Cell Stem Cell
Self-Organized Formation of Polarized
Cortical Tissues from ESCs and Its Active
Manipulation by Extrinsic Signals
Mototsugu Eiraku,1Kiichi Watanabe,1Mami Matsuo-Takasaki,1Masako Kawada,1Shigenobu Yonemura,2
Michiru Matsumura,1Takafumi Wataya,1Ayaka Nishiyama,1Keiko Muguruma,1and Yoshiki Sasai1,*
1Organogenesis and Neurogenesis Group
2Electron Microscope Laboratory
RIKEN Center for Developmental Biology, Kobe 650-0047, Japan
Here, we demonstrate self-organized formation of
apico-basally polarized cortical tissues from ESCs
using an efficient three-dimensional aggregation cul-
ture (SFEBq culture). The generated cortical neurons
are functional, transplantable, and capable of form-
ing proper long-range connections in vivo and
in vitro. The regional identity of the generated pallial
tissues can be selectively controlled (into olfactory
bulb, rostral and caudal cortices, hem, and choroid
plexus) by secreted patterning factors such as Fgf,
Wnt, and BMP. In addition, the in vivo-mimicking
birth order of distinct cortical neurons permits the
selective generation of particular layer-specific neu-
rons by timed induction of cell-cycle exit. Impor-
tantly, cortical tissues generated from mouse and
human ESCs form a self-organized structure that in-
cludes four distinct zones (ventricular, early and late
apico-basal direction. Thus, spatial and temporal
aspects of early corticogenesis are recapitulated
and can be manipulated in this ESC culture.
During embryogenesis, the telencephalic anlage is subdivided
cortical hem and choroid plexus, and the Pax6?ventral region
(the subpallium), which forms the basal ganglia (Guillemot,
2005; Puelles et al., 2000; Rallu et al., 2002).
We recently established an efficient culture method for the
selective neural differentiation of mouse ESCs (mESCs) using
serum-free suspension culture (serum-free culture of embryoid
body-likeaggregates [SFEB]; Watanabeetal.,2005,2007).Neu-
ral differentiation in the SFEB culture occurs tissue-autono-
mously and does not require extrinsic inducers (Watanabe
et al., 2005). Importantly, SFEB-cultured mESCs differentiate
into Bf1 (FoxG1)+telencephalic progenitors (Tao and Lai, 1992;
Hanashima et al., 2004) at a moderately high frequency (Wata-
nabe et al., 2005). The majority of the SFEB-induced telence-
phalic progenitors express the pallial marker Pax6, while Shh
treatment suppresses Pax6 and induces the subpallial markers
such as Nkx2.1 (Watanabe et al., 2005). A similar observation
of telencephalic differentiation has been reported for SFEB-
cultured human ESCs (hESCs) (Watanabe et al., 2007).
In this report, first, we introduce an improved three-dimen-
of ESCs into cortical progenitors and functional projection neu-
rons. Then, using this ESC culture, we demonstrate the in vitro
recapitulation of embryonic corticogenesis and its manipulation
with respect to layer-specific neurogenesis and regional specifi-
cation. Finally, we demonstrate that cortical progenitors gener-
ated in this culture by mouse and human ESCs spontaneously
form patterned structures, mimicking the early aspect of cortico-
genesis. We discuss the remarkable ability of ESC-derived
cortical neuroepithelia with regard to self-organized tissue
Efficient Differentiation of Mouse ESCs into Polarized
Cortical-Type Neuroepithelia in SFEBq Culture
of dissociated mouse ESCs that undergo spontaneous reaggre-
gation in 1 to 2 days. After 7 days of suspension culture, the
aggregates are then subjected to adhesion culture (Watanabe
et al., 2005). In this original method, the ESC aggregates form
in varying sizes and the Bf1 induction efficiency is below 50%
(typically, 30%–35% of total cells; Watanabe et al., 2005).
In the present study, we sought to design conditions in which
the aggregate formation would be more tightly and quantitatively
controlled. Dissociated mESCs (3000 cells) were cultured in
each well of a low cell-adhesion 96-well plate (U-bottomed). In
this procedure, the cells reaggregated quickly (within a few
hours), formed uniformly sized cell masses (Figure 1A and
Figure S1A available online) and selectively differentiated into
neural cells (shown below). Importantly, under these conditions,
(up to 65%–75% of total cells on days 10 and 12; Figure 1A and
also shown below) than with the original method. Moreover, the
majority of the Bf1+cells (89%; day 10) generated in this modi-
fied SFEB culture coexpressed the bona fide cortical marker
Emx1 (Figure 1A; in contrast, 36%–40% of Bf1+cells express
Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc. 519
Figure 1. Generation of Polarized Cortical Neuroepithelia from SFEBq-Cultured mESCs
(A) Schematic of the SFEBq culture for mESCs. Uniformly sized aggregates on day 7 and high percentages of Bf1+/Emx1+cells on day 10.
(B–I) Cryosections of SFEBq-cultured mESC aggregates on day 3 (B), day 4 (C), day 5 (D–F), day 8 (G), and day 10 (H and I). Immunostaining for E-cadherin
(B and C), Sox1 (B), N-cadherin (D–G), CD133 (E), Laminin (F), Bf1 (H and I), and Emx1 (I).
(J–M) Electron microscopic analysis. Arrow, M-phase cell at the apical end (J) and an apical cilia (M). Bracket, tight junctions (K) and adherence junction (L).
Cell Stem Cell
Corticogenesis in ESC Culture
520 Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc.
Emx1 at most in the original SFEBq culture; Watanabe et al.,
2005 and data not shown). Thus, this modified quick-aggrega-
tion procedure (SFEBq, hereafter) is much more effective for
inducing the differentiation of cortical cells than is the original
SFEB culture, which relies on slow reaggregation in Petri dishes.
In addition to controlling the size of the aggregates, early res-
toration of the cell-cell interaction by quick aggregation might
also improve the selectivity and reproducibility of differentiation
in this new culture procedure. In fact, one of the remarkable fea-
tures of the SFEBq culture was that the ESC-derived neural pro-
genitors reproducibly (>95%; <30% in the original SFEB culture)
formed polarized neuroepithelial structures within the aggre-
gates during the first 5 days (Figures 1B–1F). On day 3, Sox1
(earliest neuroectodermal marker) expression was first detected
in the superficial zone of the SFEBq aggregate (still E-cadherin+;
Figure 1B), which had not yet underwent evident neuroepithelial
formation. N-cadherin+neural progenitors started to accumulate
in the Sox1+outer regions (gradually expanding inwards)
the N-cadherin+tissues exhibited partial neuroepithelial struc-
tures (Figure 1C; see Figure S1C for a high-magnification view;
the arrows indicate N-cadherin-dense apical portions while the
dotted lines show the basal side). By day 5, continuous, polar-
ized N-cadherin+neuroepithelial structures (columnar epithelia)
had formed from mESC aggregates (Figure 1D; the cavitation
in the SFEBq aggregate was partly due to apoptosis in a minority
of cells near the center; data not shown). The aggregates repro-
ducibly formed an apical surface inside, which was indicated by
denseN-cadherin accumulation; the inner area was also positive
for the apical marker CD133/Prominin1 (Weigmann et al., 1997),
aPKC, and ZO-1 (Figure 1E and Figure S1D), while a basal
surface (Laminin+; Figure S1E for in vivo expression) was located
outside (Figure 1F).
When an SFEBq aggregate was cultured in suspension
beyond day 6, the continuous neuroepithelial sheet (Figure 1D)
gradually reformed into several round clusters (rosettes) around
aggregate peaked around day 10 (Figures 1H and 1I and Figures
S1F–S1J). Electron microscopic analysis (Figures 1J–1M)
showed the presence of tight (Figure 1K) and adherence (Fig-
ure 1L) junctions as well as apical cilia structures (Figure 1M;
9+0 tubulin arrangement in Figure S1K) on the lumenal side.
Consistent with this epithelial polarity, the apical-end markers
(gamma-tubulin+, CD133+) were localized at the center of each
rosette facing to the lumen (Figures 1N and 1O). Phosphorylated
histone H3 (pH3), a marker for the G2-M phases, was detected
in Emx1+nuclei close to the center (Figure 1P); this distribution
was reminiscent of the in vivo position of pH3+cortical neuroepi-
thelial nuclei in the apical zone (Figure S1L). Individual neuroepi-
thelial cells in the ESC-derived rosettes, visualized by mixing
Sox1::GFP-knockin ESCs with nonlabeled ESCs in the reaggre-
gation culture on day 0, showed a typical morphology of
the neural progenitors in the early embryonic neuroepithelium
and expressed aPKC, an apical marker, on the luminal side
Thus, the SFEBq culture efficiently and reproducibly induces
the formation of Bf1+/Emx1+neuroepithelia with an evident
polarity similar to the embryonic neuroepithelial structure.
SFEBq-Induced Neuroepithelia Generate Cortical-Type
In dissociation culture, the SFEBq-induced neural progenitors
(dissociated on day 12) differentiated into Tuj1+postmitotic
neurons expressing characteristic cortical markers such as
Emx1(82%)andVGLUT1 (83%),aswellasTelencephalin (Mitsui
et al., 2007), CamKIIa (Kinney et al., 2006), and GluR1 (Figure 2A
and Figures S2A–S2D). They also expressed subtype-specific
markers such as Tbr1, Ctip2, and Brn2 (Figures S2E–S2H and
data not shown).
We then performed an organotypic coculture with embryonic
forebrain tissues using ESCs with Venus-GFP (Nagai et al.,
2002) knocked in at the Bf1 gene locus (see Figures S2I–S2N
and Supplemental Experimental Procedures). The ESC-derived
Bf1::Venus+cell masses (day 14 of culture) were placed within
and these conjugates were then subjected to culture on a porous
filter for an additional 3 days. A large number of Bf1::Venus+cells
28). Incontrast,theBf1::Venus+cellsrarely invadedthe subpallial
or diencephalic region (Figures 2D and 2E; Figure 2F shows the
polarized arrangement of cells migrating toward the pia). Emx1,
Tbr1, Ctip2, and Brn2 were expressed in >96%, 3%, 65%, and
21% of Venus+neurons in the intermediate and mantle zones,
respectively (most Tbr1+neurons remained within the SFEBq ag-
gregate; data not shown). Efficient integration of the Bf1::Venus+
neurons was also observed when the day 14 aggregates were
cocultured with postnatal (P1) forebrain slices (70% of the conju-
gates, n = 20; Figures 2G–2I; arrowheads in Figure 2I indicate ex-
tending axons; the distribution of integrated neurons expressing
subtype-specific markers is shown in Figures S2O–S2R).
brain. Bf1::Venus+cells were dissociated and injected into the
neonatal cortex (P2; see the Supplemental Experimental Proce-
ron-like cells with a well-developed apical dendrite and several
basal dendrites were observed (Figures 2J–2L; 71% of Venus+
cells). When Bf1::Venus+cell aggregates were grafted en bloc
Venus+axons were observed in the deep cortex zone, corpus
callosum, striatum (observed as fascicles), thalamus, cerebral
peduncle (pyramidal tract), and pontine nuclear regions, consis-
tent with the corticofugal axons projecting from the embryonic
cortical neurons (Figures 2P–2U). In addition, an overlay study
on P1 slices showed that the Venus+axons from the SFEBq ag-
gregates actively turned their directions toward the subcerebral
regions and away from the cortex (Figures S2T–S2W).
These findings supported the idea that the SFEBq-induced
Bf1+progenitors can function as bona fide progenitors for corti-
cal projection neurons.
(N–Q) Immunostaining of rosettes in the SFEBq aggregates for Bf1 (N and O), gamma-tubulin (N), CD133 (O and P), phosphorylated H3 (P), and aPKC ([Q];
asterisk, lumen). In (Q), Sox1::GFP ESCs were mixed with nonlabeled ESCs upon reaggregation on day 0.
Scale bars: 100 mm in (B)–(D), (G), and (H); 50 mm in (E) and (F); 10 mm in (J); 0.2 mm in (K) and (L); 0.5 mm in (M); and 20 mm in (N)–(Q).
Cell Stem Cell
Corticogenesis in ESC Culture
Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc. 521
SFEBq-Induced Neurons, such as Neonatal Cortical
Neurons, Show Spontaneous Ca2+Oscillations
that Expand over Long Distances
We next performed Ca2+imaging studies (Ikegaya et al., 2005)
using long-term cultured SFEBq-treated mESC aggregates
(Bf1::Venus) (Figure 3A). On day 21, individual Venus+neurons
showed patterned spontaneous activity that differed from
neuron to neuron in frequency, regularity, and duration (Figures
3B and 3C; see Movie S1 for time-lapse image). Regardless
of the patterns, this spontaneous neuronal activity (Ca2+surges)
increased in frequency following glutamate application (Fig-
ure 3D) and was inhibited by TTX (tetrodotoxin; Figures 3E
and 3F), indicating that it was dependent on a local synaptic
Previous studies have reported locally coupled spontaneous
neuronal activity that is unique to the neonatal cortex during
the first postnatal week: a large-scale, very fast oscillatory
Ca2+wave that synchronously activates neurons over a long dis-
tance (one to several millimeters; Garaschuk et al., 2000; Adels-
berger et al., 2005). To our surprise, such locally-coupled Ca2+
transients were also reproducibly observed in the ESC-derived
Bf1::Venus+tissues. The fast-wave Ca2+oscillation was seen al-
most simultaneously over a large area (up to about 1 mm in the
long axis; >1.3 mm/sec) in the aggregate (Figures 3G–3I; Figures
S3A and S3B for 60 point simultaneous recording; Movies S2
and S3 for time-lapse images of this and another view field). As
seen in the neonatal cortex (Garaschuk et al., 2000; Adelsberger
et al., 2005), this oscillation was blocked by CNQX (a glutamate
antagonist; Figure 3J) or TTX, but not by bicuculline (a GABA an-
tagonist; data not shown). The large-scale, fast-wave Ca2+oscil-
lations were observed more frequently on day 24 in culture than
on day 21 (p < 0.01, Student’s t test; Figure 3K), probably reflect-
ing the maturation of neuronal networks over a large area in the
aggregate). A similar long-range (>1 mm) coupling of neuronal
activity was observed by MED64 multigrid electrode analysis
(field potential; data not shown). In addition to these long-range
oscillations, short-range, slow-wave Ca2+oscillations (Yuste
et al., 1995) were also frequently observed (Figure S3C and
These findings show that the SFEBq-induced cortical tissues,
at least in part, mimic neuronal activity features characteristic of
neonatal cortical tissues.
Figure 2. Intergration of SFEBq-Induced Cortical Neurons with
(A) An example of Ctip2+stellate-shaped neurons receiving numerous Synap-
tophysin+presynaptic inputs on the well-arborized CamKII+dendrites in long-
term dissociation culture, day 42.
(B) SFEBq-cultured mESC aggregates (Bf1::Venus) on day 12.
(C) Schematic of the coculture study using forebrain slices (E14.5).
(D and E) Invasion of Venus+neurons into the cortical region of the E14.5 slice.
(F) A high magnification view of migrating Venus+neurons.
(G–I) Invasion of Venus+neurons into the cortical region of the P1 slice. Immu-
nostaining for Venus (anti-GFP, green) and L1 (red). (I) High magnification
confocal image. Arrows, extending axons.
(J–L) An example of typical pyramidal neurons generated from the SFEBq-
derived cortical progenitors (day 14) dissociated and injected into the P2
munostaining for Bf1::Venus-GFP and Tbr1. (L) Immunostaining for Bf1::
Venus-GFP and synaptophysin.
(M–U) Bf1::Venus+cell aggregates were grafted en bloc into the frontal cortex.
(M) Schematic of coronal sections. (N and O) The graft in the frontal cortex.
(P–U) Distributions of the graft-derived Venus+axons in the boundary between
the cortex (Cx) and white matter (P), the corpus callosum (Q), the striatum (R),
the thalamus (S), the cerebral peduncle (T), and the pontine nucleus (U).
Scale bars: 50 mm in (A), (F), and (I); 500 mm in (B), (D), (E), and (M); 1 mm in (G);
60 mm in (J); 20 mm in (K) and (L); 300 mm in (O)–(R); and 100 mm in (S)–(U).
Cell Stem Cell
Corticogenesis in ESC Culture
522 Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc.
Manipulation of the Regional Identity of ESC-Derived
Cortical and Noncortical Pallial Tissues
These results led us to ascertain that the SFEBq-induced pro-
genitors had the ability to produce cortical neurons. We then
sought to modify the regional and temporal identities of the in-
duced cortical tissues by manipulating the extrinsic conditions
(Figures 4 and 5).
The pallium is patterned in a region-specific fashion during
embryogenesis (O’Leary et al., 2007). The rostral-most part of
the cortex is the olfactory bulb, which is an extension of the me-
dial pallium. The cortical hem and choroid plexus arise from the
caudal-dorsal end of the pallium (Monuki and Walsh, 2001). The
neocortex is located between these areas and itself is patterned
along the rostral-caudal axis by the rostral-low, caudal-high gra-
dient of the transcription factor COUP-TF1 (Figure 4A and
Figure S4A, left; O’Leary et al., 2007; Zhou et al., 2001). Immu-
nostaining revealed that 47% of the SFEBq-induced Bf1:Venus+
cells (sorted by FACS on day 7 and subjected to further reaggre-
gation culture for 3 days; Figure 4B) were strongly positive for
tical cells represent heterogeneous regional identities along the
Fgf8 has been shown to function as an inducer of rostral cor-
tical regions (Hebert et al., 2003b; Shimogori and Grove, 2005).
Interestingly, the treatment of Bf1::Venus+cell aggregates with
Fgf8 (day 7) strongly suppressed COUP-TF1 expression (12%
of total cells; Figures 4D and 4F and Figure S4B). In contrast,
COUP-TF1+cells were substantially increased in number by
treating the aggregates with the Fgf inhibitor FGFR3-Fc (81%
of of total cells; Figures 4E and 4F). These COUP-TF1+cells
coexpressed the cortical marker Emx1 (Figure S4C), indicating
that the FGF signal attenuation induced a caudal cortical fate.
Treatment with Fgf8 or FGFR3-Fc largely did not affect the
Bf1::Venus expression in the aggregates on day 10 (Figures
These findings showed that Fgf signaling induces rostraliza-
tion of early cortical tissues in vitro, as it does in vivo. Consistent
teristic of the projection neurons in the olfactory bulb (Yoshihara
et al., 2005), a derivative of the rostral-most cortex (Figures 4J
and 4K), was induced by Fgf8 (100% of Fgf8-treated aggregates
and 0% of untreated controls; Figures 4L–4N; day 21; see the
The in vivo development of the cortical hem and choroid
plexus requires Wnt and BMP signals (Lee et al., 2000; Hebert
et al., 2003a). During midcorticogenesis, pallial Otx2 expression
is found exclusively in the proximity of the hem (TTR?) and in the
choroid plexus (TTR+) (Figure 4O). Treatment of Bf1::Venus+cell
aggregates with Wnt3a promoted the generation of Otx2+/TTR?
hem-type cells (Figures 4P and 4Q). These Otx2+cells coex-
pressed another hem/choroid-plexus marker, Lmx1a (Mangale
et al., 2008) (Figure 4R and Figure S4D), while Lmx1a was not ef-
ficiently induced by FGFR3-Fc (Figure S4E). In contrast, BMP4
treatment induced Otx2+/TTR+choroid-plexus-like epithelial
cells in addition to Otx2+/TTR?cells, and this induction was fur-
ther enhanced when Wnt3a and BMP4 treatments were com-
bined (Figures 4S–4U and Figure S4F).
Collectively, these findings demonstrate that SFEBq culture
enables considerable in vitro control of regional pallial (cortical
and noncortical) tissue induction from ES-derived progenitors
by the manipulation of these embryologically relevant patterning
signals (Figure S4A).
Temporal Control of the Generation of Layer-Specific
Cortical Neurons and their Enrichment by In Vitro
The neocortex consists of six distinct layers, and the embryonic
cortical neuroepithelia generate layer-specific neurons by the
progressive commitments of cortical stem cell populations
(Hevner et al., 2003a; Shen et al., 2006). During corticogenesis,
the neurons of layers II–VI are born sequentially in an ‘‘inside
early, outside late’’ order (inside-out pattern), with the exception
of the Reelin+/Bf1?Cajal-Retzius cells in layer I (Soriano and Del
Rio, 2005; Stoykova et al., 2003), which are born the earliest (at
around E10 in mice). Cajal-Retzius cells are derived from the pe-
ripheral regions of the pallium such as the Bf1?hem or Bf1+pal-
lial-subpallial boundary regions (the hem anlage is also initially
Bf1+) and migrate into the superficial-most cortical zone (Bielle
et al., 2005; Frantz and McConnell, 1996; Hevner et al., 2003b;
Hanashima et al., 2004; Molyneaux et al., 2007). Then, precur-
sors of the layer VI neurons (later demarcated by Tbr1+/Bf1+)
and layer V neurons (Ctip2+/Emx1+) are sequentially born and
form the early cortical plate (CP), which later becomes the lower
CP (Figure S5A; Hevner et al., 2001; Arlotta et al., 2005). During
late corticogenesis, the late/upper CP neurons and their progen-
itors, occupying the VZ/SVZ during mid-late gestation, express
Brn2 (Figure S5B; Hevner et al., 2003a; Frantz and McConnell,
1996; Molyneaux et al., 2007).
As illustrated in Figure 5A, the sequence of marker expression
onset in the SFEBq culture generally paralleled that seen during
embryogenesis. Reelin+and Tbr1+/Reelin?/Bf1+neurons ap-
peared on days 7 and 8, while substantial numbers of Ctip2+/
neurons were seen only on and after day 10
(Figure 5B). Brn2+/Bf1+cells showed substantially increased
numbers during days 10–13 (Figure 5C; initially mitotic), and
most became postmitotic (TuJ1+) by day 15 (Figure 5D) These
Brn2+neurons also expressed other upper CP markers such
as Cux1 and Satb2 (Figures S5C and S5D and data not shown).
et al., 1997), the SFEBq-induced neural progenitors were incu-
bated overnight with BrdU during days 8–14 in culture, and the
BrdU-labeled cells were analyzed with neuron-type-specific
markers on day 16 (Figures 5E–5I). We found that the birth (the
timing of the last S phase) of Reelin+(layer I), Tbr1+/Bf1+(layer
VI), Ctip2+/Emx1+(layer V), and Brn2+/Tuj1+(layer II/III) neurons
(Figure 5I), showing that the layer-specific neurons from mESC-
derived cortical progenitors were preferentially generated in the
same temporal order as seen in the embryonic mouse cortex
(Gonzalez et al., 1997). These observations are in accordance
with the intrinsic temporal program of cortical neurogenesis
shown with cortical progenitors derived from the embryonic
brain (Shen et al., 2006) and, very recently, with those from
ESCs (Gaspard et al., 2008).
Based on these findings, we next asked whether very early-
born Reelin+neurons and early CP neurons (Ctip2+), respec-
tively, could be preferentially generated by the timed control of
neuronal induction in SFEBq culture. Bf1::Venus ESCs were
Cell Stem Cell
Corticogenesis in ESC Culture
Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc. 523
Cell Stem Cell
Corticogenesis in ESC Culture
524 Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc.
cultured by SFEBq for different times, at which point the Venus+
cells were sorted by FACS and quickly reaggregated, followed
by suspension culture in the presence of the Notch inhibitor
DAPT, which promotes neuronal differentiation from mitotic pro-
genitors (Nelson et al., 2007; Figure 5J and Figures S1P and
were sorted on day 9 and treated with DAPI (Figures 5K and 5N).
In contrast, the majority of cells became Ctip2+(strongly positive
in 66% of the cells) when the Bf1::Venus+cells were induced to
differentiate into neurons on day 12 and cultured for another
7 days (Figures 5L and 5O; cortical neuronal differentiation oc-
curred slightly more slowly in the Bf1::Venus ESCs and Ctip2 ex-
pression appeared about 0.5 to 1 day later than in the wild-type
ESCs that were used in Figure 5I; data not shown). The Ctip2+
cells coexpressed Emx1 (Figure 5M) and GAD?(Figure S5E),
indicating that these cells represented early CP neurons. In
after this late-step induction (5% without DAPT treatment;
Figure S5F). Only a marginal increase in Brn2+neurons was ob-
served by neuronal induction on day 12 as compared to the
Brn2+proportion after the day 9 induction (Figure 5P).
Thus, SFEBq culture not only recapitulates the temporal as-
pects of cortical neurogenesis, but also is applicable to selective
generation of layer-specific neurons in vitro (Figure S5H).
Generation of Cajal-Retzius Cells and Cortical-Plate
Neurons from Polarized Cortical Neuroepithelial
Tissues in SFEBq Culture
We next examined how such cell-type-specific differentiation
occurred in the structural context of the SFEBq aggregate.
On day 7, about a half of the Reelin+cells were found in Bf1?
cell masses located adjacent to, and often continuous with, Bf1+
rosettes (Figure 6A), while the Bf1+rosettes also contained
Reelin+cells, most of which were weakly positive (Figure 6A
and Figure S6A). On day 8, the majority of Reelin+cells were
clearly Bf1?, only loosely clustered and located mainly outside
of the Bf1+rosettes (Figure 6B and Figures S6B and S6C). On
day 10, Reelin+/Tbr1+neurons, which represented ?15% of the
total cells and coexpressed another Cajal-Retzius cell marker
Calretinin (Figures S6E–S6H; Englund et al., 2005; Hevner
et al., 2003b; Soriano and Del Rio, 2005; Stoykova et al.,
2003), were found in the nonrosette regions and the thin super-
ficial-most zones of the Emx1+cortical rosettes (Figure 6C).
On day 10, each SFEBq-cultured ESC aggregate contained
several Emx1+rosettesthat had Pax6+areas in the central zones
(Figure 6D). These inner portions (Bf1+/Pax6+) were occupied
by mitotic (Ki67+) neuroepithelial cells (Figure S6I). In the Bf1+
that were Emx1+, Tbr1+, Pax6?, and Reelin?(Figures 6C and 6D
and data not shown); this marker expression profile is character-
istic of the early/lower CP neurons during early in vivo cortico-
genesis (Figure S6K; Englund et al., 2005; Hevner et al., 2001).
Tbr2 is a marker for the progenitors and precursors of late/up-
per CP neurons (Englund et al., 2005; Hevner et al., 2001). Prior
to the formation of the late/upper CP in the embryonic cortex,
Tbr2+late CP precursors are mostly mitotic (Ki67+and Pax6+)
and occupy the cortical subventricular zone (Figure S6J; En-
glund et al., 2005). In the SFEBq culture, Tbr2 appeared later
(days 9 and 10) than the early CP marker Tbr1 (days 7 and 8;
Figure 5A). On day 10, in contrast to the postmitotic Tbr1+cells
surrounding the rosette, most of the Tbr2+cells were found par-
ticularly in theouter portion ofthe Pax6+zone(oradjacent partof
the Tbr1+regions; Figures 6E). This distribution pattern of Tbr2+
cells inthe Pax6+zoneisreminiscent of thein vivo situation (sub-
ventricular location; Figure 6F and Figures S6J and S6K).
After day 12 in SFEBq culture, polarized rosette structures
gradually began to disintegrate as the neuroepithelial progeni-
tors (i.e., the radial glia of cortical tissues) decreased in number.
Most of the Pax6+neuroepithelial progenitors disappeared and
underwent neuronal differentiation by day 16 (Figure 5D and
Figure S1O). Unlike in the embryonic neocortex, the ‘‘inside-
out pattern’’ of layer-specific neurons was not observed, even
at this late phase of SFEBq culture. For instance, double staining
for the early CP (Ctip2) and late CP (Satb2, Brn2) markers on day
16 showed that the early and late CP markers were mostly found
in the outer and inner portions of the cortical cell clusters, re-
spectively, suggesting that the layer inversion (early, deep; late,
superficial) did not occur (Figures 6G–6I, and data not shown);
the situation was similar on day 19 (Figures S6L–S6O).
Taken together, these findings indicate that SFEBq-induced
cortical rosettes have the apico-basal arrangement of four dis-
tinct zones (Pax6+, Tbr2+, Tbr1+, and Reelin+), mimicking struc-
tural aspects ofthe earlycorticaltissue prior tothe upper/late CP
Application of SFEBq Culture to hESC Differentiation
Lastly, we tested whether the SFEBq culture procedure is appli-
cable to the formation of polarized cortical neuroepithelial
tissues in human embryonic stem cell (hESC) culture. In this
case, ROCK inhibitor treatment (see Experimental Procedures
and Figure S7A) was essential for circumventing the hESC-spe-
cific apoptosis that is induced by cellular dissociation and sus-
pension (Watanabe et al., 2007). Like the mESCs, the SFEBq-
cultured hESCs formed N-cadherin+/BLBP+neuroepithelia that
Figure 3. Spontaneous Ca2+Oscillations in mESC-Derived Cortical Tissues
(A) Schematic of Ca2+imaging. The Venus signals were much weaker than Fluo4 signals and, therefore, negligible in this image.
(B and C) Analysis of Ca2+surges in individual neurons on day 21. Numbers correspond to individual neurons analyzed simultaneously by Ca2+imaging. (B) Fluo4
fluorescence image at the peak signal point.
(D–F) Effects of glutamate (100 mM) (D) and TTX (1 mM) (E and F) on Ca2+surges in individual neurons. ACSF, artificial CSF.
(G–I) Long-scale synchronous Ca2+oscillations on day 24. (G) Pseudocolor images after baseline signal subtraction. (H and I) Neurons A–E correspond to
individual cells analyzed simultaneously by Ca2+imaging.
(J) Effects of CNQX (10 mM) on large-scale Ca2+coupling.
(Kand L)Comparison of percentages of aggregates showing long-range Ca2+oscillations between days21and 24(K). **p < 0.01, t test. In contrast, no significant
difference was seen for the appearance of single Ca2+transients in the aggregates (L).
Scale bars: 100 mm in (B) and 200 mm in (G) and (H). The bars in the graphs represent standard errors.
Cell Stem Cell
Corticogenesis in ESC Culture
Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc. 525
Figure 4. Active Manipulation of the Regional Specification in the SFEBq-Induced Cortical Progenitors by the Extrinsic Conditions
(A) Immunostaining of COUP-TF1 in the E12.5 mouse cortex (parasagittal section; rostral, left).
(B) Schematic of the experiments of the patterning factor treatments.
Cell Stem Cell
Corticogenesis in ESC Culture
526 Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc.
were highly polarized (>95% aggregates; Figures 6J–6L and
Figure S7B) and expressed Bf1, Emx1, and Pax6 (Figures 6M
and 6N). Interestingly, unlike the mESC-derived tissues, the
hESC-derived Bf1+/Emx1+/Pax6+neuroepithelium did not re-
form into many small rosettes but instead formed one or a few
relatively large torus-like or mushroom-shaped structures even
after 46 days in differentiation culture (Figures 6J–6N; found
in >90% of the aggregates).
On day 46, Tbr1+neurons were mainly found in the TuJ1+
zones surrounding the Bf1+/Pax6+neuroepithelia (Figures 6N–
6P). In the hESC-derived cortical neuroepithelial domain, the
superficial-most layer of the Tbr1+zone contained Reelin+/
Tbr1+/Bf1–Cajal-Retzius-like cells (Figure 6O and Figure S7C).
The Bf1+/TuJ1+postmitotic neurons in the outer zone also ex-
pressed another early CP marker Ctip2 (Figures 6Q). In contrast,
the late CP marker Tbr2 was mostly seen within the Pax6+ven-
to the Tbr1+regions (Figures 6R). Thus, the early spatial expres-
sion patterns of these markers were consistent between the
mESC- and hESC-derived cortical tissues (Figure S7D).
Later around day 60, as with the mESC-derived tissues, the
continuous cortical neuroepithelia broke into several relatively
large rosettes (Figure S7F and data not shown; day 59), which
itors as seen on day 46 (Figures S7F–S7I). On day 59, however,
the Pax6+progenitor zone became much thinner (typically 5–7
nucleus thickness; Figure S7G) than that on day 46 (11–14 nu-
Region- and Timing-Specific Control of the Generation
of Cortical/Pallial Tissues from ESC-Derived
In this report, we have described a highly reproducible three-
dimensional culture of ESCs following their quick dissociation
and reaggregation and shown the efficient formation in this cul-
ture of polarized Bf1+/Emx1+neuroepithelia that generate multi-
ple kinds of cortical-type neurons in a temporally and spatially
coordinated fashion. As shown in this and other studies (Wata-
nabe et al., 2005, 2007; Gaspard et al., 2008), the telencephalic
differentiation of ESCs is largely dependent on cell-intrinsic
mechanisms and weak/basal endogenous extracellular signals,
but not on strong inductive signals. Probably because of this
feature, the telencephalic differentiation of ESCs appears to be
quite sensitive to relatively minor changes in the initial culture
conditions such as the cell density, aggregate formation, cell
adhesion, and basal culture medium (Watanabe et al., 2005;
Wataya et al., 2008).
Our SFEBq approach permits a particularly faithful recapitula-
tion of early corticogenesis. Furthermore, several aspects of
nal manipulation in this culture. By modifying extrinsic signals,
SFEB-induced Bf1::Venus+cells can be specified into a wide
In the absence of exogenous Fgf8 or its antagonist, about a half
of the Bf1::Venus+cells express COUP-TF1, suggesting that
SFEBq culture per se induces a mixed cortical progenitor popu-
lation with respect to the rostral-caudal specification. This may
be due to the low but significant level of endogenous Fgf signals,
since treatment with FGFR3-Fc can efficiently induce COUP-
TF1+caudal cells. The present study has also demonstrated
that Fgf8 is a ‘‘sufficient’’ factor for the olfactory bulb induction
directly from Bf1+progenitors.
Importantly, SFEBq-induced Bf1::Venus+progenitors are
cortical; hem and choroid plexus) by Wnt and BMP signals.
These findings demonstrate remarkable direct changeability of
the fate of Bf1+cortical progenitors into noncortical pallial tis-
suesbysimple signaling manipulation during earlydifferentiation
stages in vitro.
Taking advantage of the sequential commitment of layer-spe-
cific neurons in the SFEBq culture, we have also demonstrated
the preferential generation of the very early neuronal lineage
(Reelin+) versus the early CP lineage (Ctip2+/Emx1+) (Figure 5).
Forced cell-cycle exit at an early time point facilitates the gener-
ation of Reelin+Cajal-Retzius-type neurons, which areborn early
in vivo. In this case, forced neuronal differentiation by DAPT is
essential for efficient Reelin induction from FACS-sorted
Bf1::Venus+cells (Figure S5F, lanes 1 and 2). Conversely, even
without FACS sorting, DAPT treatment alone (day 8) could
efficiently generate Reelin+cells (up to two-thirds of total cells;
Figures S1R and S1S).
In contrast, neuronal induction at a later time point results in
the preferential generation of Ctip2+/Emx1+neurons, whose
birth date in vivo is later than that of Cajal-Retzius cells. At this
late point, DAPT treatment does not substantially induce Reelin+
cells any more (Figure S5E, lanes 3 and 4), probably because the
competence of ESC-derived cortical progenitors to generate
Reelin+cells has already been reduced by then.
Self-Organized Formation of Polarized Cortical
One of the important features of the SFEBq culture is the repro-
ducible formation of the well-polarized cortical-type neuroepi-
thelia (Figures 1 and 6). Since dissociated ESCs have no prede-
termined positional information at the beginning of culture,
certain self-organizing mechanisms should be involved in the
(C–F) The effects of FGF signaling on COUP-TF1 expression. (C) untreated, (D) Fgf8b treatment (50 ng/ml, days 7–10), and (E) FGFR-Fc treatment (500 ng/ml,
days 7–10). (F) Quantification of the COUP-TF1+cell percentages.
(G–I) The effects of FGF signaling on the expression of Bf1::Venus.
(J–N) Induction of olfactory bulb neurons in the SFEBq-induced cortical tissues. (J and K) Tbx21 expression in the P1 mouse olfactory bulb (OB, parasagittal
section). Cx, cortex. (K) Triple-staining of Tbx21, Tbr1, and Reelin in a high magnification view. (L–N) Induction of Tbx21+neurons by Fgf8.
(O–U) Induction of hem and choroid-plexus differentiation. (O) Expression of Otx2 and TTR in the mouse hem and choroid-plexus. E12.5, parasagittal section.
(P–R) Induction of Otx2 (Q and R) and Lmx1a (R) by Wnt3a treatment (20 ng/ml, days 9–12). (S and T) Induction of TTR in the SFEBq-generated neural progeniors
by BMP4 ([S] and [T]; 0.5 nM, days 9–12) and Wnt3a (T). (U) Quantification of the Otx2+and TTR+percentages.
Scale bars: 400 mm in (A); 100 mm in (C)–(E), (G), (O), (P), and (Q); 50 mm in (K) and (R); 200 mm in (L)–(N); and 80 mm in (S) and (T). The bars in the graphs represent
Cell Stem Cell
Corticogenesis in ESC Culture
Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc. 527
Figure 5. Layer-Specific Cortical Neurons Are Generated in the Temporally Controlled Manner that Mimics Corticogenesis In Vivo
(A) The expression onset of distinct cortical neuronal markers.
(B–D) Immunostaining of Citp2 and Emx1 on day 10 (B), Brn2 and Ki67 on day 13 (C) and Brn2 and Tuj1 on day 15 (D).
(E–I) Birth-date analysis using BrdU-pulse-labeling culture on the indicated days. Cryosections of SFEBq aggregates on day 16 were immunostained for BrdU
(E–H), Ctip2 (E and G) and Brn2 (F and H) after overnight treatment with BrdU on day 10 (E and F) or day 12 (G and H). (I) Quantification of the BrdU+cells
expressing each cell-type-specific marker.
(J) Schematic of the experiment of timed neuronal induction.
(K and L) Expression of Reelin and Ctip2 in the cultured aggregates after forced induction on day 9 (K) and day 12 (L).
(M) Ctip2+cells in the experiment of (L) were Emx1+.
Cell Stem Cell
Corticogenesis in ESC Culture
528 Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc.
generation of patterned cortical neuroepithelial tissues from
patternless ESC aggregates.
Such remarkable apical-basal arrangement is not observed in
neurosphere culture (subjected to quick reaggregation into
similarly-sized spheres), whether the cell source is dissociated
mouse cortical tissues (E13–E15) or dissociated SFEBq-induced
cortical cells (day 10) (data not shown). Therefore, a simple self-
sorting mechanism (sorting via selective cell adhesion; Townes
and Holtfreter, 1955) of different cortical components within
the aggregate alone cannot fully explain the formation of this or-
ganized structure, suggesting that the continuous presence and
function of polarized mitotic neuroepithelia are essential for this
self-organization. Notch may provide one of the key signals that
supports the maintenance of such neuroepithelia, give that,
for instance, the disappearance of the rosette structures is ac-
celerated by DAPT treatment during days 8–10 (Figures S1P
With respect to the neuroepithelial structures, there is an in-
triguing difference between the mESC- and hESC-derived corti-
cal tissues. In the mESC-derived cortical tissues, the continuous
lia after day 7 of SFEBq culture (Figure 1G), and these rigid
rosette structures gradually become less organized after day 12
as mitotic neuronal progenitors decrease. In contrast, the hESC-
derived tissues retain a more continuous neuroepithelial struc-
ture even on day 46 (Figure 6). In some aggregates, a single
(often folded) continuous neuropithelial structure occupies the
most area in the aggregate even after such long-term culture
(Figure S7E). One interpretation of this difference in structural
stability is that the balance between differentiation and self-
renewal may be different between hESC- and mESC-derived
progenitors, in a manner that favors the hESC-derived neuroepi-
thelia (note that the Pax6+mitotic zones in the day 46 human
neuroepithelia are much thicker than the mouse ones on day 10;
Figure 6). Another possible explanation is that hESC-derived
neural progenitors generally have a greater mechanic tendency
to form rigid epithelial structures, which may be relevant to the
large size of the human cortex.
Ascomparedtopreviousapproaches forESCdifferentiation, the
SFEBq culture isunique in that it supports the tissue formation of
cortical neuroepithelia in a spatially and temporally controlled
pattern, rather than just the differentiation of certain desired neu-
rons. The in vitro generation of cortical tissues from ESCs by
SFEBq culture will help elucidate various basic aspects of corti-
cogenesis by providing a versatile in vitro system for studying
complex events such as neocortical arealization, axonal guid-
ance, and neuronal migration.
The in vitro generation of six-layered neocortical tissues is an
intriguing but challenging task for future investigation. One of the
ability to form an inside-out pattern in SFEBq culture (Figure 6;
either, and our preliminary double-pulse labeling analyses using
BrdU/IdU (day 8/9 or day 9/10) have shown that later-born neu-
rons are not necessarily located in the relatively outer zone of the
rosette on day 12 (Figure S6P) or day 13 (data not shown). Fur-
ther mechanistic understanding of elementary regulatory factors
is essential for a future success in the in vitro production of well-
stratified cortical tissues, as well as major technical innovations
enabling the long-term, three-dimensional culture of thick corti-
of cortical tissues by the current SFEBq method should provide
a valuable resource of functional cortical neurons and tissues for
use in future medical applications. Since SFEBq-induced corti-
cal tissues form self-organized long-range neuronal networks
(Figures 2 and 3), they should be much more useful in determin-
ing pathogenesis, drug discovery and regenerative medicine for
intractable brain diseases than isolated neurons are. Moreover,
pathogenesis and drug discovery studies could be improved
by combing the SFEBq culture with disease-specific human iPS
cells (Takahashi et al., 2007), given that the SFEBq procedure
with the ROCK inhibitor (used in this study for human ESCs) is
also applicable to human iPS cell culture (see the Supplemental
With respect to regenerative medicine, given that the SFEBq
culture allows both Fgf8-induced rostral cortex specification
one attractive topic of future investigation is to finely direct the
differentiation and selection of Ctip2+layer-V neurons of the mo-
tor cortex (part of the Coup-TF1?rostral cortex) from hESCs,
which are lost in the ALS patient. In fact, Fgf8-treated SFEBq
tissues (from mESCs) contain numerous Coup-TF1?/Ctip2+
layer-V neurons (Figures S4G–S4I). Detailed mechanistic under-
standing of neocortical arealization should be vital to promote
this direction of application.
Mouse ESCs (EB5) and Bf1::Venus ESCs (number 2-1) were maintained as
described (Watanabe et al., 2005). Differentiation Medium was prepared as
follows: G-MEM supplemented with 10% Knockout Serum Replacement
(KSR; Invitrogen), 2 mM glutamine, 1 mM pyruvate, 0.1 mM nonessential
amino acids, 0.1 mM 2-ME, 250 ng/ml recombinant human Dkk-1, and
1 mg/ml recombinant human Lefty-1 (which can be replaced by 10 mM
SB431542). For SFEBq culture, ESCs were dissociated to single cells in
0.25% trypsin-EDTA (Invitrogen) and quickly reaggregated in differentiation
medium (3000 cells/150 ml/well) using 96-well low cell-adhesion plates
(Sumilon Spheroid Plates, Sumitomo; Lipidure-coat U96w from Nunc can
also be used). On day 7, cell aggregates were transferred to a 10 cm bacte-
rial-grade dish in N2 medium (DMEM/F12 supplemented with N2). The day
on which ESCs were seeded to differentiate is defined as differentiation
hESCs were maintained and cultured as described (Watanabe et al., 2007).
Differentiation Medium for SFEBq was the same with that for mESCs except
that G-MEM/10% KSR is replaced with DMEM/F12/20% KSR for hESCs.
Additional details are in the Supplemental Experimental Procedures.
Immunostaining, FACS Sorting, and Reaggregation Culture
Immunocytochemistry was performed as described (Watanabe et al., 2005,
2007) using antibodies listed in the Supplemental Experimental Procedures.
Values shown on graphs represent the mean ± SE. For FACS analysis, cells
(N–P) The cell percentages of Reelin+(N), Ctip2+(O), and Brn2+(P).
Scale bars: 100 mm in (B), (C), and (D); 100 mm in (K) and (L); and 60 mm in (M). The bars in the graph represent standard errors.
Cell Stem Cell
Corticogenesis in ESC Culture
Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc. 529
were counted with FACSAria (BD), and the data were analyzed with the
FACSDiva software (BD). The sorted cells were collected in ice-cold N2
medium containing 10% FBS and quickly reaggregated using low cell-adhe-
sion 96-well culture plates (5000 cells/well). Recombinant mouse FGF8b
(50 ng/ml), mouse FGFR3-fc (500 ng/ml), human BMP4 (0.5 ng/ml), and
Figure 6. Self-Organized Formation of Polarized
Cortical Tissues from SFEBq-Cultured mESCs
(A and B) Cryosections of SFEBq-cultured mESC aggre-
gates on day 7 (A) and day 8 (B) were immunostained
with Reelin and Bf1.
(C) Reelin+/Tbr1+neurons and Emx1+rosettes on day 10.
Arrowheads, Reelin+/Tbr1+neurons in the superficial zone
(facing to the aggregate surface) of the Emx1+rosettes.
(D and E) Immunostaining of the cortical markers Pax6,
Tbr1 (D), and Tbr2 (E) on day 10.
(F) Schematic of the cellular distribution pattern of the
SFEBq-induced cortical rosette.
(G–I) The distribution of the early/lower CP marker Ctip2
(H and I) and the late/upper CP marker Cux1 (H) and
Satb2 (I) in Emx1+clusters (G).
(J–R) Cryosections of SFEBq-cultured hESC aggregates
on day 46 were stained for cortical markers, as indicated
in each panel. In (J) and (L)–(N), confocal images of serial
sections at the same region are shown. (K) shows a high
power image of the perilumenal portion. The hESC-
derived Bf1+/N-cadherin+neuroepithelial structure coex-
pressed Pax6 (M) and Emx1 (N). Tbr1 (N) and Ctip2 (Q)
were expressed in the TuJ1+zone located superficially
to the Bf1+/Pax6+neuroepithelium, while Tbr2+cells
were mostly found in or adjacent to the Pax6+neuroepi-
Scale bars: 30 mm in (A) and (B); 50 mm in (C), (D), and (E);
60 mm in (G)–(I); 20 mm in (K); and 100 mm in (J), (L)–(O), (P),
human Wnt3a (20 ng/ml) were added to the culture me-
dium. For forced neuronal differentiation, DAPT (10 mM)
was added on the next day of FACS sorting. For induc-
tion of Tbx21+neurons, Bf1::Venus+cells were sorted
by FACS and reaggregated on day 7, treated with
FGF8 (50 ng/ml), and DAPT (10 mM) from day 8 and cul-
tured for 2 weeks. Tbx21+neurons were not induced by
DAPT treatment alone.
For in vitro birth-date analysis (Ajioka and Nakajima,
2005), aggregates were treated with BrdU (5 mg/ml) on
day 8, 9, 10, 12, or 14 and rinsed with medium to re-
move it on the next day. On day 16, cell aggregates
were fixed and cryosectioned. Sections were immuno-
stained for BrdU and each layer-specific marker as indi-
cated in Figure 5. The percentages of BrdU+cells in the
Bf1+/Ctip2+(most of the Ctip2+cells were Emx1+), Bf1+/
Brn2+, Reelin+, or Tbr1+cells were quantified. For the
quantification, 20–25 aggregates were examined for
each experiment, which was repeated at least three
Brain Slice Coculture Assay
For coculture with forebrain slices, E14.5 or P1 mouse
brains were excised, and coronal sections (200 mm) were
prepared usinga vibratome(F1000SL,Leica). Bf1::Venus+
neuronal masses were cut out in appropriate sizes from
the SFEBq cell aggregates under a fluorescent dissecting
microscope and cocultured with the forebrain slice (by in-
serting the masses into the ventricle space so that the inserted aggregates
were incontact with bothpallial and subpallial walls) for 3or 6days on a Trans-
well culture insert (Corning) containing the slice-culture medium (DMEM/F12,
N2 supplement,15% FBS, and penicillin-streptomycin) under 40%O2and 5%
Cell Stem Cell
Corticogenesis in ESC Culture
530 Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc.
ForCa2+imaging,cell aggregates were subjectedto filtercultureusing a Trans-
gregate was incubated in fluo4-AM solution as described previously (Ikegaya
The Supplemental Data include Supplemental Experimental Procedures,
seven figures, and four movies and can be found with this article online at
to Dr. A. Miyawaki for Venus cDNA, to Dr. Y. Yoshihara for Tbx21 antibody, to
Dr.Y. Ono for Lmx1aantibody,to Dr.K. Mori for Telencephalin antibody, toDr.
M. Ikeya for advice on homologous recombination, and to members of the
Sasai lab for discussion and advice. This work was supported by grants-
in-aid from MEXT, the Kobe Cluster Project, and the Leading Project (Y.S.).
Received: July 31, 2008
Revised: September 6, 2008
Accepted: September 9, 2008
Published: November 5, 2008
Adelsberger, H., Garaschuk, O., and Konnerth, A. (2005). Cortical calcium
waves in resting newborn mice. Nat. Neurosci. 8, 988–990.
Ajioka, I., and Nakajima, K. (2005). Birth-date-dependent segregation of the
mouse cerebral cortical neurons in reaggregation cultures. Eur. J. Neurosci.
Arlotta, P., Molyneaux, B.J., Chen, J., Inoue, J., Kominami, R., and Macklis,
J.D. (2005). Neuronal subtype-specific genes that control corticospinal motor
neuron development in vivo. Neuron 45, 207–221.
Bielle, F., Griveau, A., Narboux-Neme, N., Vigneau, S., Sigrist, M., Arber, S.,
Wassef, M., and Pierani, A. (2005). Multiple origins of Cajal-Retzius cells at
the borders of the developing pallium. Nat. Neurosci. 8, 1002–1012.
Englund,C.,Fink,A.,Lau,C., Pham,D.,Daza,R.A.,Bulfone, A.,Kowalczyk,T.,
and Hevner, R.F. (2005). Pax6, Tbr2, and Tbr1 are expressed sequentially by
radial glia, intermediate progenitor cells, and postmitotic neurons in develop-
ing neocortex. J. Neurosci. 25, 247–251.
Frantz, G.D., and McConnell, S.K. (1996). Restriction of late cerebral cortical
progenitors to an upper-layer fate. Neuron 17, 55–61.
Garaschuk, O., Linn, J., Eilers, J., and Konnerth, A. (2000). Large-scale oscil-
latory calcium waves in the immature cortex. Nat. Neurosci. 3, 452–459.
Gaspard, N., Bouschet, T., Hourez, R., Dimidschstein, J., Naeije, G., van den
Ameele, J., Espuny-Camacho, I., Herpoel, A., Passante, L., Schiffmann, S.N.,
et al. (2008). An intrinsic mechanism of corticogenesis from embryonic stem
cells. Nature 455, 355–357.
Gonzalez, J.L., Russo, C.J., Goldowitz, D., Sweet, H.O., Davisson, M.T., and
Walsh, C.A. (1997). Birthdate and cell marker analysis of scrambler: a novel
sci. 17, 9204–9211.
Guillemot, F. (2005). Cellular and molecular control of neurogenesis in the
mammalian telencephalon. Curr. Opin. Cell Biol. 17, 639–647.
Hanashima, C., Li, S.C., Shen, L., Lai, E., and Fishell, G. (2004). Foxg1 sup-
presses early cortical cell fate. Science 303, 56–59.
Hebert, J.M., Hayhurst, M., Marks, M.E., Kulessa, H., Hogan, B.L., and
McConnell, S.K. (2003a). BMP ligands act redundantly to pattern the dorsal
telencephalic midline. Genesis 35, 214–219.
Hebert, J.M., Lin, M., Partanen, J., Rossant, J., and McConnell, S.K. (2003b).
FGF signaling through FGFR1 is required for olfactory bulb morphogenesis.
Development 130, 1101–1111.
Hevner, R.F., Daza, R.A., Rubenstein, J.L., Stunnenberg, H., Olavarria, J.F.,
and Englund, C. (2003a). Beyond laminar fate: toward a molecular classifica-
tion of cortical projection/pyramidal neurons. Dev. Neurosci. 25, 139–151.
Hevner, R.F., Neogi, T., Englund, C., Daza, R.A., and Fink, A. (2003b). Cajal-
Retzius cells in the mouse: transcription factors, neurotransmitters, and birth-
days suggest a pallial origin. Brain Res. Dev. Brain Res. 141, 39–53.
Hevner, R.F., Shi, L., Justice, N., Hsueh, Y., Sheng, M., Smiga, S., Bulfone, A.,
Goffinet, A.M., Campagnoni, A.T., and Rubenstein, J.L. (2001). Tbr1 regulates
differentiation of the preplate and layer 6. Neuron 29, 353–366.
Ikegaya, Y., Le Bon-Jego, M., and Yuste, R. (2005). Large-scale imaging of
cortical network activity with calcium indicators. Neurosci. Res. 52, 132–138.
Kinney, J.W., Davis, C.N., Tabarean, I., Conti, B., Bartfai, T., and Behrens,
M.M. (2006). A specific role for NR2A-containing NMDA receptors in the main-
J. Neurosci. 26, 1604–1615.
is required for development of the mammalian hippocampus. Development
Mangale, V.S., Hirokawa, K.E., Satyaki, P.R., Gokulchandran, N., Chikbire, S.,
Subramanian, L., Shetty, A.S., Martynoga, B., Paul, J., Mai, M.V., et al. (2008).
Lhx2 selector activity specifies cortical identity and suppresses hippocampal
organizer fate. Science 319, 304–309.
Mitsui, S., Saito, M., Mori, K., and Yoshihara, Y. (2007). A transcriptional en-
hancer that directs telencephalon-specific transgene expression in mouse
brain. Cereb. Cortex 17, 522–530.
Molyneaux,B.J., Arlotta, P., Menezes, J.R., and Macklis, J.D. (2007). Neuronal
subtype specification in the cerebral cortex. Nat. Rev. Neurosci. 8, 427–437.
ing in mice and humans. Nat. Neurosci. 4 (Suppl), 1199–1206.
Nagai, T., Ibata, K., Park, E.S., Kubota, M., Mikoshiba, K., and Miyawaki, A.
(2002). A variant of yellow fluorescent protein with fast and efficient maturation
for cell-biological applications. Nat. Biotechnol. 20, 87–90.
Nelson, B.R., Hartman, B.H., Georgi, S.A., Lan, M.S., and Reh, T.A. (2007).
Transient inactivation of Notch signaling synchronizes differentiation of neural
progenitor cells. Dev. Biol. 304, 479–498.
O’Leary, D.D., Chou, S.J., and Sahara, S. (2007). Area patterning of the mam-
malian cortex. Neuron 56, 252–269.
Puelles, L., Kuwana, E., Puelles, E., Bulfone, A., Shimamura, K., Keleher, J.,
Smiga, S., and Rubenstein, J.L. (2000). Pallial and subpallial derivatives in
the embryonic chick and mouse telencephalon, traced by the expression of
the genes Dlx-2, Emx-1, Nkx-2.1, Pax-6, and Tbr-1. J. Comp. Neurol. 424,
Rallu, M., Corbin, J.G., and Fishell, G. (2002). Parsing the prosencephalon.
Nat. Rev. Neurosci. 3, 943–951.
Shen, Q., Wang, Y., Dimos, J.T., Fasano, C.A., Phoenix, T.N., Lemischka, I.R.,
Ivanova, N.B., Stifani, S., Morrisey, E.E., and Temple, S. (2006). The timing of
cortical neurogenesis is encoded within lineages of individual progenitor cells.
Nat. Neurosci. 9, 743–751.
Shimogori, T., and Grove, E.A. (2005). Fibroblast growth factor 8 regulates
neocortical guidance of area-specific thalamic innervation. J. Neurosci. 25,
Soriano, E., and Del Rio, J.A. (2005). The cells of cajal-retzius: still a mystery
one century after. Neuron 46, 389–394.
Stoykova, A., Hatano, O., Gruss, P., and Gotz, M. (2003). Increase in reelin-
Takahashi, K., Tanabe, K., Ohnuki, M., Narita, M., Ichisaka, T., Tomoda, K.,
and Yamanaka, S. (2007). Induction of pluripotent stem cells from adult human
fibroblast by defined factors. Cell 131, 861–872.
Tao, W., and Lai, E. (1992). Telencephalon-restricted expression of BF-1,
a new member of the HNF-3/fork head gene family, in the developing rat brain.
Neuron 8, 957–966.
Cell Stem Cell
Corticogenesis in ESC Culture
Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc. 531
Townes, P., and Holtfreter, J. (1955). Directed movement and selective
adhesion of embryonic amphibian. J. Exp. Zool. 128, 53–120.
Watanabe, K., Kamiya, D., Nishiyama, A., Katayama, T., Nozaki, S., Kawasaki,
H., Watanabe, Y., Mizuseki, K., and Sasai, Y. (2005). Directed differentiation
of telencephalic precursors from embryonic stem cells. Nat. Neurosci. 8,
Watanabe, K., Ueno, M., Kamiya, D., Nishiyama, A., Matsumura, M., Wataya,
T., Takahashi, J.B., Nishikawa, S., Muguruma, K., and Sasai, Y. (2007). A
ROCK inhibitor permits survival of dissociated human embryonic stem cells.
Nat. Biotechnol. 25, 681–686.
Wataya, T., Ando, S., Muguruma, K., Ikeda, H., Watanabe, K., Eiraku, M., Ka-
wada, M., Takahashi, J., Hashimoto, N., and Sasai, Y. (2008). Minimization of
Exogenous Signals in ES Cell Culture Induces Rostral Hypothalamic Differen-
tiation. Proc. Natl. Acad. Sci. USA 105, 11796–11801.
Weigmann, A., Corbeil, D., Hellwig, A., and Huttner, W.B. (1997). Prominin,
a novel microvilli-specific polytopic membrane protein of the apical surface
of epithelial cells, is targeted to plasmalemmal protrusions of non-epithelial
cells. Proc. Natl. Acad. Sci. USA 94, 12425–12430.
Yoshihara, S., Omichi, K., Yanazawa, M., Kitamura, K., and Yoshihara, Y.
(2005). Arx homeobox gene is essential for development of mouse olfactory
system. Development 132, 751–762.
Yuste, R., Nelson, D.A., Rubin, W.W., and Katz, L.C. (1995). Neuronal domains
in developing neocortex: mechanisms of coactivation. Neuron 14, 7–17.
Zhou, C., Tsai, S.Y., and Tsai, M.J. (2001). COUP-TFI: an intrinsic factor for
early regionalization of the neocortex. Genes Dev. 15, 2054–2059.
Cell Stem Cell
Corticogenesis in ESC Culture
532 Cell Stem Cell 3, 519–532, November 6, 2008 ª2008 Elsevier Inc.