Requirement for COUP-TFI and II in the temporal specification of neural stem cells in CNS development.

Abstract

In the developing CNS, subtypes of neurons and glial cells are generated according to a schedule that is defined by cell-intrinsic mechanisms that function at the progenitor-cell level. However, no critical molecular switch for the temporal specification of CNS progenitor cells has been identified. We found that chicken ovalbumin upstream promoter-transcription factor I and II (Coup-tfI and Coup-tfII, also known as Nr2f1 and Nr2f2) are required for the temporal specification of neural stem/progenitor cells (NSPCs), including their acquisition of gliogenic competence, as demonstrated by their responsiveness to gliogenic cytokines. COUP-TFI and II are transiently co-expressed in the ventricular zone of the early embryonic CNS. The double knockdown of Coup-tfI/II in embryonic stem cell (ESC)-derived NSPCs and the developing mouse forebrain caused sustained neurogenesis and the prolonged generation of early-born neurons. These findings reveal a part of the timer mechanisms for generating diverse types of neurons and glial cells during CNS development.

Full text

Page 1

© 2008 Nature Publishing Group http://www.nature.com/natureneuroscience Requirement for COUP-TFI and II in the temporal
family of cytokines, which activate the JAK-STAT pathway through the
leukemia inhibitory factor receptor/gp130complex4,5, the activation of
Smad transcription factors by bone morphogenetic protein (BMP) 2/4
through their heterotrimeric serine/threonine kinase receptors6,7, and
Notch signaling8. All of these are known extrinsic signals for astro-
gliogenesis. Moreover, both STAT3 (in a transcriptional complex with
Smad1 and co-activator p300/CBP) and a Notch signaling effector,
CBF1/RBP-J,directlyactivatethetranscriptionofastrocyticgenes,such
as Gfap9,10.
Gliogenesis can also be initiated by the attenuation of neurogenesis
caused by the downregulation of neurogenic proneural genes, such as
the neurogenin (Neurog) genes, which encode helix-loop-helix tran-
scription factors, in a process that does not seem to require inducing
factors. Several in vitro studies support a model of neurogenesis
promotion by Neurog1, which inhibits gliogenesis by sequestering
the Smad/CBP/p300 transcriptional co-activator complex, thus pre-
venting its activation by gliogenic cytokines through activated STAT3
(ref. 11). Notably, even combinatorial deletions of proneural genes do
not consistently induce precocious differentiation of NSPCs into
astrocytes during the early neurogenic period, although they do so
specification of neural stem cells in CNS development
Hayato Naka1,2, Shiho Nakamura1,2, Takuya Shimazaki1,2& Hideyuki Okano1,2
In the developing CNS, subtypes of neurons and glial cells are generated according to a schedule that is defined by cell-intrinsic
mechanisms that function at the progenitor-cell level. However, no critical molecular switch for the temporal specification of CNS
progenitor cells has been identified. We found that chicken ovalbumin upstream promoter-transcription factor I and II (Coup-tfI
and Coup-tfII, also known as Nr2f1 and Nr2f2) are required for the temporal specification of neural stem/progenitor cells
(NSPCs), including their acquisition of gliogenic competence, as demonstrated by their responsiveness to gliogenic cytokines.
COUP-TFI and II are transiently co-expressed in the ventricular zone of the early embryonic CNS. The double knockdown of
Coup-tfI/II in embryonic stem cell (ESC)-derived NSPCs and the developing mouse forebrain caused sustained neurogenesis
and the prolonged generation of early-born neurons. These findings reveal a part of the timer mechanisms for generating
diverse types of neurons and glial cells during CNS development.
Neurogenesis largely precedes gliogenesis during CNS development in
vertebrates, and specific types of neurons are born in a spatiotem-
porally regulated manner1,2. The temporal regulation of CNS cytogen-
esis seems to be largely dependent on temporal changes in the
differentiation potential of NSPCs. Multipotent NSPCs isolated from
the early mouse embryonic cortex can sequentially generate specific
types of neurons and glial cells in vitro in their proper in vivo order3.
Several extrinsic and intrinsic mechanisms for the initiation of
gliogenesis by NSPCs have been proposed2. These include the IL-6
later in development, suggesting that NSPCs may have little
gliogenic potential at this time12,13. In addition, the expression of
these proneural genes is probably restricted to specified or committed
intermediate progenitors rather than to pluripotent progenitors13,
making them unlikely candidate regulators for the temporal specifica-
tion of NSPCs.
The neurogenic SHP2-MEK-ERK-Rsk pathway14and the anti-
gliogenic Neureglin-1–ErbB2/ErbB4 pathway15,16are involved in the
timing of gliogenesis. However, the disruption of these pathways only
accelerates astrocyte differentiation at lateembryonicstages, suggesting
that they regulate the timing of glial differentiation, but not the
gliogenic competency of NSPCs.
Recent studies support the idea that NSPCs change their ability to
respondtoenvironmental cues over time through epigenetic modifica-
tions, including chromatin remodeling and DNA methylation of glia-
specific genes, during the switch from neurogenesis to gliogenesis17,18.
The Brahma-related gene 1 (Brg1, also known as Smarca4) subunit of
the SWI/SNF chromatin remodeling protein complexes, which are
requiredforgliogenesisandthemaintenanceofNSPCs19,isacandidate
factor for these epigenetic changes. However, the mechanisms for the
induction of such epigenetic modifications of glia-specific genes in
NSPCs remain unknown.
The transition to gliogenesis also requires appropriate transcription
factor codes. For example, the deletion of Sox9, an HMG-type
transcription factor that is expressed in the ventricular zone of
embryonicspinal cord resultsinprolongedgenerationofmotoneurons
and V2 interneurons and concomitant inhibition of gliogenesis20. In
addition, nuclear factors 1A and 1B (NFIA/B), induced at the onset of
gliogenesis, appear to be essential gliogenic factors in the developing
Received 5 May; accepted 23 June; published online 24 August 2008; doi:10.1038/nn.2168
1Department of Physiology, Keio University School of Medicine, 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582, Japan.2Core Research for Evolutional Science
Technology, Solution-Oriented Research for Science and Technology, Japan Science and Technology Agency, Saitama 332-0012, Japan. Correspondence should be
addressed to H.O. (hidokano@sc.itc.keio.ac.jp) or T.S. (shimazak@sc.itc.keio.ac.jp).
1014VOLUME 11 [ NUMBER 9 [ SEPTEMBER 2008 NATURE NEUROSCIENCE
ARTICLES

Page 2

© 2008 Nature Publishing Group http://www.nature.com/natureneuroscience
chick spinal cord21. Thus, many genes and signaling pathways are
involved in different aspects of the timing of gliogenesis.
In contrast with what is known about gliogenesis, the molecular
mechanisms underlying the sequential generation of different types of
neurons in the developing CNS of vertebrates are largely unknown,
although some transcription factors that are expressed in restricted
CNS regions have been shown to be involved3,22,23.
In this study, we identified COUP-TFI and II as key regulators
of NSPC temporal specification during mouse CNS development.
Coup-tfI/II have been shown to be involved in patterning, differentia-
tion, cell migration, axonal projection and cortical arealization in the
developing CNS24–27. However, little is known about their functions
in NSPCs. Our in vitro and in vivo data from Coup-tfI/II double
knockdownmutantsprovideevidenceforarequirementforCoup-tfI/II
in the proper temporal specification of NSPCs, including their acquisi-
tion of gliogenic competency.
RESULTS
Neurogenic-to-gliogenic transition requires Coup-tfI/II
To identify the genes involved in the temporal specification of NSPCs,
we developedan invitro culture system that recapitulatesinvivo mouse
CNS development using mouse ESCs (Fig. 1a), in which NSPCs are
generatedfrom ESCsthrough embryoidbody formation and selectively
amplified as neurospheres28. Primary neurospheres that are derived
from embryoid bodies differentiate exclusively into neurons, and glio-
genesis is activated in subsequent generations of neurospheres, similar
to the situation invivo. This system enabled us
to retrospectively determine the differentiative
potential of NSPCs at each time point.
To identify genes that are involved in reg-
ulating the initiation of this temporal change,
we first compared the global gene-expression
profiles of primary and secondary neuro-
spheres by DNA microarray. Of the 20,280
EB formation
Neurospheres formation
Lentivirus infection
ESCs
a
c
ef
g
h
d
b
EB
1. Primary neurosphere
2. Secondary neurosphere
3. Tertiary neurosphere
Differentiation
Differentiation
Differentiation
0.5
0.0
Positive
100
Oligodendrocyte
Astrocyte
Neuron
80
60
40
20
0
control
Percentage of cell types
Percentage of
neurons
Primary neurosphere
Cortex
Lateral ganglionic eminence
Negative
control
CT
KD I
KD II
KD I + II
CTKD I KD II KD
I + II
1.0
1.5
******* *** *
NSNSNS NS
COUP-TFI
COUP-TFII
Expression level
20
15
10
COUP-TFI
Neurosphere
Culture stage
12ESC EB
**
**
**
3
COUP-TFII
α-tubulin
5
CT KD CTKD
E10.5E12.5 E14.5
COUP-TFI / COUP-TFII / Hoechst
LV
LV
LV
LV
LV
LV
LV
LV
VZ
VZ
VZ
VZ
E16.5 P0P2
CTKD I
GFP / βIII-tubulin / GFAP
KD IIKD I + II
GFP / βIII-tubulin / GFAP
NS
0
Figure 1 Coup-tfI/II double knockdown inhibits
the neurogenic-to-gliogenic transition in vitro.
Coup-tfI/II knockdown in ESC-derived
neurospheres resulted in exclusive neurogenesis,
even at the gliogenic stage. Cells infected with
lentivirus encoding control or Coup-tfI/II shRNA
are designated as CTor KD, respectively.
(a) Strategy for the functional screening of
candidate genes involved in the temporal
specification of NSPCs using ESCs. (b) Western
blot of COUP-TFI/II in stable 293T transformants
infected with lentiviruses expressing shRNAs
that target both Coup-tfI and II (KD I + II),
Coup-tfI (KD I) or Coup-tfII (KD II) (t test, n ¼ 3;
*P o 0.05, **P o 0.01 and ***P o 0.001
versus positive control). (c) Representative
micrographs of the immunocytochemistry of
differentiated tertiary neurospheres infected with
lentiviruses (GFP-positive cells). (d) Percentage of
bIII-tubulin–positive neurons, GFAP-positive
astrocytes and O4-positive oligodendrocytes
among virus-infected cells from tertiary
neurospheres (t test, n (Z200 cells) ¼ 3,
**P o 0.01 versus CT). (e,f) Coup-tfI/II
knockdown in tertiary neurospheres at the onset
of differentiation did not induce any phenotypic
change (t test, n (Z300 cells) ¼ 3; NS,
P 4 0.05 versus CT). (g) Expression of COUP-
TFI/II in ESCs, embryoid bodies and primary to
tertiary neurospheres. (h) Expression of COUP-
TFI/II in the developing forebrain. At E12.5,
COUP-TFI/II are co-expressed widely and appear
to be localized both in cytoplasm and nucleus in
the cortex at this stage. After E14.5, they are
localized only in nucleus. After E16.5, their
expression in the ventricular zone diminished with
development, whereas that in the marginal zone
remained but was segregated into distinct
populations. Co-expression of COUP-TFI/II was
also observed in undifferentiated neurospheres
(see also Supplementary Fig. 1). EB, embryoid
body; LV, lateral ventricle; VZ, ventricular zone.
Scale bars represent 50 mm (c,e) and 100 mm (h).
Error bars represent s.e.m.
NATURE NEUROSCIENCE VOLUME 11 [ NUMBER 9 [ SEPTEMBER 20081015
ARTICLES

Page 3

© 2008 Nature Publishing Group http://www.nature.com/natureneuroscience
genes that were screened by microarray analysis, we found 617 with a
more than twofold higher expression level in primary neurospheres
compared with secondary neurospheres and 838 that were similarly
elevated in secondary versus primary neurospheres. We then carried
out functional screening of genes that were expressed more highly in
primary neurospheres by lentivirus-mediated overexpression and
knockdown by short-hairpin RNAs (shRNAs) in ESC-derived neuro-
spheres (Supplementary Table 1 online).
The development of pluripotent stem cells requires genetic and
epigenetic regulation29, prompting us to focus on transcription regu-
lators as candidate molecules in the functional screening. We infected
primary neurospheres with lentiviruses bearing coding sequences or
knockdown shRNAs for these factors at the time of plating and
cultured the neurospheres to the tertiary stage (Fig. 1a). Some neuro-
spheres were allowed to differentiate at each passage and were immu-
nostained with antibodies specific to bIII-tubulin, GFAP and O4, to
identify neurons, astrocytes and oligodendrocytes, respectively. Among
the genes that we examined, only Coup-tfI/II knockdown resulted in a
significant phenotype, whichwas defined as a high neurogenic potential
being retained even in tertiary neurospheres (Fig. 1b–d), without sub-
stantial change in the efficiency of neurosphere formation (data not
shown). A more severe phenotype was observed in double-knockdown
compared with single-knockdown neurospheres (Fig. 1b–d). In
contrast, control neurospheres, which were
transduced with a lentivirus vector expressing
control shRNA, mainly differentiated into glial
cells.Notably,overexpressionofCoup-tfI/IIdid
not alter the differentiation of neurospheres.
NotethatknockdownofCoup-tfI/IIduringthe
differentiation of tertiary neurospheresdid not
alter glial differentiation or survival (Fig. 1e,f).
The expression domains of Coup-tfI and II
are reported to partially overlap, but are dis-
tinct at the cellular level in the developing
CNS25,30. However, our immunohistochem-
ical analysis revealed that COUP-TFI and II
were widely co-expressed in the embryonic
day (E) 12.5 mouse brain. Their expression in
the ventricular zone was transiently upregu-
lated at E12.5 and was then diminished with
development before the appearance of astro-
cytes, a finding that is consistent with their
expressioninESC-derived
in vitro. As development progressed, the
expression of both Coup-tfI/II persisted in
the marginal zone, but it was segregated into
distinct cell populations at E14.5 and later
time points (Fig. 1g,h and Supplementary
Fig. 1 online). Given that, in terms of their
molecular functions, including their DNA-
binding specificity, COUP-TFI and II are
highly conserved and
from each other31, these results suggest that
Coup-tfI and II can compensate for each
other’s function in the developing NSPCs;
thus, their double knockdown resulted in
the preservation of the highly neurogenic
potential of ESC-derived NSPCs over time.
To confirm the specificity of the Coup-tfI/II
knockdown that was responsible for the neu-
rogenic phenotype, we carried out a rescue
neurospheres
indistinguishable
experiment using mutant COUP-TFI/II (mut–COUP-TFI/II) with
silent mutations in the target nucleotide sequences for the shRNAs;
thus, the mutant COUP-TFI/II were not knocked down (Fig. 2a and
Supplementary Fig. 2 online). In this experiment, we set a temporal
gap between the starting expression times of the knockdown shRNAs
and the rescue constructs. The time lag was intended to confirm that
the knockdown phenotype was not the result of the selective amplifica-
tion of neuron-restricted precursors. However, this was unlikely, as
committed neuronal precursors generate almost no neurospheres32.
For the rescue experiment, we infected primary neurospheres with
the Coup-tfI/II knockdown lentivirus and cultured them to the qua-
ternary passage. They were then infected with lentivirus encoding mut–
COUP-TFI/II and a hygromycin-resistance gene. The infected NSPCs
were selected for with hygromycin until senary (sixth generation)
neurospheres were formed (Fig. 2b). Control senary neurospheres
differentiated predominantly into astrocytes, but the neurospheres
expressing Coup-tfI/II shRNAs still predominantly differentiated into
neurons. However, the delayed expression of mut–COUP-TFI/II com-
pletely rescued the knockdown phenotype (Fig. 2c,d) without any
significant change in the amount of cell death, as assessed by cleaved
caspase-3 immunocytochemistry (P 4 0.05; Supplementary Fig. 3
online). Taken together, these results indicate that the expression of
COUP-TFIandIIinNSPCsisnecessaryfortheinductionofgliogenesis.
ab
c
ESCs
EB
mut–COUP-TFI / II
1. Primary neurosphere
2. Secondary neurosphere
3. Tertiary neurosphere
4. Quaternary neurosphere
5. Quinary neurosphere
Hygromycin selection
6. Senary neurosphere
Hygromycin selection
Differentiation
shRNA Coup-tfI/II
GAC AAG TCG AGC GGC AAG CAC TAC GGC
D K S S G K H Y G
GAC AAG AGT TCC GGA AAG CAT TAC GGC
D K S S G K H Y G
shRNA Coup-tfI/II
COUP-TFI / II
mut-COUP-TFI / II
KD
mut–COUP-TFI mut–COUP-TFII
GFP / βlll-tubulin / GFAP
mut–COUP-TFI + II
CT
d
0
20
40
60
80
100
0
20
40
60
80
NS
NS
NS
100
136
Culture stageCulture stageCulture stage
Percentage of neurons
Percentage of astrocytes
Percentage of oligodendrocytes
**
***
**
6
136
0
10
20
30
40
50
13
CT
KD
mut–COUP-TFI
mut–COUP-TFII
mut–COUP-TFI + II
Figure 2 Coup-tfI/II knockdown phenotype is rescued by expression of knockdown-resistant mutant
COUP-TFI/II. (a) Silent mutations in mutant COUP-TFI/II (mut–COUP-TFI/II) that prevented their
knockdown by Coup-tfI/II–specific shRNA. (b) Schema of the rescue experiment using lentivirus-
mediated expression of mut–COUP-TFI/II. (c,d) Delayed expression of the mut–Coup-tfI/II rescued the
defect in gliogenesis caused by Coup-tfI/II knockdown. The percentage of bIII-tubulin–positive neurons
was reduced to the control level after infection by mut–COUP-TFI/II lentivirus, whereas the percentages
of GFAP-positive astrocytes and O4-positive oligodendrocytes were increased to the levels of control
cultures (t test, n (Z200 cells) ¼ 3, P 4 0.05; **P o 0.01, ***P o 0.001 versus CT). Scale bar
represents 50 mm. Error bars represent s.e.m.
1016VOLUME 11 [ NUMBER 9 [ SEPTEMBER 2008 NATURE NEUROSCIENCE
ARTICLES

Page 4

© 2008 Nature Publishing Group http://www.nature.com/natureneuroscience
The acquisition of gliogenic competency requires Coup-tfI/II
Coup-tfI/IImight elicit gliogenesis by two different processes: temporal
specification of NSPCs to permit gliogenesis and/or glial fate specifica-
tion concomitant with or via the inhibition of neurogenesis. Given the
expression patterns of COUP-TFI/II in NSPCs in vivo and in vitro,
which peak before gliogenesis, we tested whether Coup-tfI/II were
required for NSPCs to acquire gliogenic competency by analyzing the
epigenetic status of the STAT3-binding site in the Gfap promoter. This
site is responsible for the JAK-STAT pathway–dependent expression
of Gfap33and is epigenetically silenced in early neurogenic CNS
progenitors, rendering them unresponsive to astrocyte-inducing
environmental signals17,18.
We carried out chromatin immunoprecipitation (ChIP) analyses for
the histone H3 modification at the STAT3-binding site in ESC-derived
tertiary neurospheres using antibodies specific for acetylated histone
H3 (Ac-H3) and dimethylated H3K4 (2Me-H3K4), which are both
associated with gene activation, and for dimethylated H3K9 (2Me-
H3K9), which is associated with gene silencing. The ChIP results
indicated that there were significantly more 2Me-H3K9 (P o 0.05)
and significantly less Ac-H3 and 2Me-H3K4
(P o 0.01 and P o 0.05, respectively) in the
Coup-tfI/II knockdown neurospheres than in
thecontrolneurospheres,indicatingthat there
was greater silencing of the STAT3-binding
site in the Gfap promoter of the Coup-tfI/II
knockdown neurospheres (Fig. 3a,b). The
extent of CpG methylation in the Gfap pro-
moter, including the STAT3-binding site, was
also significantly higher in Coup-tfI/II knock-
down neurospheres (P o 0.01 in total ten
CpG sites and P o 0.05 in CpG site of the
STAT3-binding region; Fig. 3c–e).
We further assessed the ability of Coup-tfI/II
knockdown neurospheres to respond to the
gliogenic cytokines LIF and BMP2. Consistent
with our analysis of the epigenetic statusof the
Gfappromoter,Coup-tfI/IIknockdownneuro-
spheres resisted the promotion of astroglio-
genesis by these cytokines (Fig. 3f–i). These
results suggest that Coup-tfI/II are required for
thetemporallyregulatedacquisitionofgliogeniccompetencybyNSPCs.
Coup-tfI/II affect the temporal change of the neuropotency
We next sought to determine whether Coup-tfI/IIwere also involved in
the temporal change in the neuropotency of NSPCs. In our culture
system,primaryneurospheresderivedfromESCsshowedforebrain-to-
hindbrain identities, as determined by the expression of region-specific
markers (Supplementary Fig. 4 online), and generated a large number
of Islet-1–expressing neurons.
Isl1 is a LIM-domain homeobox gene that isspecificallyexpressed in
a subpopulation of early-born neurons in the developing ventral
forebrain, hindbrain and spinal cord34–36, and the generation of these
neurons from ESC-derived neurospheres markedly decreased over
successive generations of neurospheres (Fig. 4a,b). We examined
whether Coup-tfI/II knockdown altered their generation. Coup-tfI/II
knockdown neurospheres showed a significantly increased and
sustained generation of Isl1-positive neurons (P o 0.05; Fig. 4a,b).
As the efficiency of neurosphere formation was not altered by
Coup-tfI/II knockdown, the sustained generation of Isl1-positive
ab
f
g
CT
KD
InputIgG
Ac-H3
2Me-H3K42Me-H3K9
LIF
BMP
Primary
Secondary
CTKDCT
GFP / βlll-tubulin / GFAP
KD
Tertiary
h
c
de
CTKD
i
LIF + BMP
CTKD
0.0
0.5
1.0
1.5
Ac-H32Me-
H3K4
2Me-
H3K9
Fold change
CT
*
KD
***
0
20
40
60
80
100
0
20
40
60
80
100
CT KDCT KD
Percentage of cytosine
methylation
**
Percentage of
cytosine methylation
*
0
20
40
60
80
100
123
Culture stage Culture stageCulture stage
Percentage of astrocytes
CT
KD
CT
KD
CT
KD
*
**
0
20
40
60
80
100
0
20
40
60
80
100
123
Percentage of astrocytes
**
**
123
Percentage of astrocytes
**
*
Figure 3 Coup-tfI/II are required to free Gfap
promoter from epigenetic silencing and to confer
responsiveness to gliogenic cytokines. (a,b) Coup-
tfI/II knockdown altered the epigenetic status of
histone H3 in the Gfap promoter in ESC-derived
neurospheres. Ac-H3, 2Me-H3K4 and 2Me-H3K9
in tertiary neurospheres were analyzed by ChIP
assays. Representative PCR results and
densitometric quantification are shown in a
and b, respectively (t test, n ¼ 3). (c–e) CpG
methylation status of the Gfap promoter was
analyzed by bisulfite sequencing. Arrows in c
indicate the CpG site in the STAT3-binding
sequence. The extent of CpG methylation in a
total of ten sites and in the STAT3-binding region
is shown in d and e, respectively (t test, n (Z13
clones) ¼ 3). (f–i) Coup-tfI/II knockdown NSPCs
resist the promotion of gliogenesis by LIF (10 ng
ml–1, g), BMP (100 ng ml–1, h) or LIF + BMP (i)
(t test, n (Z200 cells) ¼ 3). Scale bars represent
50 mm. Error bars represent s.e.m. *P o 0.05,
**P o 0.01 versus CT.
NATURE NEUROSCIENCE VOLUME 11 [ NUMBER 9 [ SEPTEMBER 20081017
ARTICLES

Page 5

© 2008 Nature Publishing Group http://www.nature.com/natureneuroscience
neurons cannot be attributed to the selective expansion of a specific
neurosphere-formingprogenitor
extinguished during development. Furthermore, the decrease in
the proportion of Isl1-positive neurons in tertiary knockdown
neurospheres, despite maintenance of the neurogenic phenotype
(Fig. 1c,d), suggested that the sustained
generation of Isl1-positive neurons was
also not the result of an alteration of sub-
type specification of neuronal precursors dur-
ing differentiation.
populationthatis normally
We also tested whether Coup-tfI/II knock-
down could juvenilize advanced-generation
NSPCs, which usually do not generate Isl1-
positive neurons. We observed a limited recovery of the early neuro-
potentialforgeneratingIsl1-positiveneurons(Fig.4c,d),suggestingthat
a subpopulation of NSPCs was juvenilized by Coup-tfI/II knockdown.
Taking into account the fact that the expression levels of Coup-tfI/II
in NSPCs were highest in primary neurospheres and gradually
a
Primary TertiarySecondary
CT
KD
GFP / Islet-1
Quinary
KD
QuaternarySenary
GFP / Islet-1
c
b
d
0
5
10
15
20
456
Culture stage
Percentage of
Isl-1–positive neurons
CT
KD
***
0
20
40
60
80
100
123
Culture stage
Percentage of
Isl-1–positive neurons
CT
KD
*
**
*
Figure 4 Coup-tfI/II are involved in the temporal
restriction of NSPC neuropotency. (a,b) Coup-tfI/II
knockdown in ESC-derived neurospheres
extended the production period of Islet-1–
positive neurons. Representative micrographs
of the immunocytochemistry of differentiated
neurospheres (primary to tertiary) and the
proportion of Islet-1–positive population among
GFP-positive neurons are shown in a and b,
respectively (t test, n (Z200 cells) ¼ 3).
(c,d) Coup-tfI/II knockdown revived the generation
of Islet-1–positive neurons from ESC-derived
neurospheres, even in quaternary and subsequent
generations of neurospheres, to a limited extent
(quaternary neurospheres were infected by shRNA
lentivirus at the time of their formation, and we
quantified the Islet-1–positive population among
GFP-positive neurons; t test, n (Z200 cells) ¼ 3).
Scale bars represent 200 mm (a) and 50 mm (c).
Error bars represent s.e.m. *P o 0.05,
**P o 0.01 versus CT.
Others
Astrocyte
Neuron
Others
Neuron
OPC
80
100
60
KD
CT
Percentages of cell types
Percentage of cell types
40
20
CT KD
CTKD
Sox10 / Olig2
GFP / Sox10
GFP / GFAP / Hoechst GFP / NeuN / Hoechst
CTKD
Medial ganglionic eminence
Cortex
ab
cd
GFP / Olig2
**
*
*
**
0
80
100
60
40
20
0
Figure 5 Coup-tfI/II knockdown inhibits initiation
of gliogenesis in developing forebrain. (a–d) CTor
KD lentivirus was microinjected in utero into the
cerebral ventricle of mouse embryos at E12.5
(a,b) or E10.5 (c,d) and the fates of the infected
cells at P20 (a,b) or E16.5 (c,d) were determined
by immunohistochemical analysis. Coup-tfI/II
knockdown significantly increased the percentage
of cells that had differentiated into NeuN-positive
neurons and decreased the percentage of cells
that differentiated into GFAP-positive astrocytes in
the cerebral cortex at P20 (t test, n (Z200 cells,
Z independent three sections) ¼ 5 independent
embryos, **P o 0.01 versus CT; a,b). Arrowheads
show non-neuronal cells, which include GFAP-
positive astrocytes. Most GFAP-negative non-
neuronal cells, counted as ‘Others’, looked like
immature astrocytes. Higher magnification images
of the cells indicated with arrows are shown in
insets. The number of Sox10 and Olig2 double-
positive OPCs in the medial ganglionic eminence
was significantly decreased by Coup-tfI/II
knockdown, whereas the number of NeuN-positive
neurons was increased (t test, n (Z100 cells) ¼
3, *P o 0.05 versus CT; c,d). Arrowheads
indicate GFP, Sox10 and Olig2 triple-positive
cells. Arrows indicate cells shown in insets. Scale
bars represent 100 mm (lower magnification
images in a), 20 mm (insets in a and c) and
50 mm (lower magnification images in c). Error
bars represent s.e.m.
1018VOLUME 11 [ NUMBER 9 [ SEPTEMBER 2008 NATURE NEUROSCIENCE
ARTICLES

Page 6

© 2008 Nature Publishing Group http://www.nature.com/natureneuroscience
decreasedwith time(Fig.1g), these resultssuggest thatCoup-tfI/IImay
function as part of the trigger for the temporal change in the
neuropotency of NSPCs.
Coup-tfI/II knockdown phenotype in the developing brain
To confirm the in vivo role of Coup-tfI/II that was revealed by our in
vitro analyses, we studied the effects of Coup-tfI/II knockdown in the
developing mouse brain. Consistent with our in vitro results, cells
infected with the Coup-tfI/II knockdown lentivirus at E12.5 were
exclusively fated to become neurons (86.5 ± 3.0%) in the cerebral
cortex at P20, whereas only 42.2% of the cells infected with control
lentivirus did so (Fig. 5a,b); furthermore, the generation of GFAP-
positive astrocytes was significantly reduced by Coup-tfI/II knockdown
compared with the control (P o 0.01). Moreover, transduction of the
knockdown construct at E10.5 resulted in a significant increase in
neurons at the expense of oligodendrocyte precursor cells defined by
the co-expression of Sox10 and the Olig2 transcription factors37(11 to
3.4%), accompanied by a compensatory increase in the proportion of
neurons (73.8 to 80.4%) in the medial ganglionic eminence at E16.5
(Fig. 5c,d and Supplementary Fig. 5 online).
In addition, transduction of the knockdown construct at E10.5
induced prolonged generation of neurons expressing markers for
early-born neurons in the forebrain (for example, Isl1, DARPP-32
andTbr1)38,39.The knockdowninducedlateectopicgenerationofIsl1-
positive or DARPP-32–positive neurons in the striatum, which were
recognized by BrdU incorporation at E15.5. Similarly, an unusual
sustained production of early-born Tbr1 single-positive neurons
(which are normally barely generated after E15.5) at the expense
of late-born neurons expressing Brn2 (ref. 39) was induced in the
caudal cerebral cortex (Fig. 6a–c and Supplementary Fig. 6 online).
This is consistent with a phenotype displayed by the cortex-specific
Coup-tfI knockout mouse, which shows the ectopic expression of Tbr1
in the upper layer of the occipital cortex at P8 (ref. 27). We did not
detect any late ectopic production of Reelin-positive Cajal-Retzius
neurons in brains infected with the knockdown lentivirus in these
experiments (data not shown). Thus, the Coup-tfI/II knockdown
altered the timing of neurogenesis and gliogenesis, even in vivo, in a
cell-autonomous fashion.
Notably, we observed a similar knockdown phenotype, that is,
enhanced neurogenesis accompanied by enhanced generation of
early-born neurons, in primary cultured neurospheres derived from
various CNS regions at E10.5. The neuronal subtypes were specific for
each region and this phenotype was not accompanied by any change in
the efficiency of neurosphere formation (Supplementary Fig. 7 online
and data not shown), suggesting that Coup-tfI/II are required for the
proper sequential generation of diverse types of neurons and the
subsequent gliogenesis throughout the CNS. This result, in combina-
tion with our findings from ESC-derived neurospheres, further sug-
gests that the prolonged generation of early-born neurons, seen in the
Coup-tfI/II knockdown cells in vivo, resulted not from the enrichment
of a specific population of NSPCs or fate-switch in differentiating
neuronal precursors, but from the altered temporal specification of
NSPCs. The region-dependent differences in the strength of the
neurogenic phenotype induced by the knockdown (fewer neurons
were generated by neurospheres from NSPCs of the hindbrain and
spinal cord, where gliogenesis starts earlier than in the rostral brain;
Supplementary Fig. 7) may reflect the stage-dependent function of
Coup-tfI/II, further supporting the possibility that they are required for
the temporalspecification of NSPCs to permit gliogenesis, but not glial
differentiation itself.
Coup-tfI/II function as transcriptional repressors
Coup-tfIand IIwere originally identified as transcriptional activators for
thechickenovalbuminpromoterandwerelaterfoundtoalsofunctionas
Striatum Cortex
a
b
Striatum
GFP / BrdU / Tbr1 / Brn2
c
80
Percentage of neurons in
GFP / BrdU / DARPP-32GFP / BrdU / Islet-1
Cortex
GFP / Tbr1
P2E16.5
CT
KD
DL
UL
DL
UL
0
20
40
60
Tbr1BrdU +
Tbr1
Brn2 BrdU +
Brn2
GFP-positive cells
CT
KD
****
0
20
40
60
80
Isl-1BrdU +
Isl-1
DARPP-32 BrdU +
DARPP-32
Percentage of neurons in
GFP-positive cells
CT
KD
*****
Figure 6 Coup-tfI/II knockdown inhibits temporal
specification of NSPCs in developing forebrain. CT
or KD lentivirus was microinjected in utero into the
cerebral ventricle of mouse embryos at E10.5 and
the fates of the infected cells were determined at
P2 or E16.5 by immunohistochemical analysis.
Cells born at E15.5 were labeled by BrdU
incorporation. (a) Left, representative confocal
micrographs from the immunohistochemical
analysis of brains infected with KD lentivirus and
examined at P2. Right, high-dose lentivirus-
infected cortices at E16.5. UL and DL indicate
upper-layer cortical plate and deep-layer cortical
plate, respectively. Ectopic expression of Tbr1 in
UL neurons, which presumably could have been
late-born neurons, was induced by knockdown.
Note that ectopically induced Tbr1-positive
neurons by high-dose KD lentivirus infection
showed abnormal multipolar morphology (see also
Supplementary Movie 1 online). (b,c) Coup-tfI/II
knockdown resulted in prolonged and ectopic
generation of early-born neurons persisting until at
least E15.5 in both striatum (Islet-1 and DARPP-
32 positive) and caudal cerebral cortex (Tbr1
positive). In the caudal cerebral cortex, Coup-tfI/II
knockdown decreased the proportion of late-born
neurons (Brn2 positive). In this analysis, we
regarded Tbr1 and Brn2 double-positive cells as
Brn2-positive neurons (t test, n (Z200 cells) ¼ 3;
*P o 0.05; **P o 0.01 versus CT). Scale bars
represent 20 mm (P2 in a) and 50 mm (E16.5
in a). Error bars represent s.e.m.
NATURE NEUROSCIENCE VOLUME 11 [ NUMBER 9 [ SEPTEMBER 20081019
ARTICLES

Page 7

© 2008 Nature Publishing Group http://www.nature.com/natureneuroscience
repressors in different contexts40. We therefore investigated whether
the transcriptional activating or repressing functions of COUP-TFI/II
were responsible for the acquisition of gliogenic competency.
Constructs encoding dominant-activating and dominant-repressing
forms of COUP-TFI/II were made by fusing the activation domain
of herpessimplex viral protein Vp16 (ref. 41) or the repression domain
of Drosophila engrailed (En) to the N termini of COUP-TFI and
II, respectively, in which the original C-terminal transcription-regulat-
ing domains had been deleted. In addition, a ‘tagged’ construct
pair was prepared by fusing the Flag tag to COUP-TFI and II with
intact C termini (Fig. 7a). We expressed either one of these constructs
or an empty vector in NSPCs by lentiviral vectors and assessed
NSPC differentiation.
The expression of the dominant-activating forms (VP-I/II) signifi-
cantly enhanced neurogenesis in tertiary neurospheres (P o 0.05),
whereas expression of the Flag-tagged wild-type proteins (F-I/II) or the
dominant-repressing forms (En-I/II) did not alter it when compared
withcontrolNSPCsthatreceivedtheemptyvector(Fig.7b,c).Notethat
neither F-I/II nor En-I/II induced precocious gliogenesis in early
neurosphere generations (data not shown). Furthermore, we carried
out a rescue experiment using En fusion mut–COUP-TFI/II constructs
as described above (Fig. 2) and found that the En fusion mut–COUP-
TFI/II could cause complete recovery of the gliogenesis by NSPCs
expressing the Coup-tfI/II shRNAs (Fig. 7d,e). These results suggest
that the transcriptional repressor function of Coup-tfI/II is required for
the gliogenesis of ESC-derived NSPCs, but is not sufficient for and does
not solely induce it. This means that COUP-TFI and II might need
specific co-regulators to induce gliogenesis.
Downstream of Coup-tfI/II
Finally,toelucidatethemolecularmechanisms
of the temporal specification involving Coup-
tfI/II, we analyzed the effects of Coup-tfI/II
knockdown on the temporal change of global gene-expression profiles
in ESC-derived neurospheres using DNA microarrays and focused on
genes that normally show temporal changes in expression levels during
development (Fig. 7f). Among such genes, the number whose expres-
sion levels changed more than twice as a result of Coup-tfI/II knock-
downwasrelativelysmall,suggestingthatCoup-tfI/IIareresponsiblefor
only some of the temporal changes that occur in the cellular context of
NSPCs. Moreover, in tertiary neurospheres,Coup-tfI/IIknockdown did
not substantially alter the expression levels of some known genes that
areprobablyinvolvedinthetimingofgliogenesis(Table1);therefore,it
is possible that none of these genes are downstream of Coup-tfI/II.
Although the levels of some proneural genes (Neurog2, NeuroD and
Mash1) appeared to increase in the tertiary Coup-tfI/II knockdown
neurospheres, their expression levels did not fully correlate with the
predicted differentiation phenotypes. These results suggest that a
certain part of the temporal specification of NSPCs is regulated by
molecular mechanisms involving Coup-tfI/II operating upstream of, at
least, the epigenetic modifications of a glia-specific gene.
DISCUSSION
Three mechanisms have been proposed so far for the temporally
regulated specification of NSPCs for the switch from neurogenesis to
gliogenesis. One is the attenuation of neurogenesis by reduced expres-
sion levels of proneural genes, such as Neurogs11. Another is the
induction of gliogenesis through the activation of proglial genes,
such as NFIs21. Finally, it has been suggested that temporally regulated
epigeneticmodificationsofglia-specificgenesregulatetheirresponseto
gliogeniccytokines17,18. Our data suggest the existence of an additional
b
a
c
COUP-TFI / II A/B C DE/F
F-I / II
VP-I / II
En-I / II
A/B C DE/F
A/B C DE
A/B C DE
VP16
Engrailed
Flag
CT
Primary
Tertiary
KD CTKD
f
d
e
VP-IVP-II
En-I
En-II
VP-IVP-II
En-IEn-II
GFP / O4
En-mIEn-mIIEn-mIEn-mII
0
Empty
20
40
60
80
100
F-I
F-II
VP-I
VP-II
En-I
En-II
Percentage of neurons
NS NSNSNS
**
0
20
40
60
80
100
0
20
40
60
80
100
1
Culture stage
36
Culture stageCulture stage
Percentage of neurons
**
NS
136
Percentage of astrocytes
***
NS
0
10
20
30
40
50
136
Percentage of oligodendrocytes
CT
KD
En-mI
En-mII
*
**
GFP / βlll-tubulin
GFP / O4 / GFAP
GFP / βlll-tubulin / GFAP
Figure 7 Transcriptional repressor function of
Coup-tfI/II is required for temporal specification of
NSPCs. (a) COUP-TF fusion proteins. Flag tag (F-I/
II), Vp16 (VP-I/II) or engrailed (En-I/II) were fused
with the N terminus of COUP-TFI/II. The
C-terminal 20 amino acids of COUP-TFI/II,
responsible for their original transcriptional
functions, were deleted in Vp16 and En fusion
proteins. (b,c) Forced expression of VP-I or II
showed a similar phenotype as that caused by
Coup-tfI/II knockdown (t test, n (Z200 cells) ¼ 3,
*P o 0.05 versus CT). (d,e) Rescue of Coup-tfI/II
knockdown phenotype by dominant-repressing
forms of COUP-TFI/II. En (En-mI/II) was fused
with the N terminus of each mut–COUP-TF. The
C-terminal 20 amino acids of mut–COUP-TF were
deleted. The rescue experiment was carried out as
described in Figure 2. Complete recovery of the
defect in gliogenesis caused by Coup-tfI/II
knockdown was observed by delayed expression of
En-mI/II (t test, n (Z200 cells) ¼ 3, P 4 0.05;
**P o 0.01, ***P o 0.001 versus CT).
(f) Coup-tfI/II knockdown partly altered temporal
changes in global gene-expression profiles during
long-term culture of ESC-derived neurospheres.
Global gene-expression profiles of primary and
tertiary CTand KD neurospheres were compared
by DNA microarray analysis. Red and green bars
indicate genes expressed at relatively high and
low levels, respectively, in each stage and
condition. Scale bars represent 50 mm.
Error bars represent s.e.m.
1020VOLUME 11 [ NUMBER 9 [ SEPTEMBER 2008 NATURE NEUROSCIENCE
ARTICLES

Page 8

© 2008 Nature Publishing Group http://www.nature.com/natureneuroscience
Table 1 Comparison of principal gene-expression profiles between control and knockdown neurospheres in primary and tertiary culture stages
Primary control
Primary knockdown
Tertiary control
Tertiary knockdown
Association
Gene name
Accession no.
Intensity
Fold change
Intensity
Fold change
Intensity
Fold change
Intensity
Fold change
D/C
Neurogenesis
bHLH
Neurog1 (Ngn1)
NM_010896
2.43
1.00
1.62
0.66
1.20
0.49
1.41
0.58
1.18
Neurog2 (Ngn2)
NM_009718
7.80
1.00
2.55
0.33
8.02
1.03
27.34
3.51
3.41
Neurod1 (NeuroD)
NM_010894
2.90
1.00
1.52
0.52
0.64
0.22
3.60
1.24
5.60
Ascl1 (Mash1)
NM_008553
17.20
1.00
9.52
0.55
14.95
0.87
30.86
1.79
2.06
Atoh1 (Math1)
NM_007500
0.53
1.00
0.23
0.43
0.18
0.35
0.20
0.39
1.12
Neurod4 (Math3)
NM_007501
0.43
1.00
0.26
0.62
0.52
1.23
0.88
2.06
1.68
Other transcriptional regulators
Pax6
NM_013627
6.25
1.00
5.26
0.84
2.40
0.38
3.19
0.51
1.33
Emx2
NM_010132
3.09
1.00
3.75
1.21
2.02
0.65
2.25
0.73
1.11
Foxg1
NM_008241
88.55
1.00
120.88
1.36
19.19
0.22
45.79
0.52
2.39
Gliogenesis
bHLH, HLH
Olig1
NM_016968
7.58
1.00
3.96
0.52
23.52
3.10
13.60
1.80
0.58
Olig2
NM_016967
4.48
1.00
2.95
0.66
15.98
3.57
14.27
3.19
0.89
Id1
NM_010495
83.86
1.00
150.46
1.79
86.67
1.03
77.57
0.93
0.90
Id2
NM_010496
103.08
1.00
126.74
1.23
91.60
0.89
82.84
0.80
0.90
Other transcriptional regulators
Nfia
NM_010905
10.96
1.00
9.90
0.90
40.53
3.70
36.16
3.30
0.89
Nfib
NM_008687
14.39
1.00
8.03
0.56
67.38
4.68
68.81
4.78
1.02
Nfic
NM_008688
7.20
1.00
7.11
0.99
47.52
6.60
21.95
3.05
0.46
Nfix
NM_010906
0.46
1.00
0.24
0.52
24.71
53.17
10.92
23.50
0.44
Tal1 (Scl)
NM_011527
11.82
1.00
4.65
0.39
11.84
1.00
18.71
1.58
1.58
Ncor1 (N-CoR)
NM_011308
16.31
1.00
14.08
0.86
17.97
1.10
16.26
1.00
0.91
Sox9
NM_011448
189.82
1.00
159.66
0.84
315.43
1.66
241.30
1.27
0.76
Sox10
BC023356
6.47
1.00
6.59
1.02
6.22
0.96
5.07
0.78
0.81
Rest
NM_011263
53.62
1.00
69.68
1.30
44.18
0.82
32.62
0.61
0.74
Smarca4 (Brg1)
NM_011417
138.93
1.00
118.41
0.85
55.50
0.40
55.82
0.40
1.01
Growth factor signaling pathways
Erbb4
XM_136682
10.44
1.00
11.85
1.13
11.01
1.05
11.25
1.08
1.02
Erbb2
NM_001003817
23.79
1.00
23.52
0.99
23.23
0.98
15.28
0.64
0.66
Egfr
NM_207655
1.64
1.00
1.38
0.84
16.01
9.74
7.05
4.29
0.44
Cntf
NM_053007
11.33
1.00
11.27
1.00
8.02
0.71
6.37
0.56
0.79
Ptpn11 (Shp2)
NM_011202
85.34
1.00
83.69
0.98
76.72
0.90
61.07
0.72
0.80
Ctf1
NM_007795
8.72
1.00
8.85
1.01
9.52
1.09
7.51
0.86
0.79
Pdgfra
NM_011058
1.94
1.00
2.32
1.19
66.64
34.28
22.72
11.68
0.34
Notch signaling
Notch1
NM_008714
14.71
1.00
13.36
0.91
16.25
1.10
16.72
1.14
1.03
Dll1
NM_007865
127.35
1.00
89.71
0.70
30.86
0.24
67.47
0.53
2.19
Jag1
NM_013822
18.87
1.00
18.38
0.97
26.98
1.43
11.19
0.59
0.41
Hes1
NM_008235
15.46
1.00
26.31
1.70
19.55
1.26
15.92
1.03
0.81
Hes3
NM_008237
2.75
1.00
1.73
0.63
0.21
0.08
0.13
0.05
0.64
Hes5
NM_010419
493.11
1.00
634.06
1.29
254.74
0.52
173.74
0.35
0.68
Rbpj
NM_009035
24.71
1.00
25.64
1.04
16.31
0.66
15.64
0.63
0.96
Epigenetic modification
Dnmt1
NM_010066
73.53
1.00
70.62
0.96
20.80
0.28
17.66
0.24
0.85
Dnmt3a
NM_007872
44.79
1.00
50.32
1.12
21.96
0.49
24.67
0.55
1.12
Dnmt3b
NM_010068
35.80
1.00
51.79
1.45
1.64
0.05
2.91
0.08
1.77
Ehmt2 (G9a)
NM_145830
418.84
1.00
396.24
0.95
301.96
0.72
282.27
0.67
0.93
Suv39h1
NM_011514
18.46
1.00
20.16
1.09
7.99
0.43
8.78
0.48
1.10
Ezh2
NM_007971
83.71
1.00
81.84
0.98
14.20
0.17
18.75
0.22
1.32
Eed
NM_021876
44.49
1.00
49.21
1.11
15.30
0.34
16.45
0.37
1.08
Hdac1
NM_008228
235.16
1.00
252.53
1.07
116.10
0.49
89.06
0.38
0.77
Hdac3
NM_010411
71.40
1.00
67.23
0.94
53.75
0.75
42.89
0.60
0.80
NATURE NEUROSCIENCE VOLUME 11 [ NUMBER 9 [ SEPTEMBER 20081021
ARTICLES

Page 9

© 2008 Nature Publishing Group http://www.nature.com/natureneuroscience
step that requires the function of Coup-tfI/II for NSPCs to acquire
gliogenic competency, which occurs in parallel with or upstream of the
other processes.
Our proposal is supported by the following evidence. First, expres-
sion of COUP-TFI/II was transiently upregulated in the early neuro-
genic period in NSPCs and markedly decreased before the onset of
gliogenesis (Fig. 1g,h and Supplementary Fig. 1). Second, Coup-tfI/II
knockdown resulted in the maintenance of epigenetic silencing at the
Gfap gene (Fig. 3). Third, only limited neurogenic phenotypes were
induced by Coup-tfI/II knockdown in advanced-stage neurospheres,
which mainly generate glia (Fig. 4 and Supplementary Fig. 6). Finally,
Coup-tfI/II knockdown inhibited the neurogenesis-to-gliogenesis tran-
sition without substantially changing the expression dynamics of most
known genes associated with neurogenesis and gliogenesis (Table 1).
Our study also revealed an involvement of Coup-tfI/II in the timing of
neurogenesisinvariousbrainregions,atleast ina certain time window.
In Drosophila melanogaster CNS development, several transcription
factors required for the temporal specification of the embryonic neural
stem cells called neuroblasts have been identified42. The sequential
expression of these factors, including Hunchback, Kruppel, Pdm and
Castor, in each neuroblast elicits the sequential generation of specific
types of neurons, defined by the expression of each of these factors.
Recently, seven up (SVP), a Drosophila homolog of COUP-TFI/II, was
shown to terminate the expression of Hunchback, the earliest-
expressed transcription factor/marker, and is therefore involved in
regulating the subsequent generation of neuronal diversity43. Consis-
tent with our observations in mouse, the loss of svp increases the
number of neurons expressing early neuronal markers43. Thus, the
function of Coup-tfI/II in the timing mechanism for generating
neuronal diversity would be evolutionarily conserved, at least in part,
from invertebrates to vertebrates, although vertebrate counterparts for
theother transcriptionfactorshavenot yetbeenidentified.Withregard
to gliogenesis in Drosophila embryonic CNS, although one type of
neuroglioblast divides to simultaneously generate precursors with
restricted potential that give rise to either glial cells or neurons44,
another type of neuroglioblast sequentially gives rise to neurogenic
intermediate precursors and precursors generating both neurons and
glia45. Thus, it would be interesting to know whether Drosophila SVP
couldalsobeinvolvedinthesequentialneurogenesis/gliogenesisswitch
in this context, in a similar manner to its mammalian homolog.
It is still unclear whether the temporal identity transition for the
changeofneuropotencyandtheacquisitionofgliogeniccompetencyin
developing NSPCs is regulated by a single mechanism or two inde-
pendent ones that are used by COUP-TFI/II. The absence of linear
correlation between the extents of enhanced neurogenesis and the
production of early-born neurons in response to Coup-tfI/II knock-
down in vitro (Fig. 1c,d, and Fig. 4a,b) may support the two-mecha-
nism possibility. Identification of the target genes of Coup-tfI/IIwill be
necessary to further our understanding of this issue.
Taking our results and those of other groups into consideration, we
proposea model for the role ofCoup-tfI/IIin the sequentially regulated
temporal specification of NSPCs in mouse cortical development
(Supplementary Fig. 8 online). At the first stage of temporal specifica-
tion (E10–11), onlyneurons,including preplate Cajal-Retzius neurons,
are generated. At the second stage (E12–14), Coup-tfI and II are
transiently upregulated to move from early neurogenesis to late
neurogenesis and to confer gliogenic competency on NSPCs. Conse-
quently, NSPCs become able to differentiate into astrocytes in response
to gliogenic cytokines in vitro7, but gliogenesis does not begin with the
transient increase in Coup-tfI/II expression in vivo. At the third stage
(E15–18), NSPCs start to express markers for immature glia, such as
glutamine synthetase, but little GFAP expression is detected. During
thistime,theexpressionlevelsoftheCoup-tfI/IIinNSPCsquicklydrop
and neurogenesis tapers off. At the final stage (after birth), astrocyte
differentiation, defined by GFAP and S100b expression, accelerates.
In this study, we have uncovered a portion of the molecular
machinery involved in regulating the temporal specification of
NSPCs in CNS development. Further study of the mechanisms under-
lyingthetemporalspecificationofNSPCsshouldsuggest themeansfor
thecontrolledproductionofanydesiredtypeofneuronfromstemcells
both in vitro and in vivo, which should greatly contribute to the
development and advancement of regenerative medicine of the CNS.
METHODS
Cell culture. Mouse ESC (EB3) culture and embryoid body formation with
Noggin treatment were carried out as previously described46, except that
embryoid bodies were cultured at 105cells ml–1. Embryoid bodies were
dissociated into single cells as previously described46and were cultured in
media hormone mix (MHM) medium with 20 ng ml–1fibroblast growth factor
2 to generate neurospheres47. Neurospheres were mechanically dissociated and
passaged every 6 d. A portion of the dissociated neurosphere cells in each
generation was plated and induced to differentiate by culturing without
fibroblast growth factor 2 for 5 d. Primary neurospheres from mouse embryo-
nic CNS tissues were generated and allowed to differentiate in the same way.
Lentivirus preparation. Lentiviruses were produced by transient transfection
of 293T cells with the lentivirus constructs pCMV-VSV-G-RSV-Rev and pCAG-
HIVgp48using FuGENE 6 (Roche) according to the manufacturer’s instruc-
tions. High-titer, concentrated stocks prepared by ultracentrifugation and
resuspension in Dulbecco’s phosphate-buffered saline (2.68 mM KCl,
1.47 mM KH2PO4, 136.89 mM NaCl and 8.1 mM Na2HPO4) were used to
obtain efficient infection (multiplicity of infection of B25, in vitro).
shRNA. The sequences of shRNAs are shown in Supplementary Table 2 online.
Coup-tfI/II–specific shRNAs (knockdown/knockdown I + II) were previously
described49. Coup-tfI– and Coup-tfII–specific shRNAs were designed by the
online siRNA design program siDirect (http://design.rnai.jp/). The pSilencer
2.1–U6 Negative Control (Ambion) sequence was used as one of the negative
controls, and other negative control sequences were designed by replacing one,
two or three nucleotides of the knockdown shRNAs. The efficiency and
specificity of these sequences were evaluated by western blot analysis of stable
293T transformants expressing COUP-TFI and II or mut–COUP-TFI and II.
Animals and in utero virus injection. ICR mice were used in this study.
In utero microinjection of lentivirus was guided by gross visual examination at
E12.5 and by an in vivo ultrasound real-time scanner VS40 and Vevo660
(VisualSonics) at E10.5. All aspects of animal care and treatment were carried
out according to the guidelines of the Experimental Animal Care Committee of
the Keio University School of Medicine.
Immunostaining. Immunocytochemistry was carried out as previously
described46. For immunohistochemistry, cryostat and vibratome sections were
prepared by standard protocols after fixation with 4% (wt/vol) paraformalde-
hyde. Immunohistochemistry for BrdU was performed after pretreating the
sections in 1 N HCl at 37 1C for 30 min. The nuclei were stained with Hoechst
33258 (10 mg ml–1, Sigma B2883). For double labeling of COUP-TFI and II,
immunostaining was carried out using the Zenon Tricolor Mouse IgG2a
Labeling Kit #1 (Molecular Probes Z25160) and the VECTOR M.O.M.
Immunodetection Kit Basic (Vector Laboratories, BMK-2202).
ChIP. Our ChIP assay was performed using the ChIP-IT Express Enzymatic kit
(Active Motif) according to the manufacturer’s instructions. For quantitative
analysis, 150–200 mg of sheared chromatin was used for each experiment.
Signals were detected using FluorChem (Alpha Innotech) and quantified with
Multi Gauge software (Fujifilm). For the amplification of the STAT3-binding
site in the Gfap promoter, we used GFS1 (5¢-GGG ACT CAT TAG GAG AAC
CTC AGC AAG CAG-3¢), GFAS1 (5¢-TCT GCC CAT GCT TGG GCT TCT
1022VOLUME 11 [ NUMBER 9 [ SEPTEMBER 2008 NATURE NEUROSCIENCE
ARTICLES

Page 10

© 2008 Nature Publishing Group http://www.nature.com/natureneuroscience
GGT GTC TAC-3¢), GFS2 (5¢-GCC CCC AGG ACC TCC TTT TGT GCC-3¢)
and GFAS2 (5¢-TAT CCC AGG ATG CCA GGA TGT CAG-3¢) primers.
Statistical analysis. For each statistical analysis, at least three independent
experiments were carried out. Statistical significance was determined by two-
tailed t test.
Further details on plasmids, antibodies, western blotting and our DNA
microarray are described in the Supplementary Methods online.
Note: Supplementary information is available on the Nature Neuroscience website.
ACKNOWLEDGMENTS
We are grateful to H. Miyoshi (Riken BioResource Center) for the lentivirus
constructs, R.F. Hevner (University of Washington) for the antibody to Tbr1,
S. Mitani (Tokyo Women’s Medical University) for the antibody to GFP
(mFX73) and K. Shimamura (Kumamoto University) for the Vp16 and
Drosophila engrailed constructs. We also thank the members of the Okano
laboratory for discussion, technical advice and/or critical reading of the
manuscript. This study was supported by Core Research for Evolutional Science
Technology/Solution-Oriented Research for Science and Technology–Japan
Science and Technology Agency (H.O.), grants-in-aid for scientific research from
the Ministry of Education, Culture, Sports, Science and Technology (MEXT) in
Japan (T.S. and H.O.), a grant-in-aid from the 21st Century Center Of Excellence
program of MEXT to Keio University, a Keio University grant-in-aid for
encouragement of young medical scientists (H.N.), and a grant-in-aid for
Japan Society for the Promotion of Science Fellows (H.N.).
AUTHOR CONTRIBUTIONS
All experiments were designed by T.S. and H.N. T.S. guided the experimental
processes. Most of the experiments and data analyses were carried out by H.N.
Some parts of the in vitro culture assay and immunostaining were performed by
S.N. S.N. also assisted with all the experiments. The project was supervised by
T.S. and H.O.
Published online at http://www.nature.com/natureneuroscience/
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions/
1. Temple, S. The development of neural stem cells. Nature 414, 112–117 (2001).
2. Miller, F.D. & Gauthier, A.S. Timing is everything: making neurons versus glia in the
developing cortex. Neuron 54, 357–369 (2007).
3. Shen, Q. et al. The timing of cortical neurogenesis is encoded within lineages of
individual progenitor cells. Nat. Neurosci. 9, 743–751 (2006).
4. Koblar, S.A. et al. Neural precursor differentiation into astrocytes requires signaling
through the leukemia inhibitory factor receptor. Proc. Natl. Acad. Sci. USA 95,
3178–3181 (1998).
5. Nakashima,K.etal.Developmental requirementofgp130 signaling inneuronalsurvival
and astrocyte differentiation. J. Neurosci. 19, 5429–5434 (1999).
6. Gross, R.E. et al. Bone morphogenetic proteins promote astroglial lineage commitment
by mammalian subventricular zone progenitor cells. Neuron 17, 595–606 (1996).
7. Nakashima, K. et al. BMP2-mediated alteration in the developmental pathway of fetal
mousebrain cellsfromneurogenesistoastrocytogenesis.Proc.Natl.Acad.Sci.USA98,
5868–5873 (2001).
8. Grandbarbe, L. et al. Delta-Notch signaling controls the generation of neurons/glia from
neural stem cells in a stepwise process. Development 130, 1391–1402 (2003).
9. Nakashima, K. et al. Synergistic signaling in fetal brain by STAT3-Smad1 complex
bridged by p300. Science 284, 479–482 (1999).
10.Ge, W. et al. Notch signaling promotes astrogliogenesis via direct CSL-mediated glial
gene activation. J. Neurosci. Res. 69, 848–860 (2002).
11.Sun, Y. et al. Neurogenin promotes neurogenesis and inhibits glial differentiation by
independent mechanisms. Cell 104, 365–376 (2001).
12.Tomita, K., Moriyoshi, K., Nakanishi, S., Guillemot, F. & Kageyama, R. Mammalian
achaete-scute and atonal homologs regulate neuronal versus glial fate determination in
the central nervous system. EMBO J. 19, 5460–5472 (2000).
13.Nieto, M., Schuurmans, C., Britz, O. & Guillemot, F. Neural bHLH genes control the
neuronalversus glial fatedecisionincorticalprogenitors.Neuron29, 401–413 (2001).
14.Gauthier, A.S.et al. Controlof CNS cell-fate decisions bySHP-2 and its dysregulation in
Noonan syndrome. Neuron 54, 245–262 (2007).
15.Schmid, R.S. et al. Neuregulin 1-erbB2 signaling is required for the establishment of
radial glia and their transformation into astrocytes in cerebral cortex. Proc. Natl. Acad.
Sci. USA 100, 4251–4256 (2003).
16.Sardi,S.P.,Murtie,J.,Koirala,S.,Patten,B.A.& Corfas,G.Presenilin-dependentErbB4
nuclear signaling regulates the timing of astrogenesis in the developing brain. Cell 127,
185–197 (2006).
17.Song, M.R. & Ghosh, A. FGF2-induced chromatin remodeling regulates CNTF-mediated
gene expression and astrocyte differentiation. Nat. Neurosci. 7, 229–235 (2004).
18.Takizawa, T. et al. DNA methylation is a critical cell-intrinsic determinant of astrocyte
differentiation in the fetal brain. Dev. Cell 1, 749–758 (2001).
19.Matsumoto, S. et al. Brg1 is required for murine neural stem cell maintenance and
gliogenesis. Dev. Biol. 289, 372–383 (2006).
20.Stolt, C.C. et al. The Sox9 transcription factor determines glial fate choice in the
developing spinal cord. Genes Dev. 17, 1677–1689 (2003).
21.Deneen, B. et al. The transcription factor NFIA controls the onset of gliogenesis in the
developing spinal cord. Neuron 52, 953–968 (2006).
22.Hanashima, C., Li, S.C., Shen, L., Lai, E. & Fishell, G. Foxg1 suppresses early cortical
cell fate. Science 303, 56–59 (2004).
23.Jacob, J. et al. Transcriptional repression coordinates the temporal switch from motor to
serotonergic neurogenesis. Nat. Neurosci. 10, 1433–1439 (2007).
24.Park, J.I., Tsai, S.Y. & Tsai, M.J. Molecular mechanism of chicken ovalbumin upstream
promoter-transcription factor (COUP-TF) actions. Keio J. Med. 52, 174–181 (2003).
25.Tripodi, M., Filosa, A., Armentano, M. & Studer, M. The COUP-TF nuclear receptors
regulate cell migration in the mammalian basal forebrain. Development 131,
6119–6129 (2004).
26.Armentano, M., Filosa, A., Andolfi, G. & Studer, M. COUP-TFI is required for the
formation of commissural projections in the forebrain by regulating axonal growth.
Development 133, 4151–4162 (2006).
27.Armentano, M. et al. COUP-TFI regulates the balance of cortical patterning between
frontal/motor and sensory areas. Nat. Neurosci. 10, 1277–1286 (2007).
28.Reynolds, B.A. & Weiss, S. Generation of neurons and astrocytes from isolated cells of
the adult mammalian central nervous system. Science 255, 1707–1710 (1992).
29.Surani, M.A., Hayashi, K. & Hajkova, P. Genetic and epigenetic regulators of pluripo-
tency. Cell 128, 747–762 (2007).
30.Qiu, Y. et al. Spatiotemporal expression patterns of chicken ovalbumin upstream
promoter-transcription factors in the developing mouse central nervous system: evi-
dence for a roleinsegmental patterning ofthe diencephalon.Proc.Natl. Acad.Sci. USA
91, 4451–4455 (1994).
31.Tsai, S.Y. & Tsai, M.J. Chick ovalbumin upstream promoter-transcription factors (COUP-
TFs): coming of age. Endocr. Rev. 18, 229–240 (1997).
32.Mizutani, K., Yoon, K., Dang, L., Tokunaga, A. & Gaiano, N. Differential Notch signaling
distinguishes neural stem cells from intermediate progenitors. Nature 449, 351–355
(2007).
33.Bonni, A. et al. Regulation of gliogenesis in the central nervous system by the JAK-STAT
signaling pathway. Science 278, 477–483 (1997).
34.Stenman, J., Toresson, H. & Campbell, K. Identification of two distinct progenitor
populations in the lateral ganglionic eminence: implications for striatal and olfactory
bulb neurogenesis. J. Neurosci. 23, 167–174 (2003).
35.Wang, H.F. & Liu, F.C. Developmental restriction of the LIM homeodomain transcription
factor Islet-1 expression to cholinergic neurons in the rat striatum. Neuroscience 103,
999–1016 (2001).
36.Jessell, T.M. Neuronal specification in the spinal cord: inductive signals and transcrip-
tional codes. Nat. Rev. Genet. 1, 20–29 (2000).
37.Petryniak, M.A., Potter, G.B., Rowitch, D.H. & Rubenstein, J.L. Dlx1 and Dlx2 control
neuronal versus oligodendroglial cell fate acquisition in the developing forebrain.
Neuron 55, 417–433 (2007).
38.Olsson, M., Bjorklund, A. & Campbell, K. Early specification of striatal projection
neurons and interneuronal subtypes in the lateral and medial ganglionic eminence.
Neuroscience 84, 867–876 (1998).
39.Hevner, R.F. et al. Beyond laminar fate: toward a molecular classification of cortical
projection/pyramidal neurons. Dev. Neurosci. 25, 139–151 (2003).
40.Pereira, F.A., Tsai, M.J. & Tsai, S.Y. COUP-TF orphan nuclear receptors in development
and differentiation. Cell. Mol. Life Sci. 57, 1388–1398 (2000).
41.Bailey, P., Sartorelli, V., Hamamori, Y. & Muscat, G.E. The orphan nuclear receptor,
COUP-TF II, inhibits myogenesis by post-transcriptional regulation of MyoD function:
COUP-TF IIdirectlyinteracts with p300andmyoD.NucleicAcidsRes.26, 5501–5510
(1998).
42.Isshiki, T., Pearson, B., Holbrook, S. & Doe, C.Q. Drosophila neuroblasts sequentially
express transcription factors which specify the temporal identity of their neuronal
progeny. Cell 106, 511–521 (2001).
43.Kanai, M.I., Okabe, M. & Hiromi, Y. Seven-up controls switching of transcription factors
that specify temporal identities of Drosophila neuroblasts. Dev. Cell 8, 203–213
(2005).
44.Egger,B.,Chell,J.M.&Brand,A.H.Insightsintoneuralstemcellbiologyfromflies.Phil.
Trans. R. Soc. Lond. B 363, 39–56 (2008).
45.Udolph,G.,Rath,P.& Chia,W.ArequirementforNotchinthegenesisofasubsetofglial
cells in the Drosophila embryonic central nervous system which arise through asym-
metric divisions. Development 128, 1457–1466 (2001).
46.Okada,Y.,Shimazaki,T.,Sobue,G.& Okano,H.Retinoic-acidconcentration–dependent
acquisitionofneuralcellidentityduringinvitrodifferentiationofmouseembryonicstem
cells. Dev. Biol. 275, 124–142 (2004).
47.Shimazaki, T., Shingo, T. & Weiss, S. The ciliary neurotrophic factor/leukemia inhibitory
factor/gp130 receptor complex operates in the maintenance of mammalian forebrain
neural stem cells. J. Neurosci. 21, 7642–7653 (2001).
48.Miyoshi, H., Blomer, U., Takahashi, M., Gage, F.H. & Verma, I.M. Development of a self-
inactivating lentivirus vector. J. Virol. 72, 8150–8157 (1998).
49.Jiang, X., Norman, M., Roth, L. & Li, X. Protein-DNA array-based identification of
transcription factor activities regulated by interaction with the glucocorticoid receptor.
J. Biol. Chem. 279, 38480–38485 (2004).
NATURE NEUROSCIENCE VOLUME 11 [ NUMBER 9 [ SEPTEMBER 20081023
ARTICLES