lines were provided by WiCell Research Institute (H1
line)5and BresaGen (BGN1 and BGN2 lines). H1 cells were cultivated on
irradiated MEFs in DMEM/F12 medium with 20% knockout serum
replacement (KSR), 1 mM L-glutamine, 1% nonessential amino acids,
0.1 mM β-mercaptoethanol and 4 ng/ml basic fibroblast growth factor (all
from Invitrogen). BGN1 and BGN2 cells were originally cultured in essen-
tially the same medium as for H1, but with 15% FBS (HyClone) and an ini-
tial concentration of 5% KSR instead of 20% KSR; the cells were gradually
adapted to a higher concentration of KSR to completely shift to the same
fully defined medium used for H1 cells. HESCs were subsequently cultured
on Matrigel (BD Biosciences) in medium conditioned by MEFs, and pas-
saged several times until colonies were free of contaminating MEFs9.
Normal karyotype was confirmed by the standard method (data not
shown). For in vitro experiments, HESCs were cultured for 3–7 d in condi-
tioned or nonconditioned media, in the presence or absence of mouse or
human LIF (Chemicon International) or recombinant mouse Wnt3a
(100 ng/ml; R&D Systems), which were added to fresh medium every day,
or GSK-3 inhibitors indicated in the Results.
MESCs, including the CJ7 (provided by W. Mark, Memorial Sloan-
Kettering Institute) or E14 (provided by C. Yang, The Rockefeller
University) cell lines, were maintained on MEFs in mouse ESC medium
containing knockout DMEM supplemented with 15% FBS, 100 mM MEM
nonessential amino acids, 0.55 mM 2-mercaptoethanol and 1 mM L-gluta-
mine (all from Invitrogen). To remove MEFs, cells were collected by
trypsinization and plated on 10-cm dishes for 30 min. Nonadherent cells
consisting mainly of ESCs were replated on gelatin- or Matrigel-coated
dishes (1,000 cells/cm2) and grown in mouse ESC medium supplemented
with l,400 U/ml LIF.
Blastocyst injection. We cultured CJ-GFP cells at a low density (500 cells/cm2)
on gelatin-coated 10 cm dishes in medium containing 1 µM BIO for 5 d. BIO-
treated cells (∼10–15 cells per blastocyst) were microinjected into each blasto-
cyst and transferred into surrogate C57Bl/6 mice. Mid-gestation embryos were
recovered and subjected to tissue sectioning. Chimerism of live offspring was
determined by evaluation of mixed coat color.
Ter a to ma formation. We grew CJ-GFP cells at a low density (500 cells/cm2)
on gelatin-coated, 10-cm dishes in medium containing 1 µM BIO for 7 d.
Cells were then passaged at the same density and cultured under the same
conditions for another 5 d, to further enforce differentiation. Despite this
extensive LIF-free differentiation culture protocol, most BIO-treated cells
formed round, tight colonies, similar to cells grown in medium containing
LIF (Fig. 3b). In contrast, cells grown in medium alone or MeBIO had
large, flat, differentiated morphology (data not shown). BIO-treated cells
(∼5 ×106cells/mouse) were subcutaneously injected into the left flank of
syngeneic 129 mice. After 3–5 weeks, the developed teratoma (∼20 mm in
diameter) was excised, fixed in 4% paraformaldehyde and subjected to his-
tological examination with H&E staining. All animal studies were
approved by the Animal Care and Use Committee of The Rockefeller
University.
Immunofluorescence. Cells were fixed in 4% paraformaldehyde for 20 min
at room temperature and incubated overnight at 4 °C with primary anti-
bodies to Oct-4 (BD Biosciences), β-catenin (BD Biosciences), Tuj-1
(BAbCO), cytokeratin (Sigma), GFAP (Dako), smooth muscle actin
(Research Diagnostics), α-fetoprotein (Cell Sciences) and Troma-1
(Developmental Studies Hybridoma Bank). For nuclear localization analy-
sis, the fixed samples were subjected to fluorescent digital confocal imaging
analysis using a Zeiss LSM 510 confocal microscope (Carl Zeiss). For
mouse embryonic tissue sections, mid-gestation embryos were fixed in 4%
paraformaldehyde, followed by paraffin embedding and tissue sectioning.
After deparaffinization, tissue sections were incubated with a GFP-specific
antibody (Molecular Probes) at 4 °C overnight. Antigens were localized
using Cy3-conjugated goat antibody to mouse IgG or Cy2-conjugated goat
antibody to rabbit IgG (Zymed Laboratories). For quantitative image
analysis, fluorescent images from triplicate samples were taken using the
Discovery1 system (Universal Imaging Corporation). The fluorescent
objects were selected by the threshold function and evaluated in five
regions per well by quantification of pixel intensity and object size
(MetaMorph software, Universal Imaging Corporation).
Northern analysis. Total RNA was isolated from cells by using Qiashredder
and RNeasy Mini Kit (Qiagen). The extracted RNA sample was quantified
by ultraviolet spectrophotometry, and qualified using the RNA Nano Lab
chip (Agilent Technologies). Total RNA (10 µg) was electrophoresed on 1%
agarose/formaldehyde gel and transferred onto a nylon membrane
(Stratagene). Probes specific for human Oct-4 and human Nanog were
prepared by RT-PCR using human gene–specific primer pairs (shown
below) and template cDNA generated from undifferentiated H1 cells.
Probes were radiolabeled with [32P]dCTP using the Prime-It probe label-
ing kit (Stratagene). The membrane was hybridized with the labeled probe
using Perfecthyb Plus (Sigma) and subjected to detection by
PhosphorImager (Amersham Biosciences). The primers had the following
sequences: Oct-3/4 forward, 5′-CGACCATCTGCCGCTTTGAG-3′;Oct-
3/4 reverse, 5′-CCCCCTGTCCCCCATTCCTA-3′;Nanog forward, 5′-TGC-
CTCACACGGAGACTGTC-3′;Nanog reverse, 5′-TGCTATTCTTCGGCCA
GTTG-3′.
62 VOLUME 10 |NUMBER 1 |JANUARY 2004 NATURE MEDICINE
© 2004 Nature Publishing Group http://www.nature.com/naturemedicine
ARTICLES
Western analysis. Total protein was extracted with lysis buffer (50 mM
Tris, 150 mM NaCl, 0.1% Triton X-100, 0.1 mM dithiothreitol and pro-
teinase inhibitors). Protein concentrations were determined by BCA
Protein Assay kit (Pierce). Protein (50 µg) was separated by 10% SDS-
PAGE and transferred onto a nylon membrane (BioRad). The membrane
was incubated with antibodies to Stat-3, phosphorylated Stat-3 (Tyr705),
ERK-1/2, phosphorylated ERK-1/2 (Thr202 and Thr204; Cell Signaling
Technology), followed by incubation with peroxidase-conjugated goat
antibody to mouse IgG or goat antibody to rabbit IgG (Jackson
ImmunoResearch), and developed with ECL reagent (Amersham
Biosciences).
Note: Supplementary information is available on the Nature Medicine website.
ACKNOWLEDGMENTS
We thank WiCell Research Institute and Bresagen for providing HESC lines;
H. Clevers, L. Gudas, A. Miyawaki and J. Miyazaki for providing plasmid
constructs; M. Willey and C.Yang for providing MESC lines; M. Heke and M.
Uchida for technical assistance; K. La Perle for histological report of teratoma
sections; A. North for confocal microscopic imaging; the Transgenic Core
Facility at The Rockefeller University and Memorial Sloan-Kettering Cancer
Institute for blastocyst injection; A.Vonica for providing a construct and
helpful discussion; and D. Besser and T. Tomoda for helpful advice. A.H.B. is
funded by The Rockefeller University. L.M. is supported by the Ministère de la
Recherche/INSERM/CNRS ‘Molécules et Cibles Thérapeutiques’Programme;
his sabbatical leave in the laboratory of P.G. is supported by The Rockefeller
University and the CNRS.
COMPETING INTERESTS STATEMENT
The authors declare that they have no competing financial interests.
Received 6 August; accepted 5 December 2003
Published online at http://www.nature.com/naturemedicine/
1. Smith, A.G. Embryo-derived stem cells: of mice and men. Annu. Rev. Cell Dev. Biol.
17, 435–462 (2001).
2. Rossant, J. Stem cells from the mammalian blastocyst. Stem Cells 19, 477–482
(2001).
3. Martin, G.R. Isolation of a pluripotent cell line from early mouse embryos cultured in
medium conditioned by teratocarcinoma stem cells. Proc. Natl. Acad. Sci. USA 78,
7634–7638 (1981).
4. Evans, M.J. & Kaufman, M.H. Establishment in culture of pluripotential cells from
mouse embryos. Nature 292, 154–156 (1981).
5. Thomson, J.A. et al. Embryonic stem cell lines derived from human blastocysts.
Science 282, 1145–1147 (1998).
6. Reubinoff, B.E., Pera, M.F., Fong, C.Y., Trounson, A. & Bongso, A. Embryonic stem
cell lines from human blastocysts: somatic differentiation in vitro. Nat. Biotechnol.
18, 399–404 (2000).
7. Hori, Y. et al. Growth inhibitors promote differentiation of insulin-producing tissue
from embryonic stem cells. Proc. Natl. Acad. Sci. USA 99, 16105–16110 (2002).
8. Kim, J.H. et al. Dopamine neurons derived from embryonic stem cells function in an
animal model of Parkinson’s disease. Nature 418, 50–56 (2002).
9. Sato, N. et al. Molecular signature of human embryonic stem cells and its compari-
son with the mouse. Dev. Biol. 260, 404–413 (2003).
10. Meijer, L. et al. GSK-3 selective inhibitors derived from Tyrian purple indirubins.
Chem. Biol. (in the press).
11. Xu, C. et al. Feeder-free growth of undifferentiated human embryonic stem cells.
Nat. Biotechnol. 19, 971–974 (2001).
12. Okamoto, K. et al. A novel octamer binding transcription factor is differentially
expressed in mouse embryonic cells. Cell 60, 461–472 (1990).
13. Scholer, H.R., Ruppert, S., Suzuki, N., Chowdhury, K. & Gruss, P. New type of POU
domain in germ line-specific protein Oct-4. Nature 344, 435–439 (1990).
14. Rosner, M.H. et al. A POU-domain transcription factor in early stem cells and germ
cells of the mammalian embryo. Nature 345, 686–692 (1990).
15. Niwa, H., Burdon, T., Chambers, I. & Smith, A. Self-renewal of pluripotent embryonic
stem cells is mediated via activation of STAT3. Genes Dev. 12, 2048–2060 (1998).
16. Ramalho-Santos, M., Yoon, S., Matsuzaki, Y., Mulligan, R.C. & Melton, D.A.
“Stemness”: transcriptional profiling of embryonic and adult stem cells. Science
298, 597–600 (2002).
17. Brivanlou, A.H. & Darnell, J.E., Jr. Signal transduction and the control of gene
expression. Science 295, 813–818 (2002).
18. van Es, J.H., Barker, N. & Clevers, H. You Wnt some, you lose some: oncogenes in
the Wnt signaling pathway. Curr. Opin. Genet. Dev. 13, 28–33 (2003).
19. Moon, R.T., Bowerman, B., Boutros, M. & Perrimon, N. The promise and perils of
Wnt signaling through β-catenin. Science 296, 1644–1646 (2002).
20. Doble, B.W. & Woodgett, J.R. GSK-3: tricks of the trade for a multi-tasking kinase. J.
Cell Sci. 116, 1175–1186 (2003).
21. Korinek, V. et al. Constitutive transcriptional activation by a β-catenin-Tcf complex
in APC–/– colon carcinoma. Science 275, 1784–1787 (1997).
22. Tetsu, O. & McCormick, F. Beta-catenin regulates expression of cyclin D1 in colon
carcinoma cells. Nature 398, 422–426 (1999).
23. Nagai, T. et al. A variant of yellow fluorescent protein with fast and efficient matura-
tion for cell-biological applications. Nat. Biotechnol. 20, 87–90 (2002).
24. Hosler, B.A., LaRosa, G.J., Grippo, J.F. & Gudas, L.J. Expression of REX-1, a gene
containing zinc finger motifs, is rapidly reduced by retinoic acid in F9 teratocarci-
noma cells. Mol. Cell. Biol. 9, 5623–5629 (1989).
25. Molenaar, M. et al. XTcf-3 transcription factor mediates β-catenin-induced axis for-
mation in Xenopus embryos. Cell 86, 391–399 (1996).
26. Vonica, A. & Gumbiner, B.M. Zygotic Wnt activity is required for Brachyury expres-
sion in the early Xenopus laevis embryo. Dev. Biol. 250, 112–127 (2002).
27. Chambers, I. et al. Functional expression cloning of nanog, a pluripotency sustain-
ing factor in embryonic stem cells. Cell 113, 643–655 (2003).
28. Mitsui, K. et al. The homeoprotein Nanog is required for maintenance of pluripo-
tency in mouse epiblast and ES cells. Cell 113, 631–642 (2003).
29. Itskovitz-Eldor, J. et al. Differentiation of human embryonic stem cells into embryoid
bodies compromising the three embryonic germ layers. Mol. Med. 6, 88–95 (2000).
30. Kawasaki, H. et al. Generation of dopaminergic neurons and pigmented epithelia
from primate ES cells by stromal cell-derived inducing activity. Proc. Natl. Acad.
Sci. USA 99, 1580–1585 (2002).
31. Dani, C. et al. Paracrine induction of stem cell renewal by LIF-deficient cells: a new
ES cell regulatory pathway. Dev. Biol. 203, 149–162 (1998).
32. Niwa, H. Molecular mechanism to maintain stem cell renewal of ES cells. Cell
Struct. Funct. 26, 137–148 (2001).
33. Niwa, H., Miyazaki, J. & Smith, A.G. Quantitative expression of Oct-3/4 defines dif-
ferentiation, dedifferentiation or self-renewal of ES cells. Nat. Genet. 24, 372–376
(2000).
34. Huelsken, J. et al. Requirement for β-catenin in anterior-posterior axis formation in
mice. J. Cell Biol. 148, 567–578 (2000).
35. Conacci-Sorrell, M.E. et al. Nr-CAM is a target gene of the β-catenin/LEF-1 pathway
in melanoma and colon cancer and its expression enhances motility and confers
tumorigenesis. Genes Dev. 16, 2058–2072 (2002).
36. Lloyd, S., Fleming, T.P. & Collins, J.E. Expression of Wnt genes during mouse preim-
plantation development. Gene Expr. Patterns 3, 309–312 (2003).
37. Korinek, V. et al. Depletion of epithelial stem-cell compartments in the small intes-
tine of mice lacking Tcf-4. Nat. Genet. 19, 379–383 (1998).
38. Kubo, F., Takeichi, M. & Nakagawa, S. Wnt2b controls retinal cell differentiation at
the ciliary marginal zone. Development 130, 587–598 (2003).
39. Gat, U., DasGupta, R., Degenstein, L. & Fuchs, E. De novo hair follicle morphogen-
esis and hair tumors in mice expressing a truncated β-catenin in skin. Cell 95,
605–614 (1998).
40. Merrill, B.J., Gat, U., DasGupta, R. & Fuchs, E. Tcf3 and Lef1 regulate lineage dif-
ferentiation of multipotent stem cells in skin. Genes Dev. 15, 1688–1705 (2001).
41. Chenn, A. & Walsh, C.A. Regulation of cerebral cortical size by control of cell cycle
exit in neural precursors. Science 297, 365–369 (2002).
42. Reya, T. et al. A role for Wnt signalling in self-renewal of haematopoietic stem cells.
Nature 423, 409–414 (2003).
43. Giles, R.H., van Es, J.H. & Clevers, H. Caught up in a Wnt storm: Wnt signaling in
cancer. Biochim. Biophys. Acta 1653, 1–24 (2003).
44. Kielman, M.F. et al. Apc modulates embryonic stem-cell differentiation by control-
ling the dosage of β-catenin signaling. Nat. Genet. 32, 594–605 (2002).
45. Aubert, J., Dunstan, H., Chambers, I. & Smith, A. Functional gene screening in
embryonic stem cells implicates Wnt antagonism in neural differentiation. Nat.
Biotechnol. 20, 1240–1245 (2002).
46. Ding, S. et al. Synthetic small molecules that control stem cell fate. Proc. Natl.
Acad. Sci. USA 100, 7632–7637 (2003).
47. Willert, K. et al. Wnt proteins are lipid-modified and can act as stem cell growth fac-
tors. Nature 423, 448–452 (2003).
NATURE MEDICINE VOLUME 10 |NUMBER 1 |JANUARY 2004 63
© 2004 Nature Publishing Group http://www.nature.com/naturemedicine
