ArticlePDF Available

COUP-TFI specifies the medial entorhinal cortex identity and induces differential cell adhesion to determine the integrity of its boundary with neocortex

Authors:

Abstract and Figures

Development of cortical regions with precise, sharp, and regular boundaries is essential for physiological function. However, little is known of the mechanisms ensuring these features. Here, we show that determination of the boundary between neocortex and medial entorhinal cortex (MEC), two abutting cortical regions generated from the same progenitor lineage, relies on COUP-TFI (chicken ovalbumin upstream promoter–transcription factor I), a patterning transcription factor with graded expression in cortical progenitors. In contrast with the classical paradigm, we found that increased COUP-TFI expression expands MEC, creating protrusions and disconnected ectopic tissue. We further developed a mathematical model that predicts that neuronal specification and differential cell affinity contribute to the emergence of an instability region and boundary sharpness. Correspondingly, we demonstrated that high expression of COUP-TFI induces MEC cell fate and protocadherin 19 expression. Thus, we conclude that a sharp boundary requires a subtle interplay between patterning transcription factors and differential cell affinity.
Differential gene expression in NC and MEC. Nissl-stained sagittal sections of adult (P56) mouse cortex and its higher-magnification view around the border region between NC and MEC are shown in (A) and (B). (C) RNA sequencing (RNA-seq) analyses were performed with RNA from P56 NC and MEC. Red and blue dots in dot plots show MEC-and NC-enriched genes, respectively. (D) Top GO terms and their P values are shown for these differentially expressed genes. (E) Nissl-stained sagittal section of P7 mouse cortex shows a clear border between NC and MEC (indicated by arrowhead). Images at higher magnification around border region from RNA in situ hybridization and immunostaining reveal genes with a differential expression pattern between MEC and NC. Nissl-stained sagittal sections of embryonic (E17.5) mouse cortex and its higher magnification view around the border region between NC and AC are shown in (F) and (G). (H) RNA-seq analyses were performed with RNA from E17.5 NC and AC. Red and blue dots in dot plots show AC-and NC-enriched genes, respectively. (I) Top GO terms and their P values are shown for these differentially expressed genes. Overlap of these differentially expressed genes between P56 and E17.5 is shown in (J). (K) From the list of the differentially expressed genes, a battery of genes was identified and showed significantly differential expression between MEC/AC and NC. (L) Costaining of Satb2 (NC marker) and Nurr1 (MEC marker) reveals a sharp molecular change at the border between NC and MEC in both P7 and E17.5 cortices. FPKM, fragments per kilobase of transcript per million; FDR, false discovery rate; Hp, hippocampus. Scale bars, 500 m (A), and 400 m (E, F, and L).
… 
The expression level of COUP-TFI regulates the position of the border between MEC and NC. (A) Dot plot of RNA-seq analyses from E13.5 NC and AC shows NC-and AC-enriched genes. (B) Nissl staining and RNA in situ hybridization of multiple patterning TFs were performed on sagittal sections of E13.5 wild-type mice. Relative expression intensities of the patterning TFs are shown along the anterior-posterior (A-P) axis. COUP-TFI showed a high-caudal-to-low-rostral gradient with its highest expression at the posterior end of the dorsal telencephalon (indicated by asterisk). (C) Dorsal and posterior view for P0 control, cKO-het (COUP-TFI f/+ ;Emx1-Cre), and cKO-homo (COUP-TFI f/f ;Emx1-Cre) cortices. Cortices were processed for whole-mount in situ hybridization using probes for Satb2 and Nurr1 to respectively indicate NC and MEC. (D) Immunostaining for Satb2, Wfs1, and NF-M on P7 sagittal sections of control and cKO-homo cortices. Arrowhead and asterisk indicated the border between NC and MEC in the control and cKO-homo cortices, respectively. (E) Relative intensity of Satb2 (green) and NF-M (red) staining around the NC/MEC border in control (top) and cKO-homo (bottom). (F) The length of the green (Satb2 + ) and red (NF-M + ) overlapped region is significantly increased in the cKO-homo, suggesting that the border is less clear in the cKO-homo when compared with that in control. OB, Olfactory bulb. Scale bars, 400 m (B), 250 m (C), 500 m (D, left), and 300 m (D, right).
… 
High level of COUP-TFI expression cell-autonomously induces the neuronal fate of MEC. (A) Overexpression of COUP-TFI in the control parietal region leads to entering the instability region and inducing the emergence of ectopic MEC within the NC. (B) Ectopic MEC domains were visualized by the spikes (arrowheads) in expression of MEC markers (blue) and drops in expression of NC markers (orange). (C) COUP-TFI TG/TG cortices were transfected with control or CAG-Cre plasmids with mCherry expression vector by in utero electroporation at E13.5. On sagittal sections of E18.5 electroporated cortices, transfected cells were labeled by mCherry, and transgene expression was detected by Myc and COUP-TFI. (D) Relative mCherry intensity was measured across the electroporation domains. The orange line indicates the mean mCherry intensity. In control cortices, mCherry + cells show uniform distribution across the electroporation domain, while mCherry + cells form clusters in Cre-electroporated cortices. (E) Average number of electroporated cells adjacent to a given electroporated cell (within a 15-m radius) in Cre-electroporated cortices was significantly increased compared to the control (P < 0.0001). (F and G) Expression of NC-enriched markers (Tbr1, Ctip2, and Satb2) and MEC-enriched markers (Nrp2, Kitl, and Nurr1) among mCherry + cells in control or Cre-electroporated cortices. Significantly fewer COUP-TFI-overexpressing cells displayed NC markers, including Tbr1 (P = 0.0235), Ctip2 (P = 0.0099), and Satb2 (P = 0.0167), while most expressed Nrp2 (P = 0.0112). Scale bars, 150 m (C and F) and 100 m (G).
… 
Content may be subject to copyright.
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
1 of 14
NEUROSCIENCE
COUP-TFI specifies the medial entorhinal cortex
identity and induces differential cell adhesion
to determine the integrity of its boundary
with neocortex
Jia Feng1†‡, Wen-Hsin Hsu1‡, Denis Patterson2§||, Ching-San Tseng, Hsiang-Wei Hsing1,
Zi-Hui Zhuang1, Yi-Ting Huang1, Andrea Faedo, John L. Rubenstein3,
Jonathan Touboul2, Shen-Ju Chou1*
Development of cortical regions with precise, sharp, and regular boundaries is essential for physiological func-
tion. However, little is known of the mechanisms ensuring these features. Here, we show that determination of
the boundary between neocortex and medial entorhinal cortex (MEC), two abutting cortical regions generated
from the same progenitor lineage, relies on COUP-TFI (chicken ovalbumin upstream promoter–transcription fac-
tor I), a patterning transcription factor with graded expression in cortical progenitors. In contrast with the classical
paradigm, we found that increased COUP-TFI expression expands MEC, creating protrusions and disconnected
ectopic tissue. We further developed a mathematical model that predicts that neuronal specification and differ-
ential cell affinity contribute to the emergence of an instability region and boundary sharpness. Correspondingly,
we demonstrated that high expression of COUP-TFI induces MEC cell fate and protocadherin 19 expression. Thus,
we conclude that a sharp boundary requires a subtle interplay between patterning transcription factors and dif-
ferential cell affinity.
INTRODUCTION
How the brain acquires appropriate cell diversity and robust func-
tional organization is one of the most prominent questions in the
field of neural development. It is known that within the vertebrate
central nervous system, some boundaries, such as the midbrain-
hindbrain boundary, the zona limitans intrathalmica, or the borders
between rhombomeres in the hindbrain, are established between
lineage-restricted compartments and sharpened by differential cell
adhesion (1). However, the neuromeric organization and clear
boundaries have not been delineated in the forebrain, where the ce-
rebral cortex arises; the only forebrain boundary that has been char-
acterized is the pallial-subpallial boundary located between dorsal
and ventral telecenphalon. The mammalian cerebral cortex consists
of distinct cortical regions, including neocortex (NC), archicortex
(AC), paleocortex, and transitional cortices located between these
regions, each with unique functions, cytoarchitecture, patterns of
gene expression, and input and output projection patterns. In this
study, we sought to understand how different cortical regions segre-
gate from each other, focusing on the sharp boundary separating
the medial entorhinal cortex (MEC) and NC. Notably, the NC and
MEC are two vastly distinct cortices, although both are derived from
the same pool of Emx1-lineage cortical progenitors (2).
One central paradigm, the positional information theory or
French flag model, posits that gradients of transcription factors
(TFs) instruct newly generated neurons to adopt a specific fate (3).
In line with this model, several patterning TF gradients were shown
to support neuronal specification and area patterning in the NC
(48). However, a well-known limitation of positional information
theory is related to the sharpness and robustness of boundaries,
where ambiguous differentiation can lead to irregular, unpredict-
able, or nonsharp (“salt-and-pepper”) transitions (9). Hence, various
phenomena have been proposed to account for boundary regulari-
ty, including cell-sorting mechanisms (10), combined information
from multiple gradients (11,12), or cell aggregation and adhesion
(1315). To date, though, there has been no direct evidence to indi-
cate a causal relationship between any of these processes and the
regularity of boundaries between cortical regions.
In this work, we identify key molecular mediators of neuronal
differentiation that distinguish the MEC and NC and also generate
the border between these cortical regions. We show that the con-
centration of the nuclear receptor COUP-TFI (chicken ovalbumin
upstream promoter–TF I; also called NR2F1) in progenitors plays a
central role in determining the NC/MEC border. COUP-TFI was
previously shown to be a key determinant of cortical development,
as COUP-TFI mutant mice have prominent defects in neuronal
specification and cortical and hippocampal patterning (1621).
Furthermore, de novo mutations in the human COUP-TFI gene
cause Bosch-Boonstra-Schaaf optic atrophy syndrome, which is
characterized by cortical malformation and various sensory and
cognitive deficits (2224). We confirmed that lowering the COUP-
TFI expression level caudally shifted the NC/MEC border, in agree-
ment with previous findings (18). Moreover, we demonstrated that
1Institute of Cellular and Organismic Biology, Academia Sinica, Taipei, Taiwan. 2Depart-
ment of Mathematics and Volen National Center for Complex Systems, Brandeis
University, Waltham, MA 02454, USA. 3Nina Ireland Laboratory of Developmental
Neurobiology, Department of Psychiatry, UCSF Weill Institute for Neurosciences,
University of California, San Francisco, San Francisco, CA 94158, USA.
*Corresponding author. Email: schou@gate.sinica.edu.tw
†Present address: Yong Loo Lin School of Medicine, National University of Singapore,
Singapore.
‡These authors contributed equally to this work and are listed in alphabetical order.
§These authors contributed equally to this work and are listed in alphabetical order.
||Present address: High Meadows Environmental Institute, Princeton University,
Princeton, NJ 08544, USA.
¶Present address: Axxam S.p.A; Milan; Italy.
Copyright © 2021
The Authors, some
rights reserved;
exclusive licensee
American Association
for the Advancement
of Science. No claim to
original U.S. Government
Works. Distributed
under a Creative
Commons Attribution
NonCommercial
License 4.0 (CC BY-NC).
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
2 of 14
a lack of COUP-TFI reduced the sharpness of the NC/MEC border.
On the other hand, mutant mice with COUP-TFI overexpression
exhibited not only an expansion of the MEC at the expense of
NC but also a dislocation of the boundary and emergence of protru-
sions or disconnected ectopic MEC regions. Further theoretical and
experimental investigations then revealed that differential cell ad-
hesion enhances sharpness of the boundary, and excessive adhesion
induces the emergence of irregular boundaries. Together, these
findings demonstrate that the level of patterning TF expression can
determine region-specific neuronal properties and the expression
of region-specific adhesion molecules. The adhesion molecules are
essential to the formation of boundaries between cortical regions,
but when the differential cell adhesion is too much, the boundary
may break down. Thus, the emergence of sharp boundaries invivo
requires subtle interplay between patterning TF gradients and cell
adhesion.
RESULTS
Distinct gene expression profiles between NC and MEC
To study boundary formation, we focused on the border between
the NC and the MEC. Although NC and MEC both consist of six
layers and both are derived from the Emx1 lineage (fig. S1A), a clear
cytoarchitectural border can be detected between the regions in the
adult cortex (Fig.1,AandB). Assuming that distinct cellular prop-
erties in MEC and NC segregate these two structures, we first com-
pared the gene expression profiles in NC and MEC. We found that
2039 genes were enriched in NC and 1507 genes were enriched in
MEC (Fig.1C). These genes included some involved in nervous
system development [Gene Ontology (GO):0048666], neuronal dif-
ferentiation (GO:0030182), and synaptic signaling (GO:0099536)
(Fig.1D), suggesting that adult NC and MEC show substantial dif-
ferences in neuronal properties. From the list of NC/MEC differen-
tially expressed genes, we identified a battery that labels specific
Fig. 1. Differential gene expression in NC and MEC. Nissl-stained sagittal sections of adult (P56) mouse cortex and its higher-magnification view around the border
region between NC and MEC are shown in (A) and (B). (C) RNA sequencing (RNA-seq) analyses were performed with RNA from P56 NC and MEC. Red and blue dots in dot
plots show MEC- and NC-enriched genes, respectively. (D) Top GO terms and their P values are shown for these differentially expressed genes. (E) Nissl-stained sagittal
section of P7 mouse cortex shows a clear border between NC and MEC (indicated by arrowhead). Images at higher magnification around border region from RNA in situ
hybridization and immunostaining reveal genes with a differential expression pattern between MEC and NC. Nissl-stained sagittal sections of embryonic (E17.5) mouse
cortex and its higher magnification view around the border region between NC and AC are shown in (F) and (G). (H) RNA-seq analyses were performed with RNA from
E17.5 NC and AC. Red and blue dots in dot plots show AC- and NC-enriched genes, respectively. (I) Top GO terms and their P values are shown for these differentially ex-
pressed genes. Overlap of these differentially expressed genes between P56 and E17.5 is shown in (J). (K) From the list of the differentially expressed genes, a battery of
genes was identified and showed significantly differential expression between MEC/AC and NC. (L) Costaining of Satb2 (NC marker) and Nurr1 (MEC marker) reveals a
sharp molecular change at the border between NC and MEC in both P7 and E17.5 cortices. FPKM, fragments per kilobase of transcript per million; FDR, false discovery rate;
Hp, hippocampus. Scale bars, 500 m (A), and 400 m (E, F, and L).
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
3 of 14
layers in the MEC or NC, including genes encoding TFs, such as Ctip2,
Lhx2, Tbr1, Satb2, and Nurr1; cell surface molecules, such as Reelin
(Rln); neurofilaments, such as Nef3; and genes involved in cellular
signaling and neuronal activity, Kitl, Vglut2 (Slc17a6), Calb1 (CB), Wfs1,
and Nts. Most of these factors play important roles in cortical develop-
ment (2629). More specifically, many of the genes, especially Reelin,
Wfs1, and Calb1, have been shown to label specific MEC cell types
(30,31), and their differential expression patterns could be used to
define the border between NC and MEC (Fig.1E and fig. S1C).
To determine when these genetic differences between NC and
MEC were established, we examined gene expression profiles in
these cortical structures at embryonic stages. At embryonic day 17.5
(E17.5), a time point when most cortical neurons have been gener-
ated but before cortical neurons receive stimulation from the pe-
riphery, we could not define an apparent NC/MEC border by Nissl
staining (Fig.1,FandG). However, we did find that differences in
gene expression between NC and AC (the caudal part of the dorsal
telencephalon, including hippocampus and MEC) could be detected
at E17.5, similar to those detected in adult (Fig.1H). These embryonic
NC/AC differentially expressed genes were enriched in nervous sys-
tem development (GO:0048666), neuron differentiation (GO:0030182),
cell migration (GO:0016477), and cell adhesion (GO:0007155) (Fig.1I
and fig. S1B), suggesting that distinct regional neuronal properties
are established early during neurogenesis. Comparing the NC/MEC
and NC/AC differentially expressed genes from adult and E17.5,
respectively, we found many that were present in both lists (Fig.1J).
Most of the genes showed consistent differential expression patterns
in E17.5 and adult; for example, Nurr1 and Nrp2 were enriched in
the MEC, while Satb2 and RORb were enriched in the NC (Fig.1K).
However, some genes showed different patterns in E17.5 and adult,
such as Lmo4 and Nef3 (Fig.1K).
By probing NC- and MEC-specific genes, we could detect that a
sharp transition at the border of the two structures is already present
at E17.5 (Fig.1L). Thus, we concluded that a molecular boundary
between NC and MEC was established during embryonic development,
at the position where changes in gene expression were detected.
Decreased COUP-TFI expression posteriorly shifts
and degrades the border between MEC and NC
Next, we investigated how the NC/MEC border is established in the
progenitors of the Emx1 lineage. Similar to our experiment with
E17.5 cortices, we performed RNA sequencing (RNA-seq) analyses to
compare gene expression profiles at E13.5 between the presumptive
progenitors giving rise to NC and AC. We found 2112 and 1794 genes
that were enriched in AC and NC progenitors, respectively (Fig.2A).
Among these differentially expressed genes, we focused on TFs that
show notable differential expression in NC and AC progenitors, be-
cause TFs are known to play critical roles in specifying positional in-
formation in neuronal progenitors. AC-enriched TFs, including
COUP-TFI, Emx1, Emx2, and Lhx2, could potentially regulate the
formation of NC/MEC border. By quantifying their expression gradi-
ents, we unexpectedly found a unique expression pattern for COUP-
TFI: It was most highly expressed by cortical progenitors located at
the caudal end of the dorsal telencephalic vesicle, where MEC pre-
sumably originates (Fig.2B and fig. S2A). Moreover, along the medial-
lateral axis, the high lateral expression of COUP-TFI also correlated
with the position where MEC is located relative to NC (fig. S2, B and
C). On the basis of these observations, which were corroborated by
the fact that conditional knockout (cKO) of COUP-TFI from cortical
progenitors results in reduced MEC size (18), we hypothesized that
the relatively high level of COUP-TFI expression induces the forma-
tion of MEC and therefore creates the border between MEC and NC.
To test how COUP-TFI expression level affects the border for-
mation between NC and MEC, we used different conditional alleles
to manipulate COUP-TFI expression levels by Emx1-Cre (2) and
examined the impact on the size and relative location of MEC. We
first generated cKO-het (COUP-TFIf/+;Emx1-Cre) and cKO-homo
(COUP-TFIf/f;Emx1-Cre) to lower COUP-TFI expression to differ-
ent levels in cortical progenitors and neurons (fig. S3A). By E15.5,
COUP-TFI expression was about 50% of the control level in the
cKO-het, and it was barely detectable in the cKO-homo (fig. S3,
A and B). To assess the relative positions of NC and MEC within the
intact cortical hemisphere, we used whole-mount in situ hybridiza-
tion to detect the expression of Satb2 and Nurr1, which respectively
label NC and MEC (32,33). We found complementary Satb2 and
Nurr1 expression patterns in control cortices at postnatal day 0 (P0)
(Fig.2C). Consistent with previous findings (18), the Nurr1+ domain
was markedly decreased in size, with a corresponding expansion of
the Satb2+ domain in the cKO-homo compared with control. Notably,
we found that MEC size was regulated by COUP-TFI in a dose-
dependent fashion, with the Nurr1+ domain in the cKO-het cortex
being larger than that in cKO-homo cortices and smaller than that in
control cortices (Fig.2C). Thus, we showed that decreasing COUP-
TFI expression level led to a caudal shift of the NC/MEC border. In
addition to its position, we further examined the integrity of the
NC/MEC border by examining the complementary expression of
NC- and MEC-specific marker genes. We found that the border in-
deed became less sharp, with more overlap of NC/MEC marker gene
expression in COUP-TFI cKO-homo cortices (Fig.2,EandF). Thus,
our findings suggested that COUP-TFI expression level regulates
the position and sharpness of NC/MEC border.
Increased COUP-TFI induces ectopic MEC formation
In addition to the loss-of-function experiments, we also tested the
impact of increasing COUP-TFI expression. We used Emx1-Cre to
induce the expression of a copy of the hCOUP-TFI transgene (34)
in cortical progenitors and their progeny of cTG mice (COUP-TFITG/O;
Emx1-Cre or cTG-het); the hCOUP-TFI transgene expression level
was about 50% of the endogenous mouse (mCOUP-TFI) expression
level (fig. S3, C to E). Notably, we found that MEC was rostrally
expanded in the cTG mice, as determined by the expression of MEC-
enriched genes, including Nurr1, Wfs1, and VgluT2 (Fig.3A and
fig. S4A). Furthermore, ectopic domains of MEC-enriched marker
gene expression were detected in the caudal NC, suggesting that ec-
topic MECs were generated in the NC of cTG (Fig.3,AandB). The
generation of ectopic MEC in the cTG NC was unexpected and is,
to the best of our knowledge, unprecedented. We thus further care-
fully confirmed the properties of these ectopic MEC in the COUP-TFI
cTG by comparing gene expression patterns, neuronal birthdates,
and connectivity of the ectopic structures.
Similar to the endogenous MEC (Fig.3,A,C,andD), we found the
ectopic MECs exhibited consistent layer-specific expression of Nurr1
in layers 2/3 and 5/6; Kitl and Vglut2 in layer 2/3; Calb, Lhx2, Nef3,
Rln, and Wfs1 in layer 2; and Lmo4 in layer 5/6 (Fig.3, A,E,andF,
and fig. S4, A to C). Along with the rostral expansion and ectopic ex-
pression of MEC genes, we observed down-regulation of NC-enriched
marker genes, including Satb2, Rorb, Id2, Er81, Fezf2, and Fosl2
(Fig.3B and fig. S4, A to C). Furthermore, similar to that in the cTG
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
4 of 14
(fig. S4D), the ectopic expression of MEC-enriched genes (e.g., Nrp2)
and down-regulation of NC-enriched genes (e.g., Rorb) were also
found in the D6-COUP-TFI transgenic cortices (fig. S4E), where
mCOUP-TFI is overexpressed in cortical progenitors driven by the
Dach1 promoter (35). The presence of ectopic domains expressing
MEC-specific genes in multiple lines of COUP-TFI transgenic mice
suggested that overexpression of COUP-TFI in cortical progenitors
induces fate change in the NC cells, which leads to generation of
ectopic MEC-like structures.
Although the sequential maturation of MEC neurons has been
described (36), the pattern of neurogenesis in different MEC layers
has remained uncharacterized. Therefore, we performed neuronal
birthdating to compare neurogenesis programs between MEC and
NC. Neurons in both structures were generated in an inside-out
pattern (fig. S5, A to C). In both MEC and NC, deep-layer neurons
were generated at around E11.5. However, while most NC upper-
layer neurons were generated from E15.5 to E17.5, many of the layer
2/3 MEC neurons were generated at around E13.5, and no major
numbers of MEC neurons were generated after E15.5 (fig. S5, A to
C). The early termination of neurogenesis in MEC agrees with the fact
that MEC has fewer neurons within a given cortical column than the
NC (fig. S5D). We first showed that COUP-TFI overexpression in
Fig. 2. The expression level of COUP-TFI regulates the position of the border between MEC and NC. (A) Dot plot of RNA-seq analyses from E13.5 NC and AC shows
NC- and AC-enriched genes. (B) Nissl staining and RNA in situ hybridization of multiple patterning TFs were performed on sagittal sections of E13.5 wild-type mice. Rela-
tive expression intensities of the patterning TFs are shown along the anterior-posterior (A-P) axis. COUP-TFI showed a high-caudal–to–low-rostral gradient with its highest
expression at the posterior end of the dorsal telencephalon (indicated by asterisk). (C) Dorsal and posterior view for P0 control, cKO-het (COUP-TFIf/+;Emx1-Cre), and
cKO-homo (COUP-TFIf/f;Emx1-Cre) cortices. Cortices were processed for whole-mount in situ hybridization using probes for Satb2 and Nurr1 to respectively indicate NC
and MEC. (D) Immunostaining for Satb2, Wfs1, and NF-M on P7 sagittal sections of control and cKO-homo cortices. Arrowhead and asterisk indicated the border between
NC and MEC in the control and cKO-homo cortices, respectively. (E) Relative intensity of Satb2 (green) and NF-M (red) staining around the NC/MEC border in control (top)
and cKO-homo (bottom). (F) The length of the green (Satb2+) and red (NF-M+) overlapped region is significantly increased in the cKO-homo, suggesting that the border
is less clear in the cKO-homo when compared with that in control. OB, Olfactory bulb. Scale bars, 400 m (B), 250 m (C), 500 m (D, left), and 300 m (D, right).
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
5 of 14
cTG does not affect neurogenesis by confirming that the number of
proliferating cortical progenitors is similar in cTG and control cor-
tices at E13.5 (fig. S5E), which is consistent with the similar cTG
and control cortical sizes (fig. S3L). In the cTG, we found that most
of the neurons in the ectopic MEC were generated by E13.5 and far
fewer neurons were generated at E15.5 and E17.5, when compared
with adjacent NC tissues (fig. S5F). Thus, our results showed that
the timings of neurogenesis initiation and termination in the ecto-
pic MEC domains were similar to those in the control endoge-
nous MEC.
Furthermore, in line with their distinct functions, MEC and NC
have different input and output projection patterns. The sensory
cortices in NC have reciprocal connections with thalamus, while
MEC is highly connected with hippocampus. Using DiI crystals in
the dorsal hippocampus to label neurons projecting to hippocampus,
we observed DiI-labeled neurons and neuronal projections in the
dorsal part of the MEC in control animals. However, in the cTG
hippocampus, DiI-labeled neurons and neuronal projections were
rostrally shifted and could even be detected in the ectopic MEC
(Fig.3G). To further confirm the formation of ectopic MEC in the
caudal NC in the cTG, we injected DiD in the P7 primary visual
cortex (V1) of control and cTG cortices. While DiD labeled the
reciprocal connections between V1 and thalamic dorsal lateral genic-
ulate nucleus (dLG) but not in the hippocampus in the control cortices,
we found DiD-labeled neuronal fibers in the perforant pathway in
hippocampus, in addition to the dLG, in the cTG cortices (Fig.3H).
Fig. 3. COUP-TFI overexpression in cTG cortical progenitors leads to rostral expansion of MEC and the formation of ectopic MEC. (A and B) Nissl staining, in situ
hybridization, and immunostaining were performed for MEC-enriched genes (A) and NC-enriched genes (B) on P7 sagittal sections of control and cTG (cTG-het; COUP-
TFITG/+;Emx1-Cre) cortices. The border between NC and MEC (marked by arrowhead) was rostrally shifted in the cTG cortices. Ectopic MEC domains (asterisks) were iden-
tified in the caudal NC of cTG. (C to F) Immunostaining of MEC layer markers demonstrates the similar molecular characteristics and layering structure in endogenous MEC
from control cortex (C and D) and ectopic MEC from cTG (E and F). (G) DiI crystal was placed at the dorsal hippocampus in P7 control and cTG cortices. DiI-labeled neurons
and neuronal processes could be detected in the dorsal MEC (arrowhead) in control and the rostrally expanded MEC (arrowhead) and ectopic MEC (asterisks) in cTG.
(H) DiD was placed in the primary visual cortex (V1) in P7 control and cTG cortices, and DiD-labeled neurons and neuronal processes could be detected in the dLG in both
control and cTG. DiD-labeled neuronal processes could also be detected in the perforant path (PP, arrows) in the cTG hippocampal formation (Hp). VB, ventrobasal nucle-
us. Scale bars, 500 m (A, B, G, and H) and 100 m (D and F).
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
6 of 14
In conclusion, on the basis of the expression of region-specific
layer markers, neurogenesis patterns, and connectivity, we showed
that COUP-TFI overexpression induces the formation of ectopic
MEC in the caudal NC, and these ectopic MEC domains truly resem-
bled endogenous MEC. Our results suggest that increased COUP-
TFI expression is sufficient to rostrally expand MEC and induce the
generation of ectopic MEC, at the expense of NC. These ectopic MEC
domains were not only present transiently during development, as
we found evidence for these ectopic MEC structures in the cTG caudal
NC persisted to at least P120 (fig. S6).
COUP-TFI expression level regulates the number
and location of ectopic MEC
To further examine whether COUP-TFI expression level deter-
mines the location of ectopic MECs, we generated cTG-homo mice
(COUP-TFITG/TG;Emx1-Cre), which express two copies of hCOUP-
TFI transgene, and the transgene expression level is similar to
the endogenous mouse (mCOUP-TFI) (fig. S3, C to E). Notably,
transgene expression did not significantly alter the expression level
of endogenous mCOUP-TFI (fig. S3D). However, the expression
gradient of total COUP-TFI in the cortical progenitors was altered
in the transgenic cortices, with cortical progenitors in the caudal
half of cTG-het and caudal three-quarters of cTG-homo cortices
expressing high levels of COUP-TFI (similar or higher than the
maximal COUP-TFI expression in wild type) (fig. S3, F to K). In the
cTG-homo, we found an increased overall number and more ros-
trally located ectopic MEC domains (Satb2 and Nurr1+) when com-
pared to cTG-het (Fig.4,AtoC). Together, these loss-of-function
and gain-of-function experiments suggested that the level of COUP-
TFI expression dose-dependently controls the size of MEC (fig. S3,
M and N) and ectopic MEC formation.
Mathematical model predicts that cell affinity mechanisms
can trigger the emergence of ectopic MEC
Previous studies of area patterning indicate that cortical progeni-
tors adopt a specific fate depending on the level of patterning TF
Fig. 4. The level of COUP-TFI overexpression in cortical progenitors correlates the expansion domain of ectopic MEC. Dorsal and posterior view for P0 cTG-het
(COUP-TFITG/O;Emx1-Cre) (A) and cTG-homo (COUP-TFITG/TG;Emx1-Cre) (B) cortices. Whole-mount in situ hybridization of Satb2 and Nurr1 was used to indicate NC and MEC,
respectively. (C) In situ hybridizations for Satb2 and Nurr1 on sagittal sections of P7 control, cTG-het, and cTG-homo showed ectopic Nurr1+Satb2 domains [arrowheads
in (A) to (C)] in the cTG NCs. The number of ectopic Nurr1+Satb2 domains in the NC was higher, and the domains were more rostrally located in the cTG-homo compared
to the cTG-het. (D) According to our new mathematical model, we can simulate the phenomenon similar to the experimentally observed phenotypes, even the formation
of ectopic MEC (Figs. 2C and 4, A to C). (E) The mathematical model predicts stable homogeneous MEC and NC domains (blue regions) at high concentrations of MEC and
NC patterning TFs (E and N), respectively. It also predicts Turing-like instability domain (red region) associated with ectopic domains (red) for certain ranges of 𝜌E and
𝜌N. COUP-TFI expression gradients associated with control, cKO-het, cKO-homo, cTG-het, or cTG-homo were plotted according to the data (Fig. 2B and fig. S3, F to H).
Gradients of cTG-het or cTG-homo models cut through the instability region generate ectopic domains. Scale bars, 250 m (A, B, and C).
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
7 of 14
expression. However, changes in the expression gradient of pattern-
ing TFs lead to areal shifts (8) but not to the emergence of ectopic
domains. Our observation of ectopic MEC in COUP-TFI trans-
genic mice thus challenges the classical conceptual frameworks
of cell fate determination. Previously, self-organization models
based on gene expression and diffusion were developed and shown
to accurately reflect observed shifts in the boundaries between brain
areas upon changes in the expression of patterning TFs (3739).
However, none of these models predict changes in boundary regular-
ity and integrity or the emergence of regular ectopic domains like
those that we observed in COUP-TFI transgenic cortices (figs. S11
and S12). We therefore revisited these classical models and devel oped
a system of equations combining positional information and
self-organization mechanisms with cell movement and differential
cell affinity. Our model describes the differentiation of neural pro-
genitors into NC or MEC cells as a result of three basic phenomena.
First, external cues are provided by two gradients of patterning TF,
one that promotes NC fate and an opposing gradient that pro-
motes MEC fate, the latter mimicking invivo expression of COUP-
TFI. Second, the model emulates the competition between expression
of NC and MEC genes. Eventually, the model incorporates cell move-
ment, both isotropic, via diffusion, and directional, through aggre-
gation terms that particularly reflect differential cell affinity. This
simple model showed that mechanisms based on TF gradients and
differential adhesion can support differentiation into regular re-
gions, with a boundary location dependent on patterning TF con-
centrations and sufficiently steep gradients. We further observed
that the boundaries may be sharpened by cell affinity levels (see fig.
S13), consistent with the sharp NC/MEC border that we observed
in animals (control in Fig.2D). However, quite unexpectedly, the
model also predicted that sufficiently strong differential affinity can
also cause a dynamical instability akin to that observed in the cele-
brated Turing model (40,41). The occurrence of such an instability
yields elongated ectopic domains (typically stripes or labyrinths) or
isolated aggregates depending on the expression levels of the pat-
terning TFs (Fig.4D and fig. S16). When patterning TF gradients
do not give rise to instability (as with the fitted gradients in the control),
a sharp boundary will be maintained upon changes in gradients.
However, when the patterning TF gradients intersect the instability
region (as is the case of COUP-TFI up-regulation; Fig.4E), the tran-
sition between the MEC and NC becomes irregular, yielding ectopic
MEC domains with topologies very similar to the experimentally
identified patterns (Fig.4D and fig. S20). Therefore, we used the
model to make two testable predictions (see “Pattern formation in
heterogeneous domains” section in the Supplementary Math model).
First, we validated the modeling assumption that differentiation can be
appropriately described as cell autonomous. To this end, we investig ated
whether overexpression of COUP-TFI would induce the emergence
of ectopic MEC cells in the region of overexpression. Second, the
instability arising in the theoretical model relies fundamentally on
differential cell affinities that lead cells at the MEC/NC boundary to
form clusters. We thus probed for evidence of these differential
affinities experimentally, searched for the molecular pathways
involved, and assessed the impact of COUP-TFI overexpression on
these pathways. The model predicts that a local overexpression of
COUP-TFI in the NC should induce the formation of multiple
MEC clusters (Fig.5,AandB). However, we predict that when cell
affinity mechanisms are weaker, we will observe fewer clusters with
lower cell density and less homogeneity (figs. S22 to S24).
High COUP-TFI expression cell-autonomously induces
ectopic MEC
According to the theoretical prediction, increasing COUP-TFI ex-
pression levels in cells expected to adopt the NC cell fate (e.g., in
parietal cortex) can move these cells into the instability region
(Fig.5A) and thereby induce cell-autonomous changes to their fate
and the emergence of one or multiple isolated clusters of ectopic
MEC (Fig.5B). To experimentally test this prediction, we first in-
vestigated whether high COUP-TFI expression levels could induce
MEC formation. We used in utero electroporation to transfect a
control vector or a Cre expression construct (CAG-Cre) along with
a mCherry expression vector (to label transfected cells) into the pa-
rietal cortex in COUP-TFITG/TG at E13.5 and analyzed the cortices at
E18.5. Cre expression in COUP-TFITG/TG cortices induced the ex-
pression of hCOUP-TFI in progenitors and their progeny (Fig.5C).
Most of the transfected cells in control and COUP-TFITG/TG cortices
were found in the cortical plate (CP). We found that transfected
cells in the control cortices were dispersed (Fig.5,CandD) and
expressed NC markers, such as Tbr1, Ctip2, or Satb2, but not Nrp2,
an MEC marker (Fig.5,FandG). However, induction of hCOUP-TFI
caused cell clustering (Fig.5,CandD). In these clusters, significantly
more transfected cells were found adjacent to a given transfected
cell (Fig.5E). Most of the COUP-TFI overexpressing cells did not
express NC-enriched genes, such as Tbr1, Ctip2, Satb2, or Rorb, and
instead expressed MEC-enriched genes, including Nrp2, Kitl, and
Nurr1 (Fig.5,FandG, and fig. S7, A and B). These MEC-like cell clusters
were not induced when Cre expression construct was electroporated
at E15.5 (fig. S7C), suggesting that there is a time window during
which COUP-TFI can induce NC-to-MEC cell fate change.
Furthermore, the cell clusters expressing MEC-enriched genes
induced by COUP-TFI overexpression were similar to those in the
ectopic MEC domains in the cTG cortices (Fig.3, AtoF), as pre-
dicted by our mathematic model (Fig.5B). In contrast, the model
did not predict any clustering when overexpressing COUP-TFI in
the MEC region. We experimentally confirmed that cell clustering
was not induced when Cre was electroporated into developing
MEC in COUP-TFITG/TG (fig. S7D). Thus, we provide evidence
that, within a certain time window, a local increase of COUP-TFI
expression in NC could induce MEC neuronal differentiation. In these
COUP-TFI–induced MEC cell clusters, MEC genes are up-regulated
and NC genes are repressed.
Differential affinity in MEC and NC
The model further suggested that this intriguing clustering effect
relies on differential MEC and NC cell affinities. As shown in Fig.6A,
inhibiting differential cell adhesion in the model leads to a disap-
pearance of the clusters in a region with mixed cell identity and low-
ers cell density in the clusters compared to the highly clustered and
dense ectopic regions predicted in the presence of cell adhesion
(Fig.5A). To test this prediction, we (i) assessed whether MEC and
NC cells show differentiated cell adhesion and (ii) identified the dif-
ferential cell adhesion molecules enhanced by COUP-TFI. To probe
whether differentiated MEC and NC cells spontaneously segregate,
we performed invitro cell aggregation assays with dye-labeled cells
dissociated from E13.5 presumptive NC and MEC. After 1 hour of
incubation, we assessed the propensity of cells to form aggregates in
populations of red- and green-labeled NC cells [NC (red)+NC
(green)] or red NC cells incubated with green MEC cells [NC
(red)+MEC (green)]. We found that cells from NC or MEC were
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
8 of 14
more likely to be next to cells of the same origin when compared with
cells from NC (P=0.025; Fig.6,CandD), suggesting that MEC and
NC cells spontaneously aggregate with cells of their own kind.
We next explored the mechanism underlying aggregation. The
theoretical model predicted that the differential affinities of MEC
and NC cells are key for cell clustering and boundary integrity (see
Figs.5A and 6A and “Cell Adhesion is Crucial for Ectopic Domains
Formation” section in the Supplementary Math model). We therefore
tested whether MEC and NC cells express different cell adhesion
molecules and whether high levels of COUP-TFI could induce the
expression of MEC-enriched cell adhesion molecules. On the basis
of the RNA-seq analysis of NC and AC gene expression profiles at
E13.5, we focused on a list of genes involved in homophilic cell ad-
hesion via plasma membrane adhesion (GO:0007156, P =0.00469)
(fig. S8, A and B). Using the Allen Developing Mouse Brain Atlas,
which includes expression patterns of about 2000 genes functionally
relevant to brain development, we confirmed that several homophilic
adhesion molecules have detectably differential expression along
the anterior-posterior axis in the developing telencephalon at E13.5
(fig. S8C). Comparing the expression of these genes in the anterior
and posterior ends of the E13.5 cortices, we found that Pcdh19
(encodes Protocadherin 19) was highly enriched in the caudal cor-
tex (fig. S8E) and exhibited the most notable difference in relative
expression levels between anterior and posterior cortices (fig. S8D)
(42). Because misexpression of Pcdh19 in the developing cortex is
known to cause cell clustering (4345), we first examined whether
Pcdh19 expression levels change when COUP-TFI expression is al-
tered. We found that the high-caudal–to–low-rostral expression
gradient of Pcdh19 was dose-dependently caudally shifted in the
cKO (fig. S8F) and that Pcdh19 expression is significantly increased
in the cTG (fig. S8, G and H).
We next demonstrated that COUP-TFI is able to directly bind to
a conserved Sp1/COUP-TFI binding site in the Pcdh19 promoter
region by chromatin immunoprecipitation (ChIP) (Fig.6,EandF).
Fig. 5. High level of COUP-TFI expression cell-autonomously induces the neuronal fate of MEC. (A) Overexpression of COUP-TFI in the control parietal region leads
to entering the instability region and inducing the emergence of ectopic MEC within the NC. (B) Ectopic MEC domains were visualized by the spikes (arrowheads) in ex-
pression of MEC markers (blue) and drops in expression of NC markers (orange). (C) COUP-TFITG/TG cortices were transfected with control or CAG-Cre plasmids with mCherry
expression vector by in utero electroporation at E13.5. On sagittal sections of E18.5 electroporated cortices, transfected cells were labeled by mCherry, and transgene
expression was detected by Myc and COUP-TFI. (D) Relative mCherry intensity was measured across the electroporation domains. The orange line indicates the mean
mCherry intensity. In control cortices, mCherry+ cells show uniform distribution across the electroporation domain, while mCherry+ cells form clusters in Cre-electroporated
cortices. (E) Average number of electroporated cells adjacent to a given electroporated cell (within a 15-m radius) in Cre-electroporated cortices was significantly in-
creased compared to the control (P < 0.0001). (F and G) Expression of NC-enriched markers (Tbr1, Ctip2, and Satb2) and MEC-enriched markers (Nrp2, Kitl, and Nurr1)
among mCherry+ cells in control or Cre-electroporated cortices. Significantly fewer COUP-TFI–overexpressing cells displayed NC markers, including Tbr1 (P = 0.0235),
Ctip2 (P = 0.0099), and Satb2 (P = 0.0167), while most expressed Nrp2 (P = 0.0112). Scale bars, 150 m (C and F) and 100 m (G).
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
9 of 14
Using N2a cells, in which COUP-TFI expression is relatively low,
we examined the impact of the COUP-TFI–Pcdh19 genetic path-
way on cell segregation. First, we showed that increasing COUP-
TFI expression enhanced Pcdh19 expression (fig. S9A). Next, we
performed cell aggregation assays (46), which showed that in a cluster,
control cells are dispersed, but control and COUP-TFI–overexpressing
cells were segregated with a higher aggregation score (fig. S9B). We
then used Pcdh19 short hairpin RNA (shRNA) (fig. S9C) to knock
down Pcdh19in COUP-TFI–overexpressing cells and found that
Pcdh19-knockdown COUP-TFI–expressing cells did not retain the
ability to segregate from control cells (fig. S9D). These results suggested
that COUP-TFI induces cell segregation via Pcdh19.
Coelectroporation of Cre and mCherry expression constructs
into COUP-TFITG/TG cortices would induce COUP-TFI overexpres-
sion in the transfected progenitors and their progeny, generating
more COUP-TFI–overexpressing cells than the targeted (mCherry-
expressing) cells. Therefore, to test whether a high level of COUP-TFI
expression indeed cell-autonomously induces Pcdh19 expression
invivo, we electroporated a CAG-COUP-TFI expression construct
into wild-type parietal cortex at E13.5. As expected, we found that
Fig. 6. High level of COUP-TFI expression induces Pcdh19-mediated cell clustering. (A and B) Removing differential cell adhesion from COUP-TFI overexpression [the
shaded blue region in (B)] no longer induces cell clustering and leads to the loss of instability domain (comparing with the red domain in Fig. 5A). (C and D) In vitro cell
aggregation assay with dye-labeled cells from E13.5 cortices showed that mixed medial entorhinal cortical and neocortical cells (M + N) cells segregated more than neo-
cortical cells (N + N) (P = 0.025). (E and F) COUP-TFI binds to a conserved Sp1/COUP-TFI binding site (shaded) in the Pcdh19 promoter, similarly to Rnd2 and Fabp7 promoters
(Pcdh19, P = 0.00826; Rnd2, P = 0.0009; Fabp7, P = 0.0353). (G) On sagittal sections of E18.5 cortices electroporated with indicated constructs at E13.5, most of the control
cells scattered in the CP. COUP-TFI overexpression formed Pcdh19- and Kitl-expressing cell clusters in the intermediate zone (IZ). (H and I) COUP-TFI overexpression re-
pressed Satb2 (P < 0.0001) and induced Nrp2 (P = 0.0082), regardless of Pcdh19 expression (COUP-TFI + Pcdh19KD versus COUP-TFI, Satb2, P = 0.2030; Nrp2, P = 0.0871).
(J) Analyses of adjacent cell number (COUP-TFI versus control, P = 0.0050; COUP-TFI + Pcdh19KD versus COUP-TFI, P = 0.0349), cell distance (COUP-TFI versus control,
P < 0.0001; COUP-TFI + Pcdh19KD versus COUP-TFI, P = 0.026), and cell heterogeneity (COUP-TFI + Pcdh19KD versus COUP-TFI, P < 0.001) showed that COUP-TFI overex-
pression induces cell clustering and knocking down Pcdh19 in COUP-TFI–overexpressing cells reduced COUP-TFI–induced clustering. VZ, ventricular zone; SVZ, subven-
tricular zone; IgG, immunoglobulin G; n.s., not significant. Scale bars, 10 m (C), 300 m (G), and 100 m (H).
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
10 of 14
the high level of COUP-TFI expression induced Pcdh19 expression,
and it also induced expression of other MEC-enriched genes, simi-
lar to Cre electroporation in COUP-TFITG/TG cortices (Fig.6G and
fig. S10A). Most of the COUP-TFI–overexpressing cells formed
clusters in the intermediate zone (Fig.6G) rather than the CP, as
was seen in the Cre-electroporated COUP-TFITG/TG cortices (Fig.5C).
As COUP-TFI is expressed at a higher level in CAG-COUP-TFI
electroporated cells than in Cre-electropored TG cells (fig. S10B),
we used shRNA to knock down COUP-TFI expression in CAG-
COUP-TFI transfected cells to test whether COUP-TFI overexpression
level affected neuronal migration. We found that the CAG- COUP-
TFI electroporated cells could form cell clusters in the CP when
COUP-TFI expression was reduced (fig. S10C). This finding suggests
that the COUP-TFI expression level indeed influences neuronal migra-
tion, and it must be tightly regulated during cortical development.
To directly test whether the induction of Pcdh19 expression is
required for COUP-TFI–induced cell clustering, we electroporated
Pcdh19 shRNA (fig. S9C) together with the CAG-COUP-TFI expres-
sion vector to knock down Pcdh19 in COUP-TFI–overexpressing
cells (fig. S9E). While most of the mCherry-labeled electroporated
cells in the control cortices were Satb2+ and scattered throughout
the CP, most of the COUP-TFI–overexpressing cells ectopically ex-
pressed Nrp2 and Kitl but lost the expression of Satb2 (Fig.6,HandI,
and fig. S9E). We found that knocking down Pcdh19 did not change
the ability of COUP-TFI to induce Nrp2 and Kitl expression or to
repress Satb2 expression (Fig.6,HandI, and fig. S9E), but it partially
blocked the ability of COUP-TFI to induce cell clustering (Fig.6J).
Consistent with the theoretical predictions, clusters were signifi-
cantly less dense (P=0.026) and significantly more heterogeneous
(P<0.001) than in the presence of functional Pcdh19 (Fig.6J). These
results suggest that high levels of COUP-TFI expression induces the
expression of Pcdh19, which segregates MEC cells from NC cells.
DISCUSSION
Patterning of telencephalon neuroepithelium into different progeni-
tor regions gives rise to well-segregated functional domains that are
essential to the wiring of the cerebrum and support complex cere-
bral functions. Hence, even minor defects in early patterning pro-
cesses may be associated with serious intellectual and behavioral
deficits. In this study, we discovered that the position-dependent
expression gradient of COUP-TFI is a critical determinant of the
cell fate decision to generate either NC or MEC. In particular, we
show that mutant mice with low COUP-TFI expression have re-
duced MEC and expanded NC, with a less sharp caudally shifted
NC/MEC border. We also found that COUP-TFI overexpression
expanded MEC at the expense of NC and generated ectopic MEC
domains (Figs.2 and 3) that are not accounted for by the classical
paradigm of cortical patterning. This finding then led us to further
explore the determinants of boundary regularity. We theoretically
examined the regularity of boundaries using a mathematical model
of positional information with aggregation (such as differential
adhesion). While we confirmed that cell adhesion can enhance
sharpness of the boundary, we also showed that instability arises
with excessive adhesion levels, leading to the emergence of irregular
boundaries like those seen in our experiments. In line with the pre-
dictions from our model, we further identified Pcdh19 as a COUP-
TFI–regulated factor that is responsible for MEC neuron adhesion.
Our findings suggest that during cortical development, patterning
TF gradients set up differential cell affinities to ensure the forma-
tion of sharp borders between cortical regions.
We demonstrated that the COUP-TFI expression level regulates
neuronal fate decisions, where a high level of COUP-TFI specifies
MEC neuronal fate between E10.5 (when Emx1-Cre is expressed)
and E15.5 (when MEC neurogenesis terminates). These results sug-
gest that the tight regulation of graded, position-dependent expres-
sion of COUP-TFI is crucial for cortical development. In previous
work, it was proposed that graded expression of patterning TFs is
established, in part, by Fgfs produced by an anterior midline signaling
center, the commissural plate, and by bone morphogenetic proteins
and Wnts produced by a posterior-medial signaling center, the cor-
tical hem (8,47). In addition, COUP-TFI protein level was shown to
be regulated by fibroblast growth factor 8 via miR-21 (48), but how
the expression gradient of COUP-TFI mRNA is established and
maintained requires further investigation.
The observation of ectopic MEC domains in COUP-TFI–
overexpressing cortices contrast with classical work that suggested
that changes in patterning TF expression levels only lead to alter-
ations in boundary locations and area sizes (8). We found that the
formation of ectopic MEC formation is mediated by differential cell
adhesion mechanisms involving Pcdh19. Previously, differential cell
affinity was reported to sharpen and regularize boundaries (1,49,50),
and cell type–specific combinatorial expression of adhesion mole-
cules was recently shown to mediate cell sorting and contribute to
patterning robustness (15). Our experiments and mathematical
model provide further evidence that differential cell affinity plays
an important role in shaping the boundaries within the cortex. Our
study also shows that the effects of differential cell affinity may in-
clude shattered boundaries and heterotopias. These results identify
a limit to the robustness of brain development, paving the way for
future studies on the role of excessive cell adhesion in other devel-
opmental processes.
Furthermore, although our current model was built on the as-
sumption that differential cell adhesion of MEC cells was only af-
fected by cell-autonomous changes in COUP-TFI levels, it does not
rule out a role for non–cell-autonomous interactions in boundary
formation nor the involvement of differential affinity of other cell
types. The facts that this model could reproduce the experimental
observations and that its predictions were validated experimentally
suggest that the modeled features are sufficient to explain the pheno-
type. Notably, the regularity of the solutions to our equations en-
sures that the same phenotypes will persist, even when the model is
modified to include non–cell-autonomous interactions and/or differ-
ential affinity of other cell types, as they do not dominate the dynamics
(see the Robustness of Phenotypes section in the Supplementary Math
model). For example, although COUP-TFI–associated MEC cell
adhesion is sufficient to induce the formation of an instability region,
enhancing adhesion of NC cells could make the patterning more
prominent (see fig. S21).
From a mathematical standpoint, these findings raise interesting
questions regarding partial differential equations in heterogeneous
domains. The model motivated by the cortical development prob-
lem at hand includes a spatial crossing of a Turing-like instability
that was identified for a homogeneous system (i.e., with fixed levels
of patterning TFs E and N). Spatial variations through opposing
gradients were represented as paths on the plane (E, N). For pat-
terns to appear at the boundary, it is necessary but not sufficient
that this path intersects the instability region (Fig.4E). In particular,
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
11 of 14
the length scale (and sharpness) of the gradients is expected to play
a role (see the Supplementary Math model). Identifying sufficient
conditions to create a certain pattern at the boundary, as well as
how the geometric properties of that pattern depend on the shape of
gradients, constitutes an exciting open problem in applied mathe-
matics, with far-reaching applications in various domains.
By showing that COUP-TFI is a key determinant of MEC devel-
opment, our study advances the understanding of the topic, which
is highly significant but largely underexplored. MEC plays a central
role in brain function, connecting the hippocampus to other corti-
cal regions, and it is composed of many specialized cells that are
selective for speed, head direction, or localization, serving as a hub
for neural correlates of spatial navigation (5154). Because MEC is
also one of the first cortical regions affected in Alzheimer’s disease,
patients often experience spatial disorientation in addition to mem-
ory loss (55,56). In addition, developmental defects in entorhinal
structure, such as disruption of cortical layers, heterotopic displace-
ment of neurons, and a paucity of neurons in superficial layers,
were reportedly associated with developmental defects found in
schizophrenia (57,58). Thus, it is important to understand the
unique functional properties of MEC neurons and how these prop-
erties are acquired during development. The processes that we
uncovered could potentially provide a means by which to repair
specific parts of the brain damaged by disease or injury. In addition,
the identification of differential cell adhesion as a key determinant
for precise boundary location and integrity between cortical regions
has profound implications for emerging regenerative technologies
and tissue engineering.
The formation of ectopic domains is reminiscent of heterotopia,
a condition wherein neurons do not migrate properly during fetal
development and form clusters of normal neurons in abnormal
locations. These clusters are generally thought to be the result of
disrupted progenitor proliferation or neuronal migration during
cortical development and may be one of the causes of epilepsy in
humans (59). Our findings suggest that heterotopia may also occur
by the formation of ectopic cortical domains due to dysregulation of
distinct region-specific cell adhesion properties. Particularly relevant
to our study, heterotopia-associated cell clusters were found in the
cortex of X-linked Pcdh19 heterozygous female mice, with mosaic
expression of Pcdh19 in cortical neurons. These mutant mice develop
epilepsy, similar to PCDH19 patients, who also show cortical ab-
normalities (43).
The formation of ectopic MEC in NC is also of interest from
developmental and evolutionary perspectives. It has been suggested
that the navigation system in birds and reptiles is located in the me-
dial pallium (60) and that the AC contributed to the evolution of
dorsomedial NC (61). Our findings show that simply changing the
COUP-TFI expression level can switch the fates of NC and MEC,
suggesting a close evolutionary relationship between the regions. Thus,
shaping of the COUP-TFI expression gradient in cortical progeni-
tors may have provided an intriguing molecular mechanism to reg-
ulate and fine-tune the relative sizes of functional domains in
amniote cortices during evolution.
MATERIALS AND METHODS
Animals
COUP-TFI floxed and transgenic mice were provided by M.-J. Tsai.
The COUP-TFI transgene consists of a human COUP-TFI gene under
the control of the CAG promoter and a floxed stop cassette, in the
Rosa26 locus (34). Emx1-Cre mice were provided by K. Jones. Animal
care and experimental procedures were approved by and performed
in accordance with guidelines provided by the Academia Sinica
Institutional Animal Care and Use Committee. The day of identify-
ing a vaginal plug and the day of birth were designated as E0.5 and
P0, respectively.
Nissl staining, in situ hybridization, immunohistochemistry,
and EdU labeling
Timed-pregnant mice were dissected, and embryonic cortices were
fixed in 4% phosphate-buffered paraformaldehyde (PFA); postnatal
brains were perfused with and postfixed in 4% PFA. For histological
analyses, brains were cryoprotected with 30% sucrose in phosphate-
buffered saline, embedded in Tissue-Tek OCT compound (Sakura
Finetek) and cut in 20- to 25-m sections on a cryostat (Leica). For
Nissl staining, sections were stained with 0.5% cresyl violet and then
dehydrated through graded alcohols. In situ hybridization on sec-
tions and whole mounts was performed as previously described (16).
Antisense RNA probes were labeled with digoxigenin (DIG) using a
DIG-RNA labeling kit (Roche), or with S35 radioactive isotopes.
Sections were pretreated with proteinase K (5g/ml) at room tem-
perature for 10 min, while whole mounts were pretreated with
proteinase K (10g/ml) at room temperature for 30min. The pre-
hybridization, hybridization (for overnight), and posthybridization
washes were done at 65°C, followed by incubation with anti-DIG-AP
(alkaline phosphatase) (1:2000) overnight at 4°C. 4-Nitro blue
tetrazolium chloride (NBT)/5-Bromo-4-chloro-3-indolyl phos-
phate p-toluidine salt (BCIP) (Roche) chromogenic staining was
used to visualize the distribution of specific RNA transcripts. Im-
munohistochemistry was performed as described (62). In short,
primary antibodies were incubated overnight at 4°C in blocking
solution containing 3% bovine serum albumin (Sigma-Aldrich) and
0.3% Triton X-100in phosphate buffer, followed by incubation with
Alexa-conjugated secondary antibodies (Jackson ImmunoResearch)
for 2 hours at room temperature. Cell nuclei were counterstained with
4′,6-diamidino-2-phenylindole (DAPI) (Vector). Primary antibodies
were used at the following concentratio ns: Sa tb2 ( 1:500) , Tbr 1 (1:5 00),
Ctip2 (1:300), Nrp2 (1:200), mCherry (1:500), COUP-TFI (1:200),
Myc Tag (1:500), NF-M (1:500), Vglut2 (1:500), Wfs1 (1:500), Cal-
bindin (1:500), and Nurr1 (1:150). Neuronal birthdating analyses were
performed as described (63). Briefly, EdU (5-ethynyl-2′-deoxyuridine)
(500 ng) was injected into timed-pregnant mice, and the EdU-positive
cells were detected with a Click-iT EdU imaging kit (Invitrogen).
Axonal tracing
Tracing of neuronal projections was performed as described by
Chou etal. (62). Small crystals of the fluorescent carbocyanide dyes,
DiI or DiD (Invitrogen), were inserted into hippocampus or V1 of
P7 brains. Brains were incubated for 3 to 8 weeks in 4% PFA, then
embedded in 5% low-melting agarose, cut into 100-m-thick sec-
tions on a vibratome (Leica), counterstained with DAPI (Vector),
and mounted in 0.1M phosphate buffer.
In vitro cell aggregation assay
The invitro cell aggregation assay was performed as described with
minor modifications (64). Briefly, E13.5 cortices were isolated in
Hanks’ balanced salt solution (HBSS) containing 10 mM Hepes,
and NCs and entorhinal cortices were dissected into small pieces
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
12 of 14
and collected in separate vials on ice. Then, HBSS was replaced by
dissociation buffer (Ca2+/Mg2+-free HBSS containing 10 mM EDTA)
and incubated for 15min at 37°C. Tissue pieces were then dissoci-
ated mechanically with a P200 pipette tip. After removing tissue
debris with a 70-m cell strainer, isolated cells were centrifuged for
5min at 200g and resuspended in minimum essential media (MEM)
without serum. For dye labeling, green-fluorescent (Invitrogen) and
orange-fluorescent (Invitrogen) cell trackers were diluted to final
concentrations of 2.5 and 10 mM in MEM, respectively. Cells were
stained for 30min at 37°C, and staining solution was subsequently
washed with fresh medium. For one well of a 24-well plate, 4 × 105
dye-labeled cells were mixed in 400-l MEM. The plate was incu-
bated on a rotary shaker (100 rpm) at 37°C in a 5% CO2 incubator
for 45min. After shaking, mixed cells were fixed by 4% PFA and
collected for further confocal imaging.
In utero electroporation
Expression constructs CAG-mCherry, CAG-COUP-TFI, and CAG-
Cre were generated by subcloning mCherry, COUP-TFI, or Cre
into a pCAG vector containing the cytomegalovirus early enhancer
element and chicken -actin promoter. For knockdown experiments,
shRNA designed to target the coding sequence of mouse Pcdh19
RNA: TTCTGCCCTTGTCCTAATATA or shRNA designed to
target the coding sequence of mouse COUP-TFI RNA: CACATC-
CGCATCTTTCAGGAA was cloned into the TRC2 vector
(TRC2-pLKO-puro) containing the human U6 promoter. In utero
electroporation was performed as previously described (65). In
short, E13.5 embryos were visualized through the uterus using a fiber
optic light source. Then, either CAG-Cre (1 g/l) or CAG-CreER
(1 g/l) as a control was electroporated into dorsal telencephalons of
embryos with paddle-type electrodes (CUY21EDIT II) in a series of
five square-wave current pulses (40 V, 50-ms width, and 950-ms resting
time). For mechanism exploration, a DNA solution containing CAG-
COUP-TFI (1 g/l) and sh-Pcdh19 or sh-COUP-TFI (1 g/l) was
electroporated, and here, CAG-Cre and sh-gal (1 g/l) were used as
a COUP-TFI overexpression control and a Pcdh19 knockdown control,
respectively. All DNA solutions were mixed with CAG-mcherry
(0.2 g/l) and 0.025% Fast Green (Sigma- Aldrich) and were injected
with a glass capillary into the ventricle of each embryo. Electropo-
rated embryos were allowed to develop until E18.5 and selected for
further analyses by direct visualization of mCherry expression.
RNA-seq analyses
Total RNA was extracted from NC and AC manually isolated from
E17.5 embryos and adult cortices with TRIzol reagent (Invitrogen).
To obtain enough RNA, AC tissues from three to four embryonic
brains were pooled as one sample, and three NC and AC samples
were used for biological triplicates. RNA quality was assessed using
a BioAnalyzer (Agilent Laboratories), and next-generation sequencing
was performed with Illumina HiSeq 4000. Read sequences were then
mapped to the mouse reference database (mm10) by Bowtie2 (v2.2.6).
DESeq2 was used to normalize read counts by taking into account
the gene length and the sequencing depth into FPKM (fragments
per kilobase of transcript per million) value. Differentially expressed
genes were defined by posterior probability of equal expression<
0.05. GO enrichment analysis was performed on the Gene Ontology
Resource (http://geneontology.org/) (66). The heatmaps in figs. S1
and S8 were generated with relative FPKM values, where the maxi-
mal FPKM value of each gene among all samples was set as 1.
ChIP–quantitative polymerase chain reaction
Dorsal telencephalon was dissociated from E13.5 wild-type mice.
The tissues were incubated with disuccinimidyl glutarate (Sigma-
Aldrich) to a final concentration of 2 mM and fixed in 1% formal-
dehyde, followed by cross-linking with 125 mM glycine at pH 7.2.
Cell lysates were sheared by sonication to generate chromatin fragments
with an average length of 100 to 300base pairs. Chromatin-protein com-
plexes were then immunoprecipitated with mouse anti–COUP-TFI
(R&D Systems) or Rabbit Gamma Globulin (Jackson ImmunoResearch)
overnight at 4°C. The antibody-chromatin complexes were then
incubated with Dynabeads protein G (Invitrogen) for 2 hours at 4°C.
Genomic DNA fragments were purified and subjected to quantitative
polymerase chain reaction (PCR) with specific primers [see table S1;
primers for Fabp7 and Rnd2 were used as positive controls (67)] on
real-time reverse transcription PCR (RT-PCR) using LightCycler
480 SYBR Green I Master mix (Roche).
Quantification and statistical analyses
For electroporation data, in the center of the electroporated do-
main, 300-m-wide cortical columns were cropped for quantifi-
cation of the cell numbers and marker intensity. The numbers
of mCherry+, Ctip2+, Tbr1+, Satb2+, Nrp2+, and EdU+ cells were
manually counted using ImageJ/FIJI. With a custom macro, posi-
tion coordination and fluorescence intensities of selected cells were
listed for further analyses of cell numbers, marker gene expression
intensity, and neighboring cell distance. All analyses were per-
formed with three or more biological replicates. The number of in-
dividual animals of the same genotype used is indicated as “n” in the
text and figures. Statistical analyses were performed using Graph-
Pad Prism 5 software. All quantitative data are presented as the
means±SEM. Minimal statistical significance was fixed at P<0.05
for comparisons made by unpaired t test with Welch’s correction
(for Figs.1K,2F,5G, and 6,D,F,I,andJ, and figs. S3, B, D, and E;
S5D; S7C; S8, D and H; and S9, C and D); one-way analysis of vari-
ance with Bonferroni post hoc test (for fig. S9A). Significance is rep-
resented in figures as follows: *P<0.05; **P<0.01; ***P<0.001.
Code availability
Custom MATLAB code used to analyze cortical imaging data and
perform statistical tests on these data is available at https://github.com/
Touboul-Lab/cortex_patterning. Codes were executed on MATLAB
version R2019b. FIJI macro for exporting imaging numerical value
is available at https://github.com/peggyscshu/Cell-grouping.
SUPPLEMENTARY MATERIALS
Supplementary material for this article is available at http://advances.sciencemag.org/cgi/
content/full/7/27/eabf6808/DC1
View/request a protocol for this paper from Bio-protocol.
REFERENCES AND NOTES
1. C. Kiecker, A. Lumsden, Compartments and their boundaries in vertebrate brain
development. Nat. Rev. Neurosci. 6, 553–564 (2005).
2. J. A. Gorski, T. Talley, M. Qiu, L. Puelles, J. L. R. Rubenstein, K. R. Jones, Cortical excitatory
neurons and glia, but not GABAergic neurons, are produced in the Emx1-expressing
lineage. J. Neurosci. 22, 6309–6314 (2002).
3. L. Wolpert, Positional information and the spatial pattern of cellular differentiation.
J. Theor. Biol. 25, 1–47 (1969).
4. L. C. Greig, M. B. Woodworth, M. J. Galazo, H. Padmanabhan, J. D. Macklis, Molecular logic
of neocortical projection neuron specification, development and diversity. Nat. Rev.
Neurosci. 14, 755–769 (2013).
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
13 of 14
5. J. L. Rubenstein, Annual Research Review: Development of the cerebral cortex: Implications
for neurodevelopmental disorders. J. Child Psychol. Psychiatry 52, 339–355 (2011).
6. J. M. Hebert, G. Fishell, The genetics of early telencephalon patterning: Some assembly
required. Nat. Rev. Neurosci. 9, 678–685 (2008).
7. C. Alfano, M. Studer, Neocortical arealization: Evolution, mechanisms, and open
questions. Dev. Neurobiol. 73, 411–447 (2013).
8. D. D. O'Leary, S. J. Chou, S. Sahara, Area patterning of the mammalian cortex. Neuron 56,
252–269 (2007).
9. M. Osterfield, M. W. Kirschner, J. G. Flanagan, Graded positional information:
Interpretation for both fate and guidance. Cell 113, 425–428 (2003).
10. R. R. Kay, C. R. Thompson, Forming patterns in development without morphogen
gradients: Scattered differentiation and sorting out. Cold Spring Harb. Perspect. Biol. 1,
a001503 (2009).
11. T. Gregor, D. W. Tank, E. F. Wieschaus, W. Bialek, Probing the limits to positional
information. Cell 130, 153–164 (2007).
12. J. O. Dubuis, G. Tkacik, E. F. Wieschaus, T. Gregor, W. Bialek, Positional information, in bits.
Proc. Natl. Acad. Sci. U.S.A. 110, 16301–16308 (2013).
13. B. G. Godard, C. P. Heisenberg, Cell division and tissue mechanics. Curr. Opin. Cell Biol. 60,
114–120 (2019).
14. M. S. Steinberg, Does differential adhesion govern self-assembly processes
in histogenesis? Equilibrium configurations and emergence of a hierarchy among
populations of embryonic cells. J. Exp. Zool. 173, 395–433 (1970).
15. T. Y. C. Tsai, M. Sikora, P. Xia, T. Colak-Champollion, H. Knaut, C.-P. Heisenberg,
S. G. Megason, An adhesion code ensures robust pattern formation during tissue
morphogenesis. Science 370, 113–116 (2020).
16. M. Armentano, S. J. Chou, G. Srubek Tomassy, A. Leingärtner, D. D. M. O'Leary, M. Studer,
COUP-TFI regulates the balance of cortical patterning between frontal/motor
and sensory areas. Nat. Neurosci. 10, 1277–1286 (2007).
17. C. Alfano, E. Magrinelli, K. Harb, R. F. Hevner, M. Studer, Postmitotic control of sensory
area specification during neocortical development. Nat. Commun. 5, 5632 (2014).
18. G. Flore, G. di Ruberto, J. Parisot, S. Sannino, F. Russo, E. A. Illingworth, M. Studer,
E. de Leonibus, Gradient COUP-TFI expression is required for functional organization
of the hippocampal septo-temporal longitudinal axis. Cereb. Cortex 27, 1629–1643
(2017).
19. M. Armentano, A. Filosa, G. Andolfi, M. Studer, COUP-TFI is required for the formation
of commissural projections in the forebrain by regulating axonal growth. Development
133, 4151–4162 (2006).
20. C. Zhou, Y. Qiu, F. A. Pereira, M. C. Crair, S. Y. Tsai, M. J. Tsai, The nuclear orphan receptor
COUP-TFI is required for differentiation of subplate neurons and guidance
of thalamocortical axons. Neuron 24, 847–859 (1999).
21. C. Zhou, S. Y. Tsai, M. J. Tsai, COUP-TFI: An intrinsic factor for early regionalization
of the neocortex. Genes Dev. 15, 2054–2059 (2001).
22. D. G. Bosch, F. N. Boonstra, C. Gonzaga-Jauregui, M. Xu, J. de Ligt, S. Jhangiani,
W. Wiszniewski, D. M. Muzny, H. G. Yntema, R. Pfundt, L. E. Vissers, L. Spruijt,
E. A. Blokland, C. A. Chen; Baylor-Hopkins Center for Mendelian Genomics, R. A. Lewis,
S. Y. Tsai, R. A. Gibbs, M. J. Tsai, J. R. Lupski, H. Y. Zoghbi, F. P. Cremers, B. B. de Vries,
C. P. Schaaf, NR2F1 mutations cause optic atrophy with intellectual disability. Am. J. Hum.
Genet. 94, 303–309 (2014).
23. C. A. Chen, D. G. M. Bosch, M. T. Cho, J. A. Rosenfeld, M. Shinawi, R. A. Lewis, J. Mann,
P. Jayakar, K. Payne, L. Walsh, T. Moss, A. Schreiber, C. Schoonveld, K. G. Monaghan,
F. Elmslie, G. Douglas, F. N. Boonstra, F. Millan, F. P. M. Cremers, D. McKnight, G. Richard,
J. Juusola, F. Kendall, K. Ramsey, K. Anyane-Yeboa, E. Malkin, W. K. Chung, D. Niyazov,
J. M. Pascual, M. Walkiewicz, V. Veluchamy, C. Li, F. M. Hisama, B. B. A. de Vries, C. Schaaf,
The expanding clinical phenotype of Bosch-Boonstra-Schaaf optic atrophy syndrome:
20 new cases and possible genotype-phenotype correlations. Genet. Med. 18, 1143–1150
(2016).
24. M. Bertacchi, A. L. Romano, A. Loubat, F. Tran Mau-Them, M. Willems, L. Faivre,
P. Khau van Kien, L. Perrin, F. Devillard, A. Sorlin, P. Kuentz, C. Philippe, A. Garde, F. Neri,
R. di Giaimo, S. Oliviero, S. Cappello, L. D'Incerti, C. Frassoni, M. Studer, NR2F1 regulates
regional progenitor dynamics in the mouse neocortex and cortical gyrification in BBSOAS
patients. EMBO J. 39, e104163 (2020).
25. Y. Nakagawa, D. D. O'Leary, Dynamic patterned expression of orphan nuclear receptor genes
RORalpha and RORbeta in developing mouse forebrain. Dev. Neurosci. 25, 234–244 (2003).
26. S. Bulchand, L. Subramanian, S. Tole, Dynamic spatiotemporal expression of LIM genes
and cofactors in the embryonic and postnatal cerebral cortex. Dev. Dyn. 226, 460–469 (2003).
27. Y. Nakagawa, J. E. Johnson, D. D. O'Leary, Graded and areal expression patterns
of regulatory genes and cadherins in embryonic neocortex independent
of thalamocortical input. J. Neurosci. 19, 10877–10885 (1999).
28. A. H. Kashani, Z. Qiu, L. Jurata, S. K. Lee, S. Pfaff, S. Goebbels, K. A. Nave, A. Ghosh, Calcium
activation of the LMO4 transcription complex and its role in the patterning
of thalamocortical connections. J. Neurosci. 26, 8398–8408 (2006).
29. D. Jabaudon, S. J. Shnider, D. J. Tischfield, M. J. Galazo, J. D. Macklis, ROR induces
barrel-like neuronal clusters in the developing neocortex. Cereb. Cortex 22, 996–1006
(2012).
30. T. Kitamura, M. Pignatelli, J. Suh, K. Kohara, A. Yoshiki, K. Abe, S. Tonegawa, Island cells
control temporal association memory. Science 343, 896–901 (2014).
31. S. Ray, R. Naumann, A. Burgalossi, Q. Tang, H. Schmidt, M. Brecht, Grid-layout
and theta-modulation of layer 2 pyramidal neurons in medial entorhinal cortex. Science
343, 891–896 (2014).
32. O. Saucedo-Cardenas, O. M. Conneely, Comparative distribution of NURR1 and NUR77
nuclear receptors in the mouse central nervous system. J. Mol. Neurosci. 7, 51–63 (1996).
33. H. Mashiko, A. C. Yoshida, S. S. Kikuchi, K. Niimi, E. Takahashi, J. Aruga, H. Okano,
T. Shimogori, Comparative anatomy of marmoset and mouse cortex from genomic
expression. J. Neurosci. 32, 5039–5053 (2012).
34. S. P. Wu, D. K. Lee, F. J. Demayo, S. Y. Tsai, M. J. Tsai, Generation of ES cells for conditional
expression of nuclear receptors and coregulators in vivo. Mol. Endocrinol. 24, 1297–1304
(2010).
35. A. Faedo, G. S. Tomassy, Y. Ruan, H. Teichmann, S. Krauss, S. J. Pleasure, S. Y. Tsai,
M. J. Tsai, M. Studer, J. L. R. Rubenstein, COUP-TFI coordinates cortical patterning,
neurogenesis, and laminar fate and modulates MAPK/ERK, AKT, and beta-catenin
signaling. Cereb. Cortex 18, 2117–2131 (2008).
36. F. Donato, R. I. Jacobsen, M. B. Moser, E. I. Moser, Stellate cells drive maturation
of the entorhinal-hippocampal circuit. Science 355, eaai8178 (2017).
37. A. Gierer, H. Meinhardt, A theory of biological pattern formation. Kybernetik 12, 30–39
(1972).
38. C. Quininao, A. Prochiantz, J. Touboul, Local homeoprotein diffusion can stabilize
boundaries generated by graded positional cues. Development 142, 1860–1868 (2015).
39. B. Perthame, C. Quininao, J. Touboul, Competition and boundary formation
in heterogeneous media: Application to neuronal differentiation. Math. Models Methods
Appl. Sci. 25, 2477–2502 (2015).
40. A. M. Turing, The Chemical Basis of Morphogenesis. Philos. Trans. Roy. Soc. Lond. B Biol. Sci.
237, 37–72 (1952).
41. B. t. Perthame, in Lecture Notes on Mathematical Modelling in the Life Sciences (Springer
International Publishing, 2015), p. 1.
42. Y. Gaitan, M. Bouchard, Expression of the delta-protocadherin gene Pcdh19
in the developing mouse embryo. Gene Expr. Patterns 6, 893–899 (2006).
43. D. T. Pederick, K. L. Richards, S. G. Piltz, R. Kumar, S. Mincheva-Tasheva, S. A. Mandelstam,
R. C. Dale, I. E. Scheffer, J. Gecz, S. Petrou, J. N. Hughes, P. Q. Thomas, Abnormal cell sorting
underlies the unique X-linked inheritance of PCDH19 epilepsy. Neuron 97, 59–66.e55 (2018).
44. S. Hayashi, Y. Inoue, S. Hattori, M. Kaneko, G. Shioi, T. Miyakawa, M. Takeichi, Loss
of X-linked Protocadherin-19 differentially affects the behavior of heterozygous female
and hemizygous male mice. Sci. Rep. 7, 5801 (2017).
45. X. Lv, S. Q. Ren, X. J. Zhang, Z. Shen, T. Ghosh, A. Xianyu, P. Gao, Z. Li, S. Lin, Y. Yu, Q. Zhang,
M. Groszer, S. H. Shi, TBR2 coordinates neurogenesis expansion and precise microcircuit
organization via Protocadherin 19 in the mammalian cortex. Nat. Commun. 10, 3946 (2019).
46. A. J. Bisogni, S. Ghazanfar, E. O. Williams, H. M. Marsh, J. Y. H. Yang, D. M. Lin, Tuning
of delta-protocadherin adhesion through combinatorial diversity. eLife 7, e41050 (2018).
47. T. Shimogori, V. Banuchi, H. Y. Ng, J. B. Strauss, E. A. Grove, Embryonic signaling centers
expressing BMP, WNT and FGF proteins interact to pattern the cerebral cortex.
Development 131, 5639–5647 (2004).
48. M. Terrigno, M. Bertacchi, L. Pandolfini, M. Baumgart, M. Calvello, A. Cellerino, M. Studer,
F. Cremisi, The microRNA miR-21 is a mediator of FGF8 action on cortical COUP-TFI
translation. Stem Cell Rep 11, 756–769 (2018).
49. R. A. Foty, M. S. Steinberg, Differential adhesion in model systems. Wiley Interdiscip. Rev.
Dev. Biol. 2, 631–645 (2013).
50. E. Batlle, D. G. Wilkinson, Molecular mechanisms of cell segregation and boundary
formation in development and tumorigenesis. Cold Spring Harb. Perspect. Biol. 4, a008227
(2012).
51. J. O'Keefe, J. Dostrovsky, The hippocampus as a spatial map. Preliminary evidence
from unit activity in the freely-moving rat. Brain Res. 34, 171–175 (1971).
52. T. Hafting, M. Fyhn, S. Molden, M. B. Moser, E. I. Moser, Microstructure of a spatial map
in the entorhinal cortex. Nature 436, 801–806 (2005).
53. F. Sargolini, M. Fyhn, T. Hafting, B. McNaughton, M. P. Witter, M. B. Moser, E. I. Moser,
Conjunctive representation of position, direction, and velocity in entorhinal cortex.
Science 312, 758–762 (2006).
54. M. P. Witter, T. P. Doan, B. Jacobsen, E. S. Nilssen, S. Ohara, Architecture of the entorhinal
cortex a review of entorhinal anatomy in rodents with some comparative notes. Front.
Syst. Neurosci. 11, 46 (2017).
55. H. Braak, E. Braak, Neuropathological stageing of Alzheimer-related changes. Acta
Neuropathol. 82, 239–259 (1991).
56. L. Kunz, T. N. Schroder, H. Lee, C. Montag, B. Lachmann, R. Sariyska, M. Reuter,
R. Stirnberg, T. Stocker, P. C. Messing-Floeter, J. Fell, C. F. Doeller, N. Axmacher, Reduced
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Feng et al., Sci. Adv. 2021; 7 : eabf6808 2 July 2021
SCIENCE ADVANCES | RESEARCH ARTICLE
14 of 14
grid-cell-like representations in adults at genetic risk for Alzheimer's disease. Science 350,
430–433 (2015).
57. S. E. Arnold, B. T. Hyman, G. W. Van Hoesen, A. R. Damasio, Some cytoarchitectural abnormalities
of the entorhinal cortex in schizophrenia. Arch. Gen. Psychiatry 48, 625–632 (1991).
58. P. J. Harrison, D. R. Weinberger, Schizophrenia genes, gene expression, and
neuropathology: On the matter of their convergence. Mol. Psychiatry 10, 40–68 (2005).
59. F. Watrin, J. B. Manent, C. Cardoso, A. Represa, Causes and consequences of gray matter
heterotopia. CNS Neurosci. Ther. 21, 112–122 (2015).
60. E. D. Jarvis, in Encyclopedia of Neuroscience, M. D. Binder, N. Hirokawa, U. Windhorst, Eds.
(Springer Berlin Heidelberg, 2009), pp. 1390–1400.
61. Z. Molnár, A. B. Butler, Neuronal changes during forebrain evolution in amniotes:
An evolutionary developmental perspective. Prog. Brain Res. 136, 21–38 (2002).
62. S. J. Chou, C. G. Perez-Garcia, T. T. Kroll, D. D. O'Leary, Lhx2 specifies regional fate in Emx1
lineage of telencephalic progenitors generating cerebral cortex. Nat. Neurosci. 12,
1381–1389 (2009).
63. H.-W. Hsing, Z.-H. Zhuang, Z.-X. Niou, S.-J. Chou, Temporal differences in interneuron
invasion of neocortex and piriform cortex during mouse cortical development. Cereb.
Cortex 30, 3015–3029 (2020).
64. M. Gotz, A. Wizenmann, S. Reinhardt, A. Lumsden, J. Price, Selective adhesion of cells
from different telencephalic regions. Neuron 16, 551–564 (1996).
65. C. F. Wang, H. W. Hsing, Z. H. Zhuang, M. H. Wen, W. J. Chang, C. G. Briz, M. Nieto,
B. C. Shyu, S. J. Chou, Lhx2 expression in postmitotic cortical neurons initiates assembly
of the thalamocortical somatosensory circuit. Cell Rep. 18, 849–856 (2017).
66. The Gene Ontology Consortium, Expansion of the Gene Ontology knowledgebase
and resources. Nucleic Acids Res. 45, D331–D338 (2017).
67. C. Alfano, L. Viola, J. I. T. Heng, M. Pirozzi, M. Clarkson, G. Flore, A. de Maio, A. Schedl, F. Guillemot,
M. Studer, COUP-TFI promotes radial migration and proper morphology of callosal projection
neurons by repressing Rnd2 expression. Development 138, 4685–4697 (2011).
68. B. Ermentrout, Simulating, analyzing, and animating dynamical systems: A guide to
XPPAUT for researchers and students (SIAM, 2002), vol. 14.
69. F. Hecht, New development in FreeFem++. J. Numer. Math. 20, 251–265 (2012).
70. T. Hillen, K. Painter, Global existence for a parabolic chemotaxis model with prevention
of overcrowding. Adv. Appl. Math. 26, 280–301 (2001).
71. T. Hillen, K. Painter, A user's guide to PDE models for chemotaxis. J. Math. Biol. 58,
183–217 (2009).
72. D. Horstmann, From 1970 until present: The Keller-Segel model in chemotaxis and its
consequences I. Jahresber. Deutsch. Math. Verein. 105, 103–165 (2003).
73. E. F. Keller, L. A. Segel, Initiation of slime mold aggregation viewed as an instability.
J. Theor. Biol. 26, 399–415 (1970).
74. E. F. Keller, L. A. Segel, Model for chemotaxis. J. Theor. Biol. 30, 225–234 (1971).
75. L. Wolpert, Chapter 6 Positional information and pattern formation. Curr. Top. Dev. Biol. 6,
183–224 (1971).
Acknowledgments: We thank M.-J. Tsai for providing the COUP-TFI floxed and transgenic
allele and K. Jones for Emx1-Cre. We also thank members of the Chou laboratory for help and
S.-C. P. Hsu for help on analyzing imaging data. Funding: This work was supported by Ministry
of Science and Technology (MOST 108-2311-B-001-021, S.-J.C.), Academia Sinica (AS-CDA-
107-L09, S.-J.C.), the Institute of Cellular and Organismic Biology of Academia Sinica (S.-J.C.),
and NINDS (RO1 NS099099, J.L.R.). D.P. was partially supported by the Swartz Foundation.
Author contributions: S.-J.C. designed the research. J.F., W.-H.H., C.-S.T., Z.-H.Z., H.-W.H.,
Y.-T.H., and S.-J.C. performed the research and analyzed data. A.F. and J.L.R. provided critical
materials. D.P. and J.T performed mathematic modeling and analyzed data. J.T. and S.-J.C.
wrote the paper. Competing interests: The authors declare that they have no competing
interests. Data and materials availability: All data needed to evaluate the conclusions in the
paper are present in the paper and/or the Supplementary Materials. Additional data related to
this paper may be requested from the authors.
Submitted 11 December 2020
Accepted 12 May 2021
Published 2 July 2021
10.1126/sciadv.abf6808
Citation: J. Feng, W.-H. Hsu, D. Patterson, C.-S. Tseng, H.-W. Hsing, Z.-H. Zhuang, Y.-T. Huang,
A. Faedo, J. L. Rubenstein, J. Touboul, S.-J. Chou, COUP-TFI specifies the medial entorhinal
cortex identity and induces differential cell adhesion to determine the integrity of its boundary
with neocortex. Sci. Adv. 7, eabf6808 (2021).
on July 3, 2021http://advances.sciencemag.org/Downloaded from
to determine the integrity of its boundary with neocortex
COUP-TFI specifies the medial entorhinal cortex identity and induces differential cell adhesion
Faedo, John L. Rubenstein, Jonathan Touboul and Shen-Ju Chou
Jia Feng, Wen-Hsin Hsu, Denis Patterson, Ching-San Tseng, Hsiang-Wei Hsing, Zi-Hui Zhuang, Yi-Ting Huang, Andrea
DOI: 10.1126/sciadv.abf6808
(27), eabf6808.7Sci Adv
ARTICLE TOOLS http://advances.sciencemag.org/content/7/27/eabf6808
MATERIALS
SUPPLEMENTARY http://advances.sciencemag.org/content/suppl/2021/06/28/7.27.eabf6808.DC1
REFERENCES http://advances.sciencemag.org/content/7/27/eabf6808#BIBL
This article cites 72 articles, 18 of which you can access for free
PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions
Terms of ServiceUse of this article is subject to the
is a registered trademark of AAAS.Science AdvancesYork Avenue NW, Washington, DC 20005. The title
(ISSN 2375-2548) is published by the American Association for the Advancement of Science, 1200 NewScience Advances
License 4.0 (CC BY-NC).
Science. No claim to original U.S. Government Works. Distributed under a Creative Commons Attribution NonCommercial
Copyright © 2021 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of
on July 3, 2021http://advances.sciencemag.org/Downloaded from
Article
Full-text available
The assembly and maturation of the mammalian brain result from an intricate cascade of highly coordinated developmental events, such as cell proliferation, migration, and differentiation. Any impairment of this delicate multi-factorial process can lead to complex neurodevelopmental diseases, sharing common pathogenic mechanisms and molecular pathways resulting in multiple clinical signs. A recently described monogenic neurodevelopmental syndrome named Bosch-Boonstra-Schaaf Optic Atrophy Syndrome (BBSOAS) is caused by NR2F1 haploinsufficiency. The NR2F1 gene, coding for a transcriptional regulator belonging to the steroid/thyroid hormone receptor superfamily, is known to play key roles in several brain developmental processes, from proliferation and differentiation of neural progenitors to migration and identity acquisition of neocortical neurons. In a clinical context, the disruption of these cellular processes could underlie the pathogenesis of several symptoms affecting BBSOAS patients, such as intellectual disability, visual impairment, epilepsy, and autistic traits. In this review, we will introduce NR2F1 protein structure, molecular functioning, and expression profile in the developing mouse brain. Then, we will focus on Nr2f1 several functions during cortical development, from neocortical area and cell-type specification to maturation of network activity, hippocampal development governing learning behaviors, assembly of the visual system, and finally establishment of cortico-spinal descending tracts regulating motor execution. Whenever possible, we will link experimental findings in animal or cellular models to corresponding features of the human pathology. Finally, we will highlight some of the unresolved questions on the diverse functions played by Nr2f1 during brain development, in order to propose future research directions. All in all, we believe that understanding BBSOAS mechanisms will contribute to further unveiling pathophysiological mechanisms shared by several neurodevelopmental disorders and eventually lead to effective treatments.
Article
Full-text available
The relationships between impaired cortical development and consequent malformations in neurodevelopmental disorders, as well as the genes implicated in these processes, are not fully elucidated to date. In this study, we report six novel cases of patients affected by BBSOAS (Boonstra-Bosch-Schaff optic atrophy syndrome), a newly emerging rare neurodevelopmental disorder, caused by loss-of-function mutations of the transcriptional regulator NR2F1. Young patients with NR2F1 haploinsufficiency display mild to moderate intellectual disability and show reproducible polymicrogyria-like brain malformations in the parietal and occipital cortex. Using a recently established BBSOAS mouse model, we found that Nr2f1 regionally controls long-term self-renewal of neural progenitor cells via modulation of cell cycle genes and key cortical development master genes, such as Pax6. In the human fetal cortex, distinct NR2F1 expression levels encompass gyri and sulci and correlate with local degrees of neurogenic activity. In addition, reduced NR2F1 levels in cerebral organoids affect neurogenesis and PAX6 expression. We propose NR2F1 as an area-specific regulator of mouse and human brain morphology and a novel causative gene of abnormal gyrification.
Article
Full-text available
Cerebral cortex expansion is a hallmark of mammalian brain evolution; yet, how increased neurogenesis is coordinated with structural and functional development remains largely unclear. The T-box protein TBR2/EOMES is preferentially enriched in intermediate progenitors and supports cortical neurogenesis expansion. Here we show that TBR2 regulates fine-scale spatial and circuit organization of excitatory neurons in addition to enhancing neurogenesis in the mouse cortex. TBR2 removal leads to a significant reduction in neuronal, but not glial, output of individual radial glial progenitors as revealed by mosaic analysis with double markers. Moreover, in the absence of TBR2, clonally related excitatory neurons become more laterally dispersed and their preferential synapse development is impaired. Interestingly, TBR2 directly regulates the expression of Protocadherin 19 (PCDH19), and simultaneous PCDH19 expression rescues neurogenesis and neuronal organization defects caused by TBR2 removal. Together, these results suggest that TBR2 coordinates neurogenesis expansion and precise microcircuit assembly via PCDH19 in the mammalian cortex. The T-box protein TBR2 is involved in cortical neurogenesis expansion during neurodevelopment and is preferentially enriched in intermediate progenitors. The authors show that TBR2 coordinates neurogenesis expansion and precise microcircuit assembly in the mouse cortex via PCDH19.
Article
Full-text available
The delta-protocadherins (δ-Pcdhs) play key roles in neural development, and expression studies suggest they are expressed in combination within neurons. The extent of this combinatorial diversity, and how these combinations influence cell adhesion, is poorly understood. We show that individual mouse olfactory sensory neurons express 0-7 δ-Pcdhs. Despite this apparent combinatorial complexity, K562 cell aggregation assays revealed simple principles mediate tuning of δ-Pcdh adhesion. Cells can vary the number of δ-Pcdhs expressed, the level of surface expression, and which δ-Pcdhs are expressed, as different members possess distinct apparent adhesive affinities. These principles contrast with those identified previously for the clustered protocadherins (cPcdhs), where the particular combination of cPcdhs expressed does not appear to be a critical factor. Despite these differences, we show δ-Pcdhs can modify cPcdh adhesion. Our studies show how intra- and interfamily interactions can greatly amplify the impact of this small subfamily on neuronal function.
Article
Full-text available
The morphogen FGF8 plays a pivotal role in neocortical area patterning through its inhibitory effect on COUP-TFI/Nr2f1 anterior expression, but its mechanism of action is poorly understood. We established an in vitro model of mouse embryonic stem cell corticogenesis in which COUP-TFI protein expression is inhibited by the activation of FGF8 in a time window corresponding to cortical area patterning. Interestingly, overexpression of the COUP-TFI 3'UTR reduces the inhibitory effect of FGF8 on COUP-TFI translation. FGF8 induces the expression of few miRNAs targeting COUP-TFI 3'UTR in silico. We found that the functional inhibition of miR-21 can effectively counteract the inhibitory effect of FGF8 in vitro and regulate COUP-TFI protein levels in vivo. Accordingly, miR-21 expression is complementary to COUP-TFI expression during corticogenesis. These data support a translational control of COUP-TFI gradient expression by FGF8 via miR-21 and contribute to our understanding of how regionalized expression is established during neocortical area mapping.
Article
Full-text available
Mutations in the X-linked gene Protocadherin-19 (Pcdh19) cause female-limited epilepsy and mental retardation in humans. Although Pcdh19 is known to be a homophilic cell-cell adhesion molecule, how its mutations bring about female-specific disorders remains elusive. Here, we report the effects of Pcdh19 knockout in mice on their development and behavior. Pcdh19 was expressed in various brain regions including the cerebral cortex and hippocampus. Although Pcdh19-positive cells were evenly distributed in layer V of wild-type cortices, their distribution became a mosaic in Pcdh19 heterozygous female cortices. In cortical and hippocampal neurons, Pcdh19 was localized along their dendrites, showing occasional accumulation on synapses. Pcdh19 mutants, however, displayed no detectable abnormalities in dendrite and spine morphology of layer V neurons. Nevertheless, Pcdh19 hemizygous males and heterozygous females showed impaired behaviors including activity defects under stress conditions. Notably, only heterozygous females exhibited decreased fear responses. In addition, Pcdh19 overexpression in wild-type cortices led to ectopic clustering of Pcdh19-positive neurons. These results suggest that Pcdh19 is required for behavioral control in mice, but its genetic loss differentially affects the male and female behavior, as seen in human, and they also support the hypothesis that the mosaic expression of Pcdh19 in brains perturbs neuronal interactions.
Article
Animal development entails the organization of specific cell types in space and time, and spatial patterns must form in a robust manner. In the zebrafish spinal cord, neural progenitors form stereotypic patterns despite noisy morphogen signaling and large-scale cellular rearrangements during morphogenesis and growth. By directly measuring adhesion forces and preferences for three types of endogenous neural progenitors, we provide evidence for the differential adhesion model in which differences in intercellular adhesion mediate cell sorting. Cell type–specific combinatorial expression of different classes of cadherins (N-cadherin, cadherin 11, and protocadherin 19) results in homotypic preference ex vivo and patterning robustness in vivo. Furthermore, the differential adhesion code is regulated by the sonic hedgehog morphogen gradient. We propose that robust patterning during tissue morphogenesis results from interplay between adhesion-based self-organization and morphogen-directed patterning.
Article
Establishing a balance between excitation and inhibition is critical for brain functions. However, how inhibitory interneurons (INs) generated in the ventral telencephalon integrate with the excitatory neurons generated in the dorsal telencephalon remains elusive. Previous studies showed that INs migrating tangentially to enter the neocortex (NCx), remain in the migratory stream for days before invading the cortical plate during late corticogenesis. Here we show that in developing mouse cortices, INs in the piriform cortex (PCx; the major olfactory cortex) distribute differently from those in the NCx. We provide evidence that during development INs invade and mature earlier in PCx than in NCx, likely owing to the lack of CXCR4 expression in INs from PCx compared to those in NCx. We analyzed IN distribution patterns in Lhx2 cKO mice, where projection neurons in the lateral NCx are re-fated to generate an ectopic PCx (ePCx). The PCx-specific IN distribution patterns found in ePCx suggest that properties of PCx projection neurons regulate IN distribution. Collectively, our results show that the timing of IN invasion in the developing PCx fundamentally differs from what is known in the NCx. Further, our results suggest that projection neurons instruct the PCx-specific pattern of IN distribution.
Article
The spatiotemporal organization of cell divisions constitutes an integral part in the development of multicellular organisms, and mis-regulation of cell divisions can lead to severe developmental defects. Cell divisions have an important morphogenetic function in development by regulating growth and shape acquisition of developing tissues, and, conversely, tissue morphogenesis is known to affect both the rate and orientation of cell divisions. Moreover, cell divisions are associated with an extensive reorganization of the cytoskeleton and adhesion apparatus in the dividing cells that in turn can affect large-scale tissue rheological properties. Thus, the interplay between cell divisions and tissue morphogenesis plays a key role in embryo and tissue morphogenesis.
Article
X-linked diseases typically exhibit more severe phenotypes in males than females. In contrast, protocadherin 19 (PCDH19) mutations cause epilepsy in heterozygous females but spare hemizygous males. The cellular mechanism responsible for this unique pattern of X-linked inheritance is unknown. We show that PCDH19 contributes to adhesion specificity in a combinatorial manner such that mosaic expression of Pcdh19 in heterozygous female mice leads to striking sorting between cells expressing wild-type (WT) PCDH19 and null PCDH19 in the developing cortex, correlating with altered network activity. Complete deletion of PCDH19 in heterozygous mice abolishes abnormal cell sorting and restores normal network activity. Furthermore, we identify variable cortical malformations in PCDH19 epilepsy patients. Our results highlight the role of PCDH19 in determining cell adhesion affinities during cortical development and the way segregation of WT and null PCDH19 cells is associated with the unique X-linked inheritance of PCDH19 epilepsy.